Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
1 Overview of the Thesis
How are beautiful singing voices developed? What features of the singing voice define
a voice of quality? Can these features be measured accurately and more importantly,
specifically taught? In 1967, Appleman, in The Science of Vocal Pedagogy, asserted
“vocal pedagogy cannot survive as an independent educational entity if the
physiological and physical facts, which comprise its core, remain subjects of sciolism
(superficial knowledge). Researchers must uncover and interpret scientific facts about
voice, voice quality and vocal pedagogy so that they might become realistic pedagogical
tools that may be employed by future teachers of voice” (Appelman, 1967, p. 5). This
thesis takes up Appleman’s call for research on the singing voice that both contributes
to our understanding of what constitutes a voice of quality and provides useful tools for
vocal pedagogues.
It includes published work in international peer reviewed journals (see Appendix W)
that extends previous research into the singing voice and its pedagogy. The hypotheses
are well motivated from the existing research and from established pedagogical
techniques. It forms the basis for future interdisciplinary work in assessing, describing
and training the classical singing voice.
1.1 Historical background to the thesis
Singing is the most natural of all musical arts. Wagner wrote ‘The human voice is the
practical basis of Music, and however far the latter may journey on her primal path, the
boldest combinations of the tone-setter, the most daring execution of the instrumental-
virtuoso, will always have to hark back to the purely singable, to find the law for their
1
achievements’ (Ellis, 1994). Throughout recorded history to the present day, singing has
played a significant role in cultural, social and emotional life.
1.1.1 Singing style in the western classical tradition
George Bernard Shaw said, “the notion that singing has deteriorated in the present
century is only a phase of the Good Old Times delusion … Every musical period suffers
from the illusion that it has lost the art of singing and looks back to an imaginary golden
age in which all singers had the secret of the bel canto taught by Italian magicians and
practiced in excelsis at the great Opera Houses of Europe” (Shaw, 1960).
Historical accounts of singing and singing technique do not adequately define vocal
quality. They provide information on famous singers and their teachers in each century
and give anecdotal accounts of singing training and subjective descriptions of vocal
quality. This tradition, often referred to as bel canto, broadly represents the classically
trained voice of opera and concert singers. Classical singing style has evolved over five
centuries in accordance with changing musical styles, compositional styles, cultural and
social factors and performing venues; all of which impacted on vocal production.
Scholars differ with respect to the advent of western classical singing: some attribute the
rise of the virtuoso singer in the late 16th to early 17th centuries as the beginning of
western classical singers (Duey, 1980; Stark, 1999); others date its advent later to the
development of specific operatic singing style which began in the 19th century (Potter,
1998).
2
1.1.2 Classical voice quality
1.1.2.1 Changing musical styles
The birth of opera (including favoli in musica, the earliest manifestations of opera) at
the beginning of the Baroque period marked the beginning of a new style of solo
singing. Composers like Monteverdi pioneered a new compositional style emphasising
the importance of the text, “making the words the mistress of the harmony and not the
servant”. The solo singer was the focus of the musical performance, in contrast with the
polyphonic Renaissance style, where the rules of counterpoint and voice-leading made
all voices and instruments equal (Rosand, 1989; Treitler & Strunk, 1998).
This new musical art form revived the humanist ideals of the ancient Greeks, with the
rise of the ‘musician poet’. Singing was regarded as an extension of oratorical speech.
As the singer was elevated to narrator, he had the freedom to interpret the poetic text in
his vocal performance. Singers of the day used extensive improvisation or
ornamentation to enhance their interpretation of the text and historical conventions in
these passagi (improvisations). The earliest singing treatises (Bacilly, 1668/1968;
Caccini, 1602/1930; Praetorius, 1619/1979) focussed on the art and interpretation of
ornamentation.
Musicologists attribute expressivity in early Baroque vocal music to the relationship
between text and music, but cannot ascertain the effect of these musical changes on
voice quality. However, the Italian singing style was imitated throughout 17th century
Europe. German and French writers praised the Italians’ ability to sing in both an
‘oratorical’ (Praetorius, 1619/1979) and expressive style, capturing the emotion of the
3
text (Mersenne, 1636/1979). Italians’ solo singing style was different to both the
Germanic (Praetorius, 1619/1979) and English (Playford, 1694/1972) choral singing
styles. The Italian language, with its sonorous vowel sounds, was ideal for enunciation
(Bacilly, 1668/1968) of poetic texts.
The impact of musical style in singing is demonstrated by the 17th century English
national singing style which followed an autonomous tradition, distinct from European
singing, as English opera developed a different compositional style as music drama with
spoken dialogue. By 1711, Handel’s opera Rinaldo opened in London, and England,
like the rest of Europe, was exposed to the Italian singing style.
1.1.2.2 Singers rise to fame
By the early 17th century, public performances of Italian opera were widespread.
Castrati dominated opera singing from the mid-17th century and were cast in the most
important roles in operas, while tenors and basses were reduced to minor roles. Castrato
singing was considered the highest art, superior to all other vocal qualities and these
singers were distinguished by extraordinary vocal power and flexibility unmatched by
other singers. Their powerful voices suited the new performance venues, as the first
public performances of opera were in rooms at least twice the size (Worsthorne, 1968)
of the first performance of Monteverdi’s Orfeo at the palace of Margherita Gonzaga-
Este in 1607 (Fenlon, 1984). Twentieth century critics speculate that the bel canto
tradition was synonymous with castrati and their highly florid vocal style (Duey, 1980;
Foreman, 1969).
4
1.1.2.3 Virtuosity
The earliest singing treatises (Bacilly, 1668/1968; Caccini, 1602/1930; Praetorius,
1619/1979) focussed on instructions for ornamentation, an intrinsic component of good
singing at that time. Later writers Tosi and Mancini described the art of vocal
embellishment: the “even, clear, flexible and moderately quick” trill as “indispensable
to singers” (Tosi, 1743/1967, ch 3). Tosi asserted that a singer who could “produce no
trill or only a faulty one will never become a great singer.” Tosi and Mancini also
identified distinct vocal registers with different sounds qualities, voce di petto (chest
voice) and voce di testa (head voice).
Eighteenth century composers Handel, Mozart and even Rossini incorporated the
virtuosic trend in singing into their scores, particularly for castrati and coloratura
sopranos. Singers extended this virtuosity in elaborate and improvised cadential figures.
Ornamentation as defined in the 18th century was different to the 20th century definition
of vibrato. For example, W.A. Mozart, in a letter to his father (Paris, 12 June 1778)
described the use of an ‘appropriate’ vibrato. He praised singer Joseph Meisner for his
cantabile singing (Anderson, 1991), but described his tendency to tremble, which
distorted the sound so much that Mozart could determine beats of a crotchet or quaver.
He said that “[t]he human voice trembles naturally – but in its own way – and only to
such a degree that the effect is beautiful. Such is the nature of the voice; and people
imitate it not only on wind instruments, but on stringed instruments too and even on the
clavier. But the moment the proper limit is overstepped, it is no longer beautiful –
because it is contrary to nature. It reminds me then of the organ when the bellows are
5
puffing” (Anderson, 1991, p. 552). Singing pedagogue Tosi also warned singers to sing
“without vocal trembling” (Tosi, 1743/1967).
1.1.2.4 Changing accompaniments
Changes to the ‘loudness’ of singing occurred in response to changes in sound quality
of the instruments used to accompany voice: for example, the harpsichord superseded
the lute as an accompanying instrument (Potter, 1998). “[I]nevitably, singers had to sing
louder” (Potter, 1998, p. 50). By the 19th century, voices had undoubtedly grown in size.
1.1.2.5 A new vocal style
In the 19th century, Rossini reportedly exclaimed “Alas for us, we have lost our bel
canto” (Jander & Harris, 2004). Changing vocal styles in opera, concert and oratorio
singing and a shift towards the modern idea of a classical voice gradually appeared in
19th century literature on voice (Potter, 1998).
Orchestras steadily grew in size alongside dedicated performance venues (rather than
domestic musical settings). By the mid-1800s, composers complained of voices that
could not be heard in larger halls (Cairns, 1969). Voices changed throughout the century
to suit these new performance requirements. Singers were no longer praised for their
virtuosic vocal ornamentation. Musical literature reports new classifications of voice or
‘fachs’ to define vocal timbre (tenore robusto, tenore di forza, Heldentenor, ‘Verdi
baritone’, ‘Falcon soprano’, ‘dramatic soprano’ and ‘lyrico spinto’). These fachs
describe vocal range and volume or carrying power.
6
The tenor voice, with an extended and powerful upper range, superseded the castrati
voices of previous centuries. Similarly, soprano voices extended upwards and specific
roles were written for baritones and mezzo soprani by composers such as Verdi.
Remarkable new vocal qualities emerged, which we would now recognise as ‘operatic’
For example, the first tenor high C by singer Duprez in Rossini’s William Tell in 1831
(Cairns, 1969) shocked unsuspecting audiences of the day.
Wagner and Verdi emphasised the relationship between text and music and cultivated a
speech influenced (Sprechgesang) or dramatic declamation style. Voices performing
epic Wagnerian works required unprecedented strength and endurance to be heard over
the largest orchestras. The term bel canto was used by Italians in reaction to the
Wagnerian singing style, and renewed patriotic support for the older traditions emerged
(Duey, 1980).
Surveys of the pedagogical literature (Burgin, 1973; Fields, 1947; Monahan, 1978)
identify the 1840s and Manuel Garcia II’s publications (Garcia, 1911) as the beginning
of a new teaching style based on scientific theory. Descriptions of vocal quality in
classical singing are supported by physiological instructions to achieve the quality. For
example, Garcia was first to advocate a low larynx position, and its new vocal quality.
Garcia’s observations of the larynx with the primitive laryngoscope marked the
beginning of a scientific approach to singing, which has influenced all subsequent vocal
study. By the late 19th century, the majority of pedagogical texts refer to anatomy,
physiology and the acoustics of voice production (Burgin, 1973; Fields, 1947;
Monahan, 1978).
7
1.2 Empirical research in singing voice
Vocal pedagogy in Western art music is now four centuries old, yet the definition of a
voice of quality remains as complex in the 21st century as it was in centuries before
acoustics and audio recordings. Empirical research into the singing voice uses acoustic
and perceptual methodologies to define vocal parameters associated with “good” vocal
quality and relate it to vocal functioning. Later studies report components of good
singing, identify differences between male and female singers, between singers and
speakers and between classical singers from pop or other styles of singing. These
studies also attempt to capture the qualities or features that distinguish novice from
professional singers and from local singers to singers of international renown.
While there is a proliferation of acoustic studies in singing voice throughout the
century, current writers (Callaghan, 2000; Miller, 1998) point out that it is difficult to
draw firm conclusions about the implications for singing pedagogy of many of these
scientific studies which typically use small numbers of subjects, rarely link the vocal
strategies used to perceptual results, often use student singers rather than professionals,
and rarely identify the teaching/learning approach of the subjects. Most importantly,
“The subtle individual properties that set one voice apart from another should not be
averaged out” (Miller, 1998, p. 299).
1.2.1 Vibrato
Vibrato is taken for granted as an intrinsic quality of the classical singing voice and in
pedagogical literature, as a component of tone quality in coordination, (Vennard, 1968)
richness (Seashore, 1938; Vennard, 1968) and vibrancy (Miller, 1996). Vibrato was
8
advocated as desirable in singing as a component of tone quality, justified by its
prevalent use by singers and instinctive acceptance by listeners. Vibrato is regarded,
especially in pedagogical circles, as a result of good singing, and, when successful,
almost imperceptible to the human ear. It is accepted in the singing literature that poor
vibrato is indicative of poor technique (Miller, 1996; Vennard, 1968) and inferior sound
quality. In classical singing, sound without vibrato, or straight-tone, has been described
as dull or spread (Vennard, 1968) and lacking freedom, power (Miller, 1996) and ring
(Vennard, 1968). A delayed vibrato onset indicated a faulty technique and an unnatural
voice quality (Miller, 1996).
Researchers have used acoustic, physiological and perceptual methods to study vibrato.
Vibrato is quantifiable by rate, the number of cycles per second, extent, the fluctuation
of pitch above and below the mean pitch, and onset, the delay from the onset of
phonation until the first vibrato cycle (Sundberg, 1988). Despite extensive research
(Prame, 1994, 1997; Shipp, Sundberg, & Hadlund, 1984; Sundberg, 1995a; Vennard,
1968) researchers have not reached consensus on what features constitute ideal vibrato.
Seashore (1932) defined good vibrato as “a pulsation of pitch, usually accompanied
with synchronous pulsations of loudness and timbre, of such extent and rate as to give a
pleasing flexibility, tenderness and richness to the tone” (Seashore, 1932, p. 7).
Seashore categorised vibrato as a musical ornament, invaluable for investigation
because it was used across all tones and impacted on tone quality. Using recorded
voices of famous singers, Seashore reported that vibrato was present in 95% of trained
singers’ voices and described it as a vocal ornament performed on every note. Later
9
studies investigate vibrato as an intrinsic component of classical vocal quality, distinct
from embellishment or ornamentation.
Ideal vibrato parameters vary by musical era and musical genre. Further, acoustic and
reports of appropriate vibrato do not accord with perceptual judgments of listeners
(Rothman, Diaz, & Vincent, 2000) and many studies define undesirable or defective
vibrato perceptually (Miller, 1996; Vennard, 1968) and acoustically (Keidar, Titze, &
Timberlake, 1984; Seidner, Nawka, & Cebulla, 1995) rather than attempt to identify
vibrato in voices of quality.
Physiological studies of vibrato produced varying results. Vibrato has been attributed to
crico-thyroid oscillations and 4 of 5 singers showed consistent lateral crico-arytenoid
muscle activity, oscillating at the same rate as the singers’ vibrato (Hirano, Hibi, &
Hagino, 1997). Comparing vibrato and straight tone, Hirano, Hibi & Hagino (1997)
found that some singers either widened or narrowed the supraglottic space and
hypopharynx when they produced vibrato. Vibrato has been linked to movement of the
larynx and many researchers drew comparisons to pre-existent neurological or existent
physiological tremors (Titze, Solomon, Luschei, & Hirano, 1994).
Studies seek to test singers’ conscious modification of vibrato: in the communication of
emotion (Howes, Callaghan, Davis, Kenny, & Thorpe, 2004; Sundberg, Iwarsson, &
Hagegard, 1995); loudness or sound pressure level (Michel & Myers, 1991; Titze et al.,
1999); and as consequence of vocal training (Brown, Rothman, Morris, & Sapienza,
2001; Brown, Rothman, & Sapienza, 2000; Ekholm, Papagiannis, & Chagnon, 1998;
10
Mendes, Rothman, Sapienza, & Brown, 2003; Wapnick & Ekholm, 1997). No studies
have assessed the effect of specific technical instruction on vibrato parameters.
1.2.2 Spectral energy
Classical and operatic singers produce an energy boost in their vocal quality, which is
perceived as a sound quality and also enables them to be heard without amplification.
Bartholomew (1934) first acknowledged a high energy spectral reinforcement as a
component of good voice quality. Sundberg (1974) described this energy as the
“singer’s formant”; a clustering of formants F3, F4 and F5 causing an energy peak in 2-
4 kHz. Acoustic research proposed that this energy is most likely to occur when the size
ratio of the pharynx to larynx tube/inlet is 6:1, which is achieved by lowering the larynx
in classical vocal quality (Sundberg, 1974; Titze, 1998; Titze, 2001).
The peak of high energy band between 2 and 4 kHz depends on individual singer and
also on vocal ranges (eg. soprano to bass) (Dmitriev & Kiselev, 1979). Soprano voices
do not consistently demonstrate spectral reinforcement in 2-4 kHz comparable to male
singers (Bartholomew, 1934; Sundberg, 1977, 1995b; Weiss, Brown, & Morris, 2001).
Studies show that high voices demonstrated a spectral reinforcement between 2.7 - 3.5
kHz (mezzo-soprano to high soprano) (Dmitriev & Kiselev, 1979) or 2.6 - 4.5 kHz but
concluded that this energy band was not as significant for female singers as for male
singers (Weiss et al., 2001). However, more recent studies of professional female
sopranos at a high performing level identified a stronger high level energy boost around
3 kHz (Barnes, Davis, Oates, & Chapman, 2004; Sundberg, 2001; Thorpe, Cala,
Chapman, & Davis, 2001) which may be associated with performing level and
subsequently singers’ audibility in large opera theatres.
11
Many studies consider the increase in vocal energy above 2 kHz a result of training in
vocal projection (Löfqvist & Mandersson, 1987). However, vocal projection or carrying
power is an important objective for young singers (Vurma & Ross, 2000), but which
can be developed to the detriment of vocal quality. Energy boosts above 2 kHz have
been associated with classical singers of the highest renown (Barnes et al., 2004; Thorpe
et al., 2001). Methods of measuring the energy distribution in voices give an indication
of acoustic features seen in good voices. Although neither proves the presence of a
singer’s formant, it provides a tool by which to measure and compare spectral energy in
the voice, perhaps throughout the development of a singer’s career (Lundy, Roy,
Casiano, Xue, & Evans, 2000).
More recent research has proposed methods to quantify this high range energy. Omori,
Kacker. Carroll, Riley & Blaugrund (1996) devised a method of objectively rating voice
quality by measuring the difference between the height of energy peaks associated with
fundamental frequency and singing power peak or energy boost (0-2 kHz and 2-4 kHz).
This result was identified as the singing power ratio (SPR). They compared singers with
non-singers and found that singers produced a significantly smaller SPR than non-
singers. These singers’ voices were described perceptually as demonstrating more
“ringing quality” and “richness” than non-singers. Singers with ≥ 4 years of training
were more likely to demonstrate a smaller SPR than singers with < 4 years training.
While SPR was not intended to define singer’s formant, it does provide a way in which
to compare inter and intra-subject high energy values.
12
1.2.2.1 Long term average spectra
In current singing research, the long term average spectrum (LTAS) is widely used as a
simple, non-invasive analysis for musical timbre and vocal features, both in speech and
singing. LTAS have been used to assess the effects of this characteristic energy boost >
2 kHz over time. An LTAS is the mean of all spectra over time (Baken & Orlikoff,
2000), and a measure of the dB level of the time-average of the power of the acoustic
signal at each frequency. As a research tool, LTAS facilitates the complex acoustic
evaluation of music, and captures the quantity of information produced in an entire
musical excerpt (Jansson & Sundberg, 1975).
In classical or operatic voice, an LTAS presents an energy boost around 3 kHz and this
has been linked to carrying power over an orchestra or audibility in an opera theatre
(Barnes et al., 2004; Sundberg, 1974; Thorpe et al., 2001; Vurma & Ross, 2000).
Energy in the 2-4 kHz regions which clusters the third, fourth and fifth formants, so
close in frequency that they appear as a unified peak, and amplifies a classically trained
singer to be heard over an orchestra (Sundberg, 1974). This amplification in male
voices, has been referred to the “2800” (Bartholomew, 1934), because of the frequency
level of the peak, or centre level of the singer’s formant, but is not as readily identified
in female voices (Weiss et al., 2001).
Methods used to generate LTAS curves have followed rigorous time and musical
stimuli criteria in order to reduce the effect of these variables on the final plot. LTAS
studies adhere to the recommended 20-30 seconds length of LTAS (Jansson &
Sundberg, 1975; Sundberg, 2001), although studies vary from short segments of 20
13
seconds (Howard, Szymanski, & Welch, 2002; Kenny & Mitchell, 2004, in press) to
samples lasting over a minute (Barnes et al., 2004; Mitchell & Kenny, 2004). After this
time, the plot identifies certain consistent features contained in the sound over time, and
averages out short-term variations in phonetic or articulatory strategies. This method
evens-out spectral changes associated with phonetic structures and musical stimuli and
gives a profile of the voice quality (Löfqvist & Mandersson, 1987). Sundberg proposed
that the energy boost above 2 kHz utilizes a bandwidth of spectra unfulfilled by the
orchestral energy levels (Sundberg, 1974).
LTAS have been presented as visual cues for vocal qualities and vocal genres (Borch &
Sundberg, 2002; Cleveland, Sundberg, & Stone, 2001; Sundberg, 1974). However,
quantitative interpretation of LTAS attempts to reduce the information in LTAS to a
single meaningful number to capture this energy boost in classical singing. Löfqvist and
Mandersson (1987) calculated spectral tilt, the difference between the between low (0-2
kHz) and high range (2-4 kHz) energy using the regression line between the two
prominent energy peaks. Novak and Vokral (1995) evaluated the spectral tilt of singing
students and found a steeper regression line in male voices (-3.65 – -4.93) than female
voices (2.71 – -3.33). These measurements were used, like SPR (Omori et al., 1996), to
provide an objective assessment of vocal quality. Similarly, Thorpe, Cala, Chapman &
Davis (Thorpe et al., 2001) compared the areas of energy 0-2 and 2-4 kHz rather than
the level of the energy peaks. These measurements have confirmed increased vocal
projection in high level professional singers (Barnes et al., 2004; Thorpe et al., 2001).
14
1.3 Pedagogical directions
Twentieth century vocal pedagogues asserted the need for singing teachers’
understanding of vocal functioning (Appelman, 1967; Miller, 1998). While a scientific
understanding of the voice is not a prerequisite for developing a good voice, it can
capture and document vocal pedagogy for future generations. Miller pointed out that
“for scientific research to be valid and have practical value in the studio, teachers of
singing must be involved, knowledgeable and interested. [Their] input, in areas of
expertise best understood by voice teachers, is essential” (Miller, 1998).
Singing pedagogy and singing literature in the last 25 years incorporates the new voice
science literature. Singing pedagogy conferences play an important role in the
dissemination of ideas and discussion on singing and singing technique. Bodies such as
the National Association of Teachers of Singing (NATS) in the United States, the
Australian National Association of Teachers of Singing (ANATS) and the British Voice
Association (BVA) are pivotal to the understanding and involvement of singing
pedagogues in singing research. Journals devoted to voice are directly linked to these
associations: Journal of Singing to NATS; Logopedics Phoniatrics Vocology to BVA
and Australian Voice to ANATS. Such publications facilitate dissemination of voice
research directly to singing pedagogues.
Pedagogical literature reports and disseminates scientific findings on voice. However,
some vocal historians are sceptical about the role of empirical research in singing and
they challenge researchers to undertake research relevant to the singing studio. Stark
was moved to comment that “voice scientists do not generally concern themselves with
15
elaborate historical or pedagogical constructs like bel canto, nor do they put much stock
in resonance images, ‘placing the voice’, or other traditional techniques of voice
training. Rather, they try to isolate specific physiological, acoustical, and aerodynamic
aspects of the singing voice in the controlled environment of the voice research
laboratory” (Stark, 1999).
Few empirical studies classify the components of good sound quality in singing,
although perceptual studies seek to describe good singing (Wapnick & Ekholm, 1997).
Recent singing studies have focussed on the singer’s intentions to achieve a good sound
and identified differences in singing with and without emotional connection or
enhanced projection (Barnes et al., 2004; Foulds-Elliott, Thorpe, Cala, & Davis, 2000;
Sundberg, 1997; Thorpe et al., 2001). Such studies imply that singers can consciously
manipulate their sound production.
Are great voices a natural gift or can they be developed by sound pedagogy? What
acoustic and perceptual features of the singing voice define a voice of quality? Can
these vocal features be accurately measured and taught? If they can be taught, what are
the best techniques for achieving the best vocal quality? Internationally renowned vocal
pedagogue Richard Miller stated “in any age, the main duties of a teacher of singing,
with regard to technique, have always been chiefly to (a) analyse vocal problems and
(b) design proper solutions for them.”
While singing pedagogy has absorbed research of the past 50 years, it has not directly
addressed specific vocal tools or singing techniques. This thesis seeks to provide
16
stronger links between empirical research and pedagogy in the singing studio, and
recognises the importance of singing pedagogues in this enterprise. Applied research in
singing must address the goals of singing pedagogy and systematically assess and
document the acquisition of vocal mastery. This is the first body of work to examine an
individual singing technique.
Thus, this thesis is situated at the intersection between voice research and vocal
pedagogy. It is devoted to the exploration of one commonly used vocal training
technique in classical singers – that of “open throat”.
It is the synthesising nature of a thesis that makes the last written the first presented.
The key points are presented in the first paper which summarises the body of work
presented in the thesis and ensures that it is accessible to the singers and pedagogues to
whom it is addressed.
17
REVIEW PAPER
Mitchell, H. F., & Kenny, D. T., (in review). Open throat: acoustic and perceptual support for pedagogical practice. Journal of Singing.
18
The open throat technique 1Journal of Singing, in review
Open throat: acoustic and perceptual support for pedagogical practice
Helen F. Mitchell and Dianna T. Kenny
From the Australian Centre for Applied Research in Music Performance (ACARMP), Sydney Conservatorium of Music, The University of Sydney, New South Wales 2006, Australia
Journal of Singing; in review
Music and song permeate every facet of human experience and play a significant role in cultural and social life. How are beautiful singing voices developed? What features of the singing voice define vocal beauty? Are listeners reliable in their assessments of these features? Can these features be accurately measured and taught? This paper reports on the first body of work of its kind to define and assess one feature of the technique of the classical singing voice – open throat. Such is its prominence in vocal pedagogy, Richard Miller commented ‘It would be hard to find a voice teacher who recommended singing with a closed throat’ (Miller 1996b) (p.58). The fact that a technique is universally prac-ticed does not in itself provide justification for its continued use; it must be validated empirically. Accordingly, we examined the acoustic and perceptual features of the classical singing voice while singers used and then reduced usage of open throat in a series of five studies. We confirmed that open throat technique is a useful and valid pedagogical technique that enhances vocal quality in the female classical voice. The methods and procedures developed to assess open throat can be used to assess other pedagogical training techniques, thereby contributing to the science of voice and vocal pedagogy.
Helen Mitchell, Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music, The Uni-versity of Sydney, New South Wales, Australia 2006 Phone: 61 2 9351 9644; Fax: 61 2 9351 9540 Email: [email protected]
INTRODUCTION
“Open throat” is a term regularly used in the singing studio. It is a pedagogical concept trans-mitted through the oral tradition of singing. Mitch-ell and Kenny (Mitchell et al. 2003; Mitchell and Kenny 2004a, 2004b, in press; Kenny and Mitch-ell in press, 2004) examined the technique in the modern singing studio using qualitative, acoustic, perceptual and statistical analyses to relate singing instructions, terminology and spectra to the sound qualities produced by instruction in this technique and validated its role in singing pedagogy.
Achieving vocal mastery
How are beautiful singing voices developed? What features of the singing voice define a voice of quality? Can these features be measured accu-rately and more importantly, specifically taught? Historically, teaching and evaluating singing has been guided by an oral tradition in which peda-gogical techniques are handed down from one generation of singing teachers to the next. Today, empirical research into the singing voice has the potential to benefit the singing community by doc-umenting and systematically assessing the acqui-sition of vocal mastery.
Since the 1980s technology to measure vocal
acoustics has become steadily more sophisticated and has been increasingly used in experimental research, with the findings of such research now being incorporated into texts on singing (Miller 1996b; Nair 1999; Sundberg 1977; Thurman and Welch 2000). International authority Richard Miller asserts “It is the responsibility of the sing-ing teacher in a scientific age to interpret and ex-pand vocal traditions through the means of current analysis so that the viable aspects of tradition can be communicated in a systematic way” (Miller 1998,p. 299).
However, it is often difficult to see how the general principles established in scientific stud-ies can be applied to the subtleties of developing individual vocal quality. To date, few studies link the vocal strategies used by singers to acoustic studies and perceptual judgments by pedagogues. This makes it difficult to draw firm conclusions about the implications of these scientific studies for singing pedagogy as they rarely identify the teaching/learning approach of the subjects (Miller 1998; Callaghan 2000).
In this paper, we report the first body of work to track a singing technique as practiced in the sing-ing studio using a comprehensive approach to all aspects of singing research (Mitchell and Kenny
19
The open throat technique 2Journal of Singing, in review
2004a, 2004b, in press; Mitchell et al. 2003; Ken-ny and Mitchell in press, 2004). As it is pedagogi-cally informed and verified, this series of studies has the potential to substantially progress the fields of acoustic and perceptual assessment of voice in order to benefit the wider singing community and enhance pedagogical approaches.
DEFINING OPEN THROAT TECHNIQUE
Concepts relating to open throat can be traced throughout pedagogical and scientific singing liter-ature (Burgin 1973; Fields 1947; Monahan 1978). It is defined as a complex process that is both a pedagogical instruction and a perceived sensation or action that results in a specific sound quality. Use of the technique makes a difference to vocal quality. Indeed, Vennard defined open throat as the ‘condition agreed upon by most voice teachers as desirable for resonance’ (Vennard 1968, p.252). Current support for the use of open throat in sing-ing technique is widespread (Miller 1996a; Reid 1975, 1983). It elicits a sound quality which is per-ceived as resonant (Miller 1996b; Vennard 1968), round (Joiner 1998), free (Ware 1998), pure, (Ma-rafioti 1981) rich and warm (McKinney 1982) and is attributed to freedom from ‘constrictor tensions’ (Reid 1983)(p. 83). The sound quality is linked to vocal actions: the preparation to sing or inhala-tion (Hemsley 1998; Manèn 1987; Miller 1997b, 1997a); through the surprise breath or smelling the rose imagery (Hemsley 1998; Miller 1996b; Pu-ritz 1956); and visualizing space within the throat, through an ‘air-ball’ or ‘soap bubble’ (Herbert-Caesari 1951; Manèn 1987).
Do expert singing teachers actually agree on the definition of open throat technique, and if so, on what such a technique was meant to achieve? Subjective terminology used in singing pedagogy does not always indicate a specific vocal instruc-tion or action. Often terminology and meaning are not the same for each teacher. Communication of techniques in singing pedagogy can be improved by attempts to gain consensus on the use of ter-minology. This was the goal of our first study on open throat. We interviewed fifteen expert singing pedagogues to explore current thinking regard-ing terminology, pedagogy, sound quality and the perceived physiology associated with open throat technique (Mitchell et al. 2003).
Terminology of open throat
The majority of our fifteen vocal pedagogues interviewed agreed that open throat was essential
to good singing and more specifically to classical singing. Most included the technique as fundamen-tal in their singing training. Freedom, collar and depth were suggested as alternative terms to clarify meaning or to refer to an action. Pedagogues were aware of the need to tailor their terminology and instructions in the singing studio to each student’s vocal needs and learning styles. Despite the use of different terminology to describe the technique, there was consistency in the vocal instructions to achieve it and with respect to the sound qualities it produced. Pedagogues taught conscious control of open throat using laugh, sob, correct inhalation or maintaining the posture of inhalation.
Sound quality
Open throat produces a distinctive sound qual-ity recognised by most singing pedagogues. Table 1 [after (Mitchell et al. 2003)] identifies the most common terms used to describe the sound qual-ity. Pedagogues associated open throat with both a sound quality that was characterized by freedom, warmth and openness and an action that produced balance, coordination, evenness and consistency. The terms related to vocal ‘quality’ such as ‘warm’, ‘full’ or ‘round’, as well as the ‘functional’ terms such as ‘easy’ or ‘clean’ were used interchange-ably by participants in this study. Although the 15 participants offered 18 terms to describe sound quality, there were clear associations or similari-ties in their usage and application.
Physiological action
Singing pedagogues relied on their perceptions of sound qualities to determine the physiological processes at work in the production of the sound quality. Open throat was defined as a technique to maximize pharyngeal space and/or abduct the ventricular folds. Participants who consistently mentioned a presumed “abduction” or retraction of the false vocal folds or ventricular folds report-ed seeing the action of endoscopy and linked the technique to the term retraction. Those participants who used the term and technique of retraction did so to reduce constriction or tension to achieve a healthier sound quality as well as to achieve a spe-cific sound quality.
Recommendations
This study advanced previous work in clarifying terminology related to open throat (Burgin 1973; Fields 1947; Monahan 1978). In this study, peda-gogues’ use of open throat confirmed the perceived value of the technique. Pedagogues did not sepa-
20
The open throat technique 3Journal of Singing, in review
rate quality and function descriptors and seemed comfortable with their use interchangeably in the studio. The research highlighted the usefulness of qualitative research as a tool for the generation of research questions and clarification of terminol-ogy used to describe vocal quality and to encour-age the use of anatomically correct terms in vocal pedagogy. For example, de-constriction is a more accurate term than retraction.
Having established the widespread usage of the term ‘open throat’ and its practice in the singing studio, the next challenge was to assess whether open throat had identifiable acoustic and percep-tual characteristics.
ACOUSTIC AND PERCEPTUAL VERIFICATION OF OPEN THROAT TECHNIQUE
The musical community has become increas-ingly fascinated with the link between acoustic measures of vocal quality and perceptual judg-ments of listeners. Prior to the work of Mitchell and Kenny (Mitchell and Kenny 2004a, 2004b; Kenny and Mitchell 2004, in press), two studies examined this link. Wapnick and Ekholm (Wap-
nick and Ekholm 1997) established 12 generally accepted perceptual criteria for the assessment of voice quality in classical singing (appropriate vi-brato, color/warmth, diction, dynamic range, ef-ficient breath management, evenness of registra-tion, flexibility, freedom throughout vocal range, intensity, intonation accuracy, legato line, and resonance/ring). Ekholm, Papagiannis and Cha-gnon (Ekholm, Papagiannis, and Chagnon 1998) used four of these criteria (“appropriate vibrato”, “resonance/ring”, “color/warmth”, and “clarity/focus”) and related them to objective measure-ments taken from acoustic analysis of the voice signal. Both studies required listeners to focus on specific vocal dimensions, as well as making an overall judgment of vocal quality. In both stud-ies, analysis of the listener rating scales of vocal quality revealed that the specific dimensions out-lined above were collinear and hence likely to be tapping into a single underlying construct, that of (overall) vocal quality, thereby rendering individ-ual assessments on each dimension at least partial-ly redundant. While focusing listeners by using a number of criteria may improve the consistency of judges’ responses, (Wapnick et al. 1993), studies have reported very high correlations between all dimension of voice quality studied (Ekholm, Pa-
Pedagogue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 %Balanced/Coordinated □ □ □ □ □ □ □ □ □ □ □ □ □ □ 93Free □ □ □ □ □ □ □ □ □ □ 67Open □ □ □ □ □ □ □ □ 53Even/Consistent □ □ □ □ □ □ □ □ 53Warm □ □ □ □ □ □ □ □ 53With space □ □ □ □ □ □ □ □ 40Healthy/Safe □ □ □ □ □ 33Round □ □ □ □ □ 33Overtones/formants □ □ □ □ □ 33Easy/Flexible □ □ □ □ □ 33Clear □ □ □ □ 27Full □ □ □ 20Efficient □ □ □ 20With depth □ □ □ □ 20Clean □ □ □ 20Sexy/Juicy/Lusty □ □ 13Natural Voice □ □ 13Relaxed □ 7
Table 1: Sound qualities associated with open throat – pedagogues’ individual choices indicated by □.
21
The open throat technique 4Journal of Singing, in review
pagiannis, and Chagnon 1998; Robison, Bounous, and Bailey 1994; Wapnick and Ekholm 1997). All of these dimensions were found to converge with the overall judgment of vocal quality. It may not be possible to separate individual features of good singing from the overall perception of a “good voice”.
Assessing acoustic and perceptual characteristics of open throat
Our next challenge was to determine whether this pedagogical technique could reliably produce desirable acoustic and perceptual changes in voic-es.
Six advanced female opera students, [3 sopranos, 3 mezzo-soprano] were asked to sing in three con-ditions: ‘optimal’ (O), using maximal open throat, ‘sub-optimal’ (SO), using reduced open throat and loud sub-optimal (LSO), which was the same as SO but with the additional instruction to sing as loudly as in O. Singers performed three musical tasks: messa di voce (a crescendo-diminuendo on a single note of long duration) on three pitches across their range, portions of an aria (Mozart: Ridente la Calma, K 152, bars 1-27) and a lied (Schubert: Du bist die Ruh D. 776 Op. 59, No. 3, bars 54 to 80) (Figure 1a and b). These were cho-sen as they require vocal skill and technical mas-tery (Miller 1996b) within the capacity of tertiary level students of opera (Ekholm, Papagiannis, and Chagnon 1998).
Singers’ voices were recorded to CD (Marantz CDR 630) using a high-quality microphone (AKG C-477) positioned on a head boom a constant 7 cm distance from the singer’s lips. This ensured we recorded only the voice energy, not the room reflections. We calibrated each recording in order to compare each singers’ recording with the others at the same SPL, heard as ‘loudness’. Recordings were anaysed with Soundswell (Hi-tech, Sweden) and Cool Edit software.
PERCEPTUAL DIFFERENCES OBSERVED IN OPEN THROAT
In this study, we explored whether particular sound qualities were associated with one vocal technique, open throat.
We asked singing pedagogues to judge the sing-ing of our sample of female classical singers as O (maximal open throat) and SO (reduced open throat) in 48 messa di voce and 30 song samples (including 6 repeated samples to test judges’ re-
sponse reliability). Fifteen expert singing peda-gogues made a forced choice decision (O or SO) on each sample they assessed. Correctly identified responses were counted by condition (O/SO), by judge and by singer.
The majority of listeners correctly recognised the use of open throat when it occurred. Collec-tively, judges were accurate in their identification of O (81.1% correct) and SO (91.0% correct) in messa di voce samples. In the individual song samples, listeners identified the use of open throat in 84% of O samples, and 69% of SO [Table 2 after (Mitchell and Kenny in press)]. They were more likely to make a correct identification in the Mozart task (85% correct) than in the Schubert task (68%). Listeners identified >83% of Mozart O and SO, and Schubert O. They were least reli-able in judging Schubert SO (53%).
The 15 judges identified the experimental condi-tion in 372 of 450 messa di voce samples (82.7%). Twelve of fifteen judges were moderately consist-ent in their judgments (k≥0.600); and three judges
&b8
3
Sop ‰ ‰
R
œ .œ
œ j
œ
J
œ .œœ
J
œ
Ù œœœ œœ œ œ œ œ
œ œ
œ œ
J
Ϩ
œœœ
&b
6
œ œ œ œ œœ
r
œ
r
œ
‰ ‰ ∑ ‰ ‰ ≈
R
œ .œœ
j
œ ≈
R
œ .œ
œ
j
œ≈R
œ
&b
12œ œœœ.
J
œ
R
œ
J
œ‰ ≈ R
œ .œ
œ
J
œ
j
œ
j
œ
j
œ≈ R
œ .œ
œ
J
œ
j
œ
j
œ
j
œ≈
R
œ
&b
18
J
œ
J
œ ≈
R
œ
J
œ
j
œ ≈
r
œœ œœœ.
j
œ
r
œ
j
œ
j
œ ≈
R
œ
J
œ
J
œ ≈
R
œ
J
œ
j
œ ≈
r
œ
&b
24
œ œœœ.
j
œ
rK
œ
œ
U
J
œ# œ œœ
œ œœ
œ œœœ
‰
≈
Ri - den - te la cal - ma nell' al - ma si_ des - ti, nell'
al - ma si_ desti; ne res - ti un seg - no di
sde-gno_e ti - mor. Ri - den - te la cal - ma nell' al - ma si des - ti; ne
res - ti più seg - no di sde-gno_e ti - mor, res - ti più seg - none
sde-gno_e ti - mor,
di
sde - gno_e ti - mor.di
&b
b
b 8
3
SopJ
œ .
J
œ
R
œœb
J
œ œb
J
œ œ
J
œ œ
J
œ
&b
b
b
6.œ
.œ
∑ œ œœ œ.œ œœ j
œ œ œœ œn œ œ
&b
b
b
12
œj
œ
∑ ∑
J
œ .
J
œ
R
œœb
J
œ
œb
J
œ
&b
b
b
18œ
J
œ œ
J
œ .œ.œn
∑ œ œœ œ.œ œ
&b
b
b
24œ j
œ œ œœ œn œ œ
.œ œ ‰
Dies Au - gen - zelt von dei - nem Glanz Al - lein er-
hellt, O füll es ganz, O füll es
ganz. Dies Au - gen - zelt von dei - nem
Glanz Al - lein er- hellt, O füll es
ganz, O füll es ganz!
Figure 1b: The Schubert Task, Du bist die Ruh D. 776 (Op. 59, No. 3), bars 54 to 80.
Figure 1a: The Mozart Task, Ridente la Calma, K 152, bars 1-27.
22
The open throat technique 5Journal of Singing, in review
were inconsistent in their judgements. Listeners demonstrated reliability (fair to highly consistent) in their judgements through the duration of the song task, repeating their judgement of O or SO in an average of 86% of the six repeated samples in the perceptual test.
Conclusions
Listeners recognised a specific vocal quality in singers’ sound that they associated with the use of open throat technique. These findings suggest that there is a specific vocal quality in classical singing associated with the use of open throat technique, which is a perceptual reality to sing-ing pedagogues. Perceptual verification of a single vocal technique indicates that other pedagogical strategies can be assessed and evaluated making it feasible to track the methods by which good sing-ers are trained and which produce the most effec-tive outcomes in terms of vocal quality.
Singers rely on expert listeners’ judgments in auditions, competitions and examinations. Re-search indicates that listeners show some degree of reliability and consistency in their perceptual judgments of timbre, between vocal genres (from opera to music theatre) (Sundberg, Gramming, and Lovetri 1993), and between good and poor vocal and instrumental performance (Ekholm, Papagi-annis, and Chagnon 1998; Geringer and Madsen 1998; Saunders and Holahan 1997; Wapnick et al. 1993), in the assessment of excellence in overall voice quality (Wapnick and Ekholm 1997) and in rankings and ratings of performers in competi-
tive situations (Davidson and Da Costa Coimbra 2001).
ACOUSTIC CHARACTERISTICS OF OPEN THROAT TECHNIQUE
Having verified that open throat technique has reliable perceptual qualities, we then investigat-ed whether these were associated with consist-ent acoustic characteristics. A number of voice qualities [loudness, vibrato and long term average spectra (LTAS)] were assessed for their acoustic features. .
Loudness
Pedagogues suggested that the use of open throat resulted in ‘loudness’ (Mitchell et al. 2003). We used SO and LSO in experimental conditions to demonstrate that voice quality changes from O to SO were not simply the result of changes (i.e. re-ductions) to sound pressure levels (SPL) (Foulds-Elliott et al. 2000; Rossing, Sundberg, and Tern-strom 1986). A pilot test found that LSO did not produce a sufficiently discernible voice quality to be reliably distinguished from SO and was there-fore not included in the perceptual studies.
Vibrato
Vibrato is taken for granted as an intrinsic qual-ity of the classical singing voice and in pedagogi-cal literature, as a component of tone quality in coordination (Vennard 1968) richness (Seashore 1938; Vennard 1968) and vibrancy (Miller 1996b). Despite acoustical, physiological and perceptual studies (Vennard 1968; Prame 1994, 1997; Shipp, Sundberg, and Hadlund 1984; Sundberg 1995), it has been difficult to define its most desirable parameters. Consistent vibrato occurring within specified parameters, has been associated with a beautiful sound (Robison, Bounous, and Bailey 1994) and to listeners’ overall preference (Ekholm, Papagiannis, and Chagnon 1998). In the classical singing literature, a steady and even vibrato is universally promoted (Miller 1996b) while poor vibrato is considered indicative of poor technique (Vennard 1968; Miller 1996b) and inferior sound quality. In classical singing, sound without vibra-to, or straight-tone, has been described as dull or spread (Vennard 1968) and lacking freedom, pow-er (Miller 1996b) and ring (Vennard 1968). In fact, delay in vibrato onset is argued to be indicative of a faulty technique and an unnatural voice quality (Miller 1996b).
Few studies compare the same singers’ vibrato in
Task Hit % Miss %
All O 84% 16%
All SO 69% 31%
Mozart O 86% 14%
Mozart SO 84% 16%
Schubert O 83% 17%
Schubert SO 53% 47%
Table 2: Song responses. Percentages of correct (hit) and incorrect (miss) responses to the song task, by condition (optimal and sub-optimal) and by condition and task (Mozart and Schubert).
23
The open throat technique 6Journal of Singing, in review
different tasks. When they do, they find it changes across musical styles (Easley 1932; Hakes, Shipp, and Doherty 1987), as a result of emotion (Sund-berg 1997; Rothman and Arroyo 1987; Gabriels-son and Juslin 1996), drama (Sundberg, Iwarsson, and Hagegard 1995) or indeed variations in ‘loud-ness’ (SPL) (Titze et al. 1999) or dynamics (Prame 1994). Our study is the first to assess the impact of a specific singing technique i.e. open throat on vibrato. As pedagogues associated evenness and consistency with the technique, we hypothesized that frequency modulations associated with vi-brato rate, extent and onset would vary outside acceptable or desirable parameters for SO and LSO compared to O, that is rate (VR) would be less consistent, extent (VE) would be reduced and onset (VO) would be delayed.
Vibrato results
We demonstrated reliable differences in vibrato parameters as a result of varying the degree to which singers applied the technique of open throat. Figure 2 [after (Mitchell and Kenny 2004b)] illus-trates the changes to vibrato parameters observed in spectrographs of the three experimental condi-tions. Hypotheses were confirmed for vibrato ex-tent and onset, that is, a reduction of open throat technique for these singers produced a significant decrease in VE in SO/LSO and a significant in-crease in VO in both SO/LSO. There was no statis-tically significant change for VR. However, visual inspection of the spectrographs (Figure 2) shows that reduction of open throat in the SO and LSO conditions was associated with greater irregularity of the vibrato pattern compared to O. There were no significant differences between SO and LSO on any of the vibrato parameters.
As vibrato is considered a key indicator of good singing, these findings suggest that open throat is important to the production of a good sound in classical singing. Since vibrato parameters largely define good singing technique in the literature, open throat would appear to be an essential ele-ment of sound vocal pedagogy. Inappropriate vibrato is indicative of poor singing in general; therefore further acoustic tests were required to test the differences in timbre.
Long-term average spectra (LTAS)
LTAS is widely used to represent singers’ sound (Sundberg 1974; Borch and Sundberg 2002) and its different vocal qualities based on energy changes that occur during different vocal tasks. Early re-searchers in this field have used LTAS to develop
exemplars of voice types (Sundberg 1974) or sing-ing genres (Cleveland, Sundberg, and Stone 2001; Borch and Sundberg 2002). LTAS curves have also been used to differentiate male and female voices (Mendoza et al. 1996), to note differences between singing and speaking voices (Barrichelo et al. 2001), solo and choral voice (Rossing, Sun-dberg, and Ternstrom 1987) and pop or country from opera singers (Borch and Sundberg 2002; Cleveland, Sundberg, and Stone 2001). Some LTAS have become exemplars of particular vocal qualities and of voices of quality. In classical or operatic voice, for example, an LTAS presents an energy boost around 3 kHz and this has been linked to carrying power over an orchestra or audibility in an opera theatre (Sundberg 1974; Barnes et al. 2004; Thorpe et al. 2001; Vurma and Ross 2000).
In research, we perform two simple measure-ments on LTAS. Figure 3 illustrates how the ra-tio measurements of singing power ratio (SPR) (Omori et al. 1996) and energy ratio (ER) (Thorpe et al. 2001) are derived from the LTAS curve. From the LTAS plots, the highest peaks in the 0-2 kHz and 2-4 kHz regions were labeled P1 and
0
2000
4000
6000
1 54320
0
2000
4000
6000
1 54320
1 543200
2000
4000
6000
Freq
uen
cy(H
z)
Time (seconds)
Time (seconds)
Time (seconds)
Freq
uen
cy(H
z)Fr
equ
ency
(Hz)
a
c
b
Figure 2 a-c: Examples of vibrato from subject 3 singing ‘Von deinem Glanz’ in each of the three conditions, optimal (O), sub-optimal (SO), and loud sub-optimal (LSO).
24
The open throat technique 7Journal of Singing, in review
P2 respectively and the areas below the peaks, A1 and A2. The SPR (Omori et al. 1996) is the differ-ence between the energy peaks P1 and P2 while the ER (Thorpe et al. 2001) is the difference in energy area between 0-2 kHz and 2-4 kHz. A low ER or SPR represents a greater reinforcement in the 2-4 kHz region and better balance between the spectral energy 0-2 and 2-4 kHz. When there is less reinforcement in the 2-4 kHz region, ER re-sults follow SPR.
As inconsistent vibrato is considered indicative of poor singing, it was hypothesized that testing the energy distribution in our singers’ voices in each condition would identify the timbral changes associated with open throat. Hypotheses were gen-erated regarding the LTAS plots in O compared to SO: that peaks P1 (0-2kHz) and P2 (2-4kHz) would be reduced, that overall LTAS shape would demonstrate smaller or multiple energy peaks above 2 kHz in SO/LSO compared to O. Differ-ences between spectral peak height (SPR: singing power ratio,) and spectral area (ER: energy ratio) between the 0-2 and 2-4 kHz frequency ranges were performed to assess the effect of open throat on carrying power in the voices.
LTAS results
Visual inspection of long term average spec-tra (LTAS) confirmed differences between O and SO/LSO, and the O condition produced a rounder peak between 0-2 kHz indicating a warmer sound quality compared to SO (Figure 3). Despite these
findings, there was no significant difference in measurements of SPR (peak height) (Omori et al. 1996) or ER (peak area) (Thorpe et al. 2001) be-tween O and SO/LSO. There were however, sig-nificant differences between SO and LSO for P2, SPR and ER but hypotheses were not confirmed for O. These findings did not accord with differ-ences in vibrato extent and onset between O and SO/LSO.
As these results were not consistent with the vi-brato findings they suggest that while LTAS pro-vides information on energy distribution, meas-urements performed on the LTAS were unable to differentiate between experimental conditions, whereas the human ear produced the most reliable assessment of vocal quality. Plotting the differenc-es between O and SO/LSO pairs of LTAS clearly indicates the areas of spectral change (Figure 4). This appears to be the most sensitive measure of energy distribution differences between condi-tions, but we have no way of capturing and quan-tifying such data at this time.
DO ACOUSTIC MEASURES MATCH PERCEPTUAL JUDGMENTS?
The conflicting results from the perceptual and acoustic studies of open throat technique high-lighted potential problems with these conven-tional analyses as measures of vocal quality. In the final study, we compared perceptual ratings of vo-cal quality with acoustic measures performed on
Figure 3: Long term average spectra of singer 3 singing Mozart’s Ridente la calma in three experimental con-ditions, optimal, sub-optimal and loud sub-optimal (O, SO, LSO). Singing power ratio (SPR) is the difference between the spectral peaks (P1 and P2) and energy ratio (ER) is the difference between the area under the peaks (A1 and A2).
30
50
70
90
110
0 2000 4000 6000Frequency (Hz)
SPL(dBat7cm
) Optimal
Sub-optimal
Loud sub-optimal
= SPR
P1 P2
A2
A1
25
The open throat technique 8Journal of Singing, in review
LTAS to assess whether SPR and ER were suffi-ciently sensitive to evaluate LTAS or vocal timbre and whether acoustic measures can predict a voice of quality?
Characteristic LTAS shapes, particularly of clas-sical or operatic voices of performers of the high-est regard have been accepted as an accurate visu-al representation of good sound. Operatic singing is positively associated with an increase in energy between 2-4 kHz (Omori et al. 1996; Thorpe et al. 2001; Vurma and Ross 2000; Barnes et al. 2004). While measures of SPR (Omori et al. 1996) and ER (Thorpe et al. 2001) are not intended to give information about singer’s formant, they should illustrate differences in energy distribution, or car-rying power. However, Vurma & Ross (Vurma and Ross 2000) found that increased energy above 2 kHz did not necessarily represent vocal quality judged perceptually.
To test the value of acoustic measures, we matched pedagogues’ perceptual ratings to acous-tic measures of each perceptual sample. Perceptual scores, SPR and ER were rank ordered for overall quality as defined in the singing literature (Omori et al. 1996; Thorpe et al. 2001; Barnes et al. 2004;
Vurma and Ross 2000). We then compared per-ceptual rankings with rankings of acoustic meas-ures (SPR and ER) to assess whether the acoustic characteristics matched the perceptual judgments of overall timbre.
Table 3 [after (Kenny and Mitchell 2004)] presents the rankings of each singer and the re-spective musical task and experimental condition, ordered by perceptual rank from highest to lowest. While we found the expected significant relation-ship between SPR and ER, there was no relation-ship between perceptual rankings of vocal quality of singers based on SPR or ER. LTAS measures were not consistent with perceptual ratings of vo-cal quality, and could not therefore be used to de-fine a voice of quality in our studies.
Measurements of comparative energy (SPR and ER) were inconclusive indicators of vocal qual-ity. A science of the singing voice cannot progress without addressing the problem inherent in accept-ing long-term average spectra as analogues for vo-cal quality without providing a link between per-ceptual cues and a visual representation of ‘quality in singing. This presents a major challenge to the current wisdom that acoustic parameters of voice
-20
0
20
40
60
80
100
0 1000 2000 3000 4000 5000
Optimal
Energydifference
Frequency (Hz)
SPL(dBat7cm
)
6000
Sub-optimal
Figure 4: Example of singer 1’s LTAS plots in the optimal and sub-optimal condition for the Mozart task. Below is an example of the calculation of energy difference between the optimal and sub-optimal conditions.
26
The open throat technique 9Journal of Singing, in review
must emulate the established acoustic norms in or-der to achieve overall vocal excellence.
In Figure 5a-d, we compared exemplars of LTAS sampled from voices ranked high, middle or low on the perceptual rating scale. LTAS of the high-est-ranking singers [Figure 5a-b after (Kenny and Mitchell in press)] showed an increase in energy between 2-4 kHz, whereas LTAS of middle and low ranked singers (Figure 5c-d) lacked the uni-fied peak of energy increase above 2 kHz. Previ-ous similar findings (Borch and Sundberg 2002; Miller 1998; Vurma and Ross 2000) have inter-preted these visual cues in singers’ LTAS as a good sound. However, as third equal ranking singers 1 and 3, singing in the O condition produced this
unified energy peak > 2 kHz (the most ‘masculine’ LTAS shape) they should have had the ‘best’ voic-es perceptually. However, their LTAS were differ-ent to the highest-ranking singer (singer 5), who had a wider distribution of energy above 2 kHz. The SO plots of the lower ranked singers (Fig 3d) showed a spectral roll-off more consistent with plots for speech than singing.
CONCLUSIONS AND FUTURE DIRECTIONS
In this series of studies, we have demonstrated that open throat is a technical and perceptual re-ality to singers and singing pedagogues and pro-duces a specific vocal quality in classical singing that can be reliably identified by expert listeners. Through qualitative, acoustic and perceptual stud-ies, we have defined the term ‘open throat’ as a technique, an action and a sound quality.
While we know that singing improves over time and through training specific components of the training that produce improvements in vocal qual-ity have not hitherto been isolated (Ekholm, Papa-giannis, and Chagnon 1998; Mendes et al. 2003; Robison, Bounous, and Bailey 1994; Vurma and Ross 2000; Wapnick and Ekholm 1997). This series of studies presents a first attempt to study a single technique, to identify its perceptual and acoustic characteristics, and its effect on vibrato.
Vocal quality results from a complex combina-tion of acoustical parameters. To date, no single objective evaluation captures or characterizes vocal quality in a systematic way (Omori et al. 1996; Thorpe et al. 2001). We have demonstrated that acoustic analyses such as LTAS do not reli-ably match perceptual judgments by expert lis-teners and therefore cannot be used to define or predict vocal quality. We recommend that any fu-ture acoustic analyses or visual representations of voice must emulate the human ear. A new genera-tion of acoustic recording equipment and analysis software is emerging that has the potential to pro-vide much finer acoustic and psychoacoustic rep-resentations of vocal quality. Voice research will benefit from other areas of acoustic research (e.g. room acoustical quality (Nannariello, Osman, and Fricke 2002) and audio systems (Zwicker and Fastl 1999)) oriented to listener sensation and will use binaural measurements (using a dummy head microphone) to virtually place the listening sub-ject in the same acoustical environment as in the original recording(Møller et al. 1996). A singing quality model will be more appropriately derived
Table 3: Singer, task, condition and rankings for perceptual score, singing power ratio (SPR) and energy ratio (ER), sorted by perceptual score rank-ing from highest to lowest.
Perceptual Rank
SPR Rank
ER Rank Singer Task Condi-
tion
1 16 9 5 S O2 13 3 5 M O3 10 10 1 M O4 5 2 3 M O5 9 11 1 S O6 21 20 4 S O7 6 13 4 M O8 23 23 2 S O9 19 19 3 S O10 8 6 2 M O11 11 14 6 S O12 4 4 5 S SO13 3 12 1 M SO14 24 24 2 S SO15 15 8 6 M O16 7 7 5 M SO17 20 21 3 S SO18 18 18 6 S SO19 12 16 1 S SO20 1 5 3 M SO21 22 22 4 S SO22 17 15 2 M SO23 14 17 4 M SO24 2 1 6 M SO
27
The open throat technique 10Journal of Singing, in review
by representing listener experience.
REFERENCES
Barnes, J.J., P.J. Davis, J. Oates, and J. Chapman. 2004. The relationship between professional operatic so-prano voice and high range spectral energy. J Acoust Soc Am 116 (1):530–538.
Barrichelo, V.M.O., R.J. Heuer, C.M. Dean, and R.T. Sataloff. 2001. Comparison of singer’s formant, speaker’s ring and LTA spectrum among classical singers and untrained normal speakers. Journal of Voice 15 (3):344-350.
Borch, D Z, and J Sundberg. 2002. Spectral distribu-tion of solo voice and accompaniment in pop music. Logopedics Phoniatrics Vocology 27:37-41.
Burgin, John Carroll. 1973. Teaching singing. Metuchen, N.J.: Scarecrow Press.
Callaghan, Jean. 2000. Singing and voice science. San Diego, California: Singular Pub. Group.
Cleveland, T. F., J Sundberg, and R.E. Stone. 2001.
Long-Term-Average-Spectrum characteristics of country singers during speaking and singing. Journal of Voice 15 (1):54-60.
Davidson, J.W., and D Da Costa Coimbra. 2001. In-vestigating performance evaluation by assessors of singers in a music college setting. Musicae Scientiae 5 (1):33-53.
Easley, E. 1932. A comparison of the vibrato in concert and opera singing. In The vibrato, edited by C. E. Seashore. Iowa: University of Iowa Press.
Ekholm, E., G. C. Papagiannis, and F. P. Chagnon. 1998. Relating objective measurements to expert evaluation of voice quality in Western classical sing-ing: critical perceptual parameters. Journal of Voice 12 (2):182-96.
Fields, V.A. 1947. Training the singing voice: an anal-ysis of the working concepts contained in recent contributions to vocal pedagogy. New York: King’s Crown Press.
Foulds-Elliott, SD., CW. Thorpe, SJ. Cala, and PJ. Davis. 2000. Respiratory function in operatic sing-ing: effects of emotional connection. Logopedics
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000Hz
Singer 5 Schubert OSinger 5 Mozart O
SPL(dBat7cm
)
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000Hz
SPL(dBat7cm
)
Singer 1 Mozart OSinger 3 Mozart O
Singer 1 Schubert O
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000Hz
SPL(dBat7cm
)
Singer 6 Schubert O
Singer 5 Schubert SO
Singer 1 Mozart SO
Singer 2 Schubert SO
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000Hz
SPL(dBat7cm
)
Singer 2 Mozart SO
Singer 4 Mozart SO
Singer 6 Mozart SO
a b
c d
Figure 5: LTAS of highest (a), 3= (b), middle (c) and lowest (d) perceptually ranked singers. Legends correspond to task, Mozart (M) or Schubert (S), and singer number.
28
The open throat technique 11Journal of Singing, in review
Phoniatrics Vocology 25 (4):151-68.Gabrielsson, A., and P.N. Juslin. 1996. Emotional Ex-
pression in Music Performance: Between the Per-former’s Intention and the Listener’s Experience. Psychology of Music 24:68-91.
Geringer, J.M., and C.K. Madsen. 1998. Musicians’ ratings of good versus bad vocal and string perform-ances. Journal of Research in Music Education 46 (4):522-534.
Hakes, J., T. Shipp, and E. T. Doherty. 1987. Acous-tic properties of straight tone, vibrato, trill and trillo. Journal of Voice 1 (2):148-156.
Hemsley, Thomas. 1998. Singing and imagination: a human approach to a great musical tradition. Oxford; New York: Oxford University Press.
Herbert-Caesari, Edgar F. 1951. The voice of the mind. London: Robert Hale.
Joiner, James Richard. 1998. Charles Amable Battaille: pioneer in vocal science and the teaching of singing. Lanham, Md.: Scarecrow Press.
Kenny, D.T., and H.F. Mitchell. 2004. Visual and audi-tory perception of vocal beauty: conflict or concur-rence? Paper read at 8th International Conference on Music Perception & Cognition (ICMPC8), August, at Evanston, IL, USA.
———. in press. Acoustic and perceptual appraisal of vocal gestures in the female classical voice. Journal of Voice.
Manèn, Lucie. 1987. Bel canto: the teaching of the clas-sical Italian song-schools its decline and restoration. Oxford: Oxford University Press.
Marafioti, P. Mario. 1981. Caruso’s Method of Voice Production. New York: Dover Publications. Original edition, 1922.
McKinney, James C. 1982. The Diagnosis and Correc-tion of Vocal Faults. Nashville, Tennessee: Broad-man Press.
Mendes, A.P., H. B. Rothman, C. M. Sapienza, and W. S. Brown, Jr. 2003. Effects of vocal training on the acoustic parameters of the singing voice. Journal of Voice 17 (4):529-543.
Mendoza, E, N Valencia, J Munoz, and H Trujillo. 1996. Difference in voice quality between men and women: use of the long-term average spectrum (LTAS). Jour-nal of Voice 10 (1):59-66.
Miller, R. 1998. The singing teacher in the age of voice science. In Vocal Health and Pedagogy, edited by R. T. Sataloff. San Diego, CA: Singular.
Miller, R. 1996a. Sotto Voce: Vocal Timbre in Piano Dynamic. Journal of Singing - The Official Journal of the National Association of Teachers of Singing 52 (5):23-24, 52.
Miller, Richard. 1996b. The structure of singing: sys-tem and art in vocal technique. New York: London: Schirmer Books.
———. 1997a. National schools of singing: English, French, German, and Italian techniques of singing re-visited. Rev. ed. Lanham, Md: Scarecrow Press.
———. 1997b. The Open Throat (La gola aperta). In On the Art of Singing, edited by R. Miller. New York: Oxford University Press.
Mitchell, H.F., and D.T. Kenny. 2004a. The effects of open throat technique on long term average spectra (LTAS) of female classical voices. Logopedics Pho-niatrics Vocology 29 (3):99-118.
———. 2004b. The impact of “open throat” technique on vibrato rate, extent and onset in classical singing. Logopedics Phoniatrics Vocology 29 (4):171-182.
———. in press. Can experts identify “open throat” technique as a perceptual phenomenon? Musicae Scientiae.
Mitchell, H.F., D.T. Kenny, M. Ryan, and P.J. Davis. 2003. Defining open throat through content analysis of experts’ pedagogical practices. Logopedics Phoni-atrics Vocology 28 (4):167-180.
Møller, H., M.F. M.F. Sørensen, C.B. Jensen, and D. Hammershøi. 1996. Binaural technique: do we need individual recordings? J. Audio Eng. Soc. 44:451-469.
Monahan, Brent Jeffrey. 1978. The art of singing: a compendium of thoughts on singing published be-tween 1777 and 1927. Metuchen, N.J.: Scarecrow Press.
Nair, Garyth. 1999. Voice tradition and technology: a state-of-the-art studio. San Diego: Singular Publish-ing Group.
Nannariello, J, M.R. Osman, and F.R. Fricke. 2002. Re-cent developments in the application of neural net-work analysis to architectural and building acoustics. Acoustics Australia 29:103-110.
Omori, K., A. Kacker, L. M. Carroll, W. D. Riley, and S. M. Blaugrund. 1996. Singing power ratio: quanti-tative evaluation of singing voice quality. Journal of Voice 10 (3):228-35.
Prame, E. 1994. Measurement of the vibrato rate of ten singers. J Acoust Soc Am 96 (4):1979-1984.
———. 1997. Vibrato extent and intonation in profes-sional Western lyric singing. J Acoust Soc Am 102 (1):616-621.
Puritz, Elizabeth. 1956. The teaching of Elisabeth Schumann. London: Methuen.
Reid, Cornelius L. 1975. Voice: psyche and soma. New York: Joseph Patelson Music House.
———. 1983. A dictionary of vocal terminology. New York: Joseph Patelson Music House.
Robison, C.W., B Bounous, and R Bailey. 1994. Vocal Beauty: A study proposing its acoustical definition and relevant causes in classical baritones and female belt singers. Journal of Singing 51:19-30.
Rossing, T.D., J Sundberg, and S. Ternstrom. 1986.
29
The open throat technique 12Journal of Singing, in review
Acoustic comparison of voice use in solo and choir singing. J Acoust Soc Am 79 (6):1975-1981.
———. 1987. Acoustic comparison of soprano solo and choir singing. J Acoust Soc Am 82 (3):830-836.
Rothman, H. B., and A. Antonio Arroyo. 1987. Acoustic variability in vibrato and its perceptual significance. Journal of Voice 1 (2):123-141.
Saunders, T.C., and J.M. Holahan. 1997. Criteria-spe-cific rating scales in the evaluation of High School instrumental performance. Journal of Research in Music Education 45 (2):259-272.
Seashore, Carl E. 1938. Psychology of music. New York and London: McGraw-Hill Book Company.
Shipp, T., J. Sundberg, and S. Hadlund. 1984. A model of frequency vibrato. Paper read at Transcripts of the Thirteenth Symposium: Care of the Professional Voice, at New York.
Sundberg, J. 1977. The acoustics of the singing voice. Scientific American (March):82-91.
Sundberg, J. 1974. Articulatory interpretation of the “singing formant”. J Acoust Soc Am 55 (4):838-44.
———. 1995. Acoustic and psychoacoustic aspects of vocal vibrato. In Vibrato, edited by P. H. Dejonckere, M. Hirano and J. Sundberg. San Diego: Singular Publishing Group.
———. 1997. Expressivity in singing. TMH-QPSR 2 (3):13-19.
Sundberg, J., P. Gramming, and J. Lovetri. 1993. Com-parisons of pharynx, source, formant, and pressure characteristics in operatic and musical theatre sing-ing. Journal of Voice 7 (4):301-10.
Sundberg, J., J. Iwarsson, and H. Hagegard. 1995. A
singer’s expression of emotions in sung performance. In Vocal Fold Physiology: Voice Quality Control, ed-ited by O. Fujimura and M. Hirano. San Diego, Cali-fornia: Singular Publishing Group, Inc.
Thorpe, CW., SJ. Cala, J. Chapman, and PJ Davis. 2001. Patterns of breath support in projection of the singing voice. Journal of Voice 15 (1):86-104.
Thurman, L, and G. F. Welch, eds. 2000. Bodymind & voice: foundations of voice education. Rev. ed. ed. Collegevile, Minn: VoiceCare Network.
Titze, I. R., R. Long, G. I. Shirley, E. Stathopoulos, L. O. Ramig, L. M. Carroll, and W. D. Riley. 1999. Messa di voce: an investigation of the symmetry of crescendo and decrescendo in a singing exercise. J Acoust Soc Am 105 (5):2933-40.
Vennard, William. 1968. Singing: the mechanism and the technic. 5th ed. New York: Fischer.
Vurma, A., and Y. Ross. 2000. Priorities in voice train-ing: carrying power or tone quality. Musicae Scien-tiae 4 (1):75-93.
Wapnick, J., and E. Ekholm. 1997. Expert consensus in solo voice performance evaluation. Journal of Voice 11 (4):429-36.
Wapnick, J., P Flowers, M Alegant, and L Jasinskas. 1993. Consistency in piano performance evaluation. Journal of Research in Music Education 41 (4):282-292.
Ware, Clifton. 1998. Basics of vocal pedagogy: the foundations and process of singing. New York: Mc-Graw-Hill.
Zwicker, E, and H Fastl. 1999. Psychoacoustics: facts and models. Berlin: Springer.
30
2 Defining open throat
Pedagogical literature advocates the use of open throat in singing. Historically, the term
open throat can be traced throughout vocal pedagogy literature (Burgin, 1973; Fields,
1947; Monahan, 1978) as an interpretation of freedom or lack of tension in the area of
the throat, resulting in a lack of constriction and a better tone quality. Current support
for the use of open throat in singing technique is widespread (Miller, 1996; Reid, 1975,
1983). Open throat is associated with techniques for breath and as factor of overall
resonance (Hemsley, 1998; Miller, 1996) and as an enhancement to vocal quality
(Miller, 1996; Reid, 1983; Vennard, 1968).
Terminology depicting space or openness in singing is familiar in both the singing
literature and in the singing studio. However, the definition and interpretation of the
instructions to elicit space or openness vary across singing pedagogues. In singing
pedagogy texts, there is diverse opinion as to the need and indeed success of the use of
open throat technique. The technique is used to correct constriction, necessary for the
production of certain vocal qualities and genres, as an enhancement to the sound or a
specific sound quality. Actions or gestures such as “yawn”, “sob”, “laugh” and “surprise
breath” are frequently found in the literature and may represent open throat technique.
Open throat has been associated with certain schools of singing, as a sound quality, in
certain national schools of singing throughout the literature. Different nationalistic
schools value vocal resonance for different sound qualities (Miller, 1997): the German
school favours conscious alteration of the pharyngeal size, different to speaking;
31
whereas the Italian school advocates less manipulation and a more natural sensation of
openness.
In the first study, we assessed the degree of consensus amongst singing pedagogues
regarding the definition of, and use in the singing studio of the technique called open
throat. Open throat was defined in other terms, such as “freedom”, “space”, “collar” or
“retraction”. Results indicated that all fifteen pedagogues described open throat
technique as fundamental to singing training and were positive about the sound quality
it achieved, especially in classical singing. Its use in singing has been characterised with
a sound quality which is free, warm and open. Open throat has been defined as a
technique to maximize pharyngeal space and/or abduct the ventricular folds.
This pedagogical approach has been well received in the scientific literature. Professor
David Howard, Editor of Logopedics Phoniatrics Vocology, noted in his editorial
review of our paper that our innovative study on open throat technique sat comfortably
with other international studies that examined physiological explanations for vocal
quality and acoustic synthesis of yawn and twang technique by Titze’s American
research group (Howard, 2003; Titze, Bergan, Hunter, & Story, 2003).
The results of the study were reported to the singing pedagogy community at the
Australian National Association of Teachers of Singing (ANATS) National Conference,
October 2002, Melbourne [Mitchell, H. F., Kenny, D. T., Ryan, M. Open throat:
pedagogical perceptions and practices].
32
PAPER 1
Mitchell, H. F., Kenny, D. T., Ryan, M., & Davis, P. J. (2003). Defining open throat through content analysis of experts' pedagogical practices.
Logopedics Phoniatrics Vocology, 28(4), 167-180
33
Defining ‘open throat’ through content analysis ofexperts’ pedagogical practices
Helen F. Mitchell, Dianna T. Kenny, Maree Ryan and Pamela J. Davis
From the Australian Centre for Applied Research in Music Performance (ACARMP), Sydney Conservatorium of Music,The University of Sydney, New South Wales 2006, Australia
Received 21 October 2002. Accepted 4 September 2003.
Logoped Phoniatr Vocol 2003; 28: 167�/180
‘Open throat’ is a term regularly used in the singing studio, but agreement across pedagogues as to its definition and functionhas not yet been assessed. Fifteen expert singing pedagogues participated in a qualitative study involving a semi-structuredinterview to explore current thinking regarding terminology, pedagogy, sound quality and the perceived physiology to achieveopen throat, as used in the singing studio. Most teachers included the use of the technique as a fundamental in singingtraining, and were positive about the sound quality it achieved, especially in classical singing. The purpose of the techniquewas described as a way of maximizing pharyngeal space and/or achieving abduction of the ventricular folds.
Key words: associated physiology, singing pedagogy, sound quality descriptors, terminology.
Helen F. Mitchell, Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music,University of Sydney, NSW 2006, Australia. Tel.: �/61 2 9351 1386. Fax: �/61 2 9351 9540. E-mail: [email protected]
INTRODUCTION
Historically, teaching and evaluating singing has been
guided by an oral tradition in which pedagogical
techniques are handed down from one generation of
singing teachers to the next. The oral tradition, with
its accompanying musical terminology, particularly
that used to describe performance quality, can be
confusing and indeed mystifying for those who use
and practise it (1).
Many of the terms used to describe ‘openness’,
‘space’ or ‘freedom’ as a characteristic of the singing
voice have been used inconsistently, and terminology
associated with sound quality among expert singing
pedagogues is diverse and idiosyncratic. Seashore
recommended that ‘musicians scrap a mass of the
current synonyms for tone quality, because these
words do not connote any demonstrable differences
in content. The diversity of words simply adds to the
confusion’ (2, p. 111). There is, and has always been a
recognition of an overall good sound and of its
components (3), but in this case of openness there is
need for clarification, especially in the area of profes-
sional classical singing.
Poor singing has been described as sound ‘getting
stuck’ in the throat (4). Awareness of the throat as a
technique in good singing, or the lack of awareness as
an explanation for bad singing, has appeared through-
out the singing pedagogy literature. It is believed that
sound should flow freely, with nothing impeding it,
like a ‘chimney’ for air (5), or an unobstructed passage
for a sound beam (6). Failure to achieve this free flow
has been perceived to have serious consequences for
the singer and the sound. ‘If the scholar should have
any defects, of the nose, the throat, or of the ear, let
him never sing but when the master is by, or somebody
that understands the profession in order to correct
him, otherwise he will get an ill habit, past all remedy’
(7, p. 187).
Concepts relating to open throat can be traced
throughout pedagogical and scientific singing litera-
ture. Vennard defines open throat as the ‘condition
agreed upon by most voice teachers as desirable for
resonance’ (8, p. 252). That is, the use of open throat
makes a fundamental difference to vocal quality. Titze
(9) and Titze and Story (10) described a ‘wide
pharynx’ as an acoustic enhancement to the first
formant and to the overall sound. The difference to
the sound has been described as rounder, larger (11),
free (12), pure (13), rich, warm (14), and concentrated
energy (15).
�ORIGINAL ARTICLE �
# 2003 Taylor & Francis. ISSN 1401-5439 Logoped Phoniatr Vocol 28
DOI: 10.1080/14015430310018856 34
At least two views of open throat have been
propagated in the pedagogical literature: the first
focuses on the action at laryngeal level (16), and the
other emphasizes pharyngeal involvement, at a level
extending to the soft palate or velum (6, 8). These
differing perspectives regarding the vocal mechanism
required to achieve open throat have inevitably
resulted in different teaching strategies and terminol-
ogy. Examples include actions involved in inhalation,
the surprise breath and smelling the rose (5, 17, 18).
Visualization of space within the throat, through an
‘air ball’ or ‘soap bubble’ (6, 19), and various other
configurations to alter the shape of the pharynx are
common throughout the literature. The lowered lar-
ynx, often along with widening of the oropharynx, is
also associated with open throat (6, 20, 21). Yawn is
both advocated (15, 21�/3), and also seen as poten-
tially distorting to the sound if extended too far by
creating tension and imbalance (24, p. 81). Forms of
laughing as a technique to achieve an open throat were
advocated in the late twentieth century (16, 24). Estill
(16) and Citardi et al . (25) advocate a method
involving retraction of the false vocal folds or abduc-
tion of the ventricular folds away from the midline.
Current pedagogical thought and teaching recom-
mend the use of more than one term to define the
relative openness of the pharynx and adjacent struc-
tures. Open throat and retraction are recognized in the
literature (16, 24) as well as in teaching, and a wealth
of other equivalents exist, such as throat widening,
space, or space at the back. From this point on, open
throat will be used as a generic term for all terminol-
ogy and technique descriptors.
Wapnick and Ekholm (26) advocated consistency in
the use of common terminology across singing peda-
gogues. They initiated links in the discovery and
application of terminology, perceptual judgment and
acoustic evaluation. Ekholm et al . (27) found striking
connections were made across judges when one sound
quality appeared dependent on another [resonance/
ring and clarity/focus (.94) and colour/warmth with
appropriate vibrato (.93)]. The methodology in this
study builds on these works by linking the study of a
technique with its consequent sound quality, to
ascertain the relationship between the desired sound
quality and the technique used to achieve it. This
study sought to determine agreement, links and
interdependency of descriptors on technique and
perceptual qualities in describing technique in singing.The focus of the study is on terminology and
techniques of teaching open throat as practised by
experienced singing pedagogues. In particular, we were
interested in the perceived role of open throat in
classical singing, and whether open throat was thought
to improve the overall sound quality.
METHODS
Participants
Participants were 15 experienced singing teachers with
established singing studios. All participants were
known to the researchers via affiliations with keymusic centres in Australia and were personally invited
to participate by one of the authors. Selection criteria
for inclusion in the study were based on reputation
within the singing profession, determined by renown
or prominence according to the calibre of the students
in the singing studio. Specifically, the participants:
1. taught singing at tertiary level in universities or
conservatoria;
2. had a successful private studio, defined as one
comprising of students who performed and audi-tioned at a big city or national level for positions
in opera companies, music theatre or professional
musical ensembles;
3. worked primarily in Australia or the United
Kingdom;
4. represented the range of current singing pedago-
gies in their teaching methods (investigators aimed
specifically to include both exponents and criticsof the Estill (16) method);
5. have had, or continue to have, a successful career
as a professional solo singer.
Participants were sent information about the project
and were invited to undertake a taped, semi-structured
interview regarding current vocal pedagogy, in parti-
cular, the terminology used by singing pedagogues to
explain singing techniques to their students. Ethical
approval to conduct the study was obtained from The
University of Sydney Ethics Committee. All the inter-views were conducted by the first author, and only the
first author was aware of the identity of participants.
Instruments
Questionnaire. Each participant completed a brief
questionnaire giving details of their teaching
experience and current singing students. The ques-
tionnaire was divided into three sections: demographicinformation, influences on singing pedagogy and
characteristics of their singing studio. Participants
were asked for information related to: age, profes-
sional education, number of years of teaching experi-
ence and proportion of time spent teaching or singing
professionally. They were also asked to name people
with whom they had studied, and workshops or
masterclasses they had attended that had influencedtheir teaching. Finally, pedagogues were asked to
classify themselves according to the level at which
168 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2835
their students performed. Pedagogues were asked
to use the Bunch and Chapman taxonomy (28) to
estimate the percentage of their students in eachcategory (superstar, international, national/big city,
regional/touring, local community, singing teachers,
full-time students of singing, amateur, child) and in
the primary genres (opera, contemporary music thea-
tre, musical theatre, concert/oratorio/recital).
Semi-structured interview. A semi-structured inter-
view was developed to assess the beliefs and attitudes
of singing teachers related to open throat (29). Thismethod of data collection is described by Smith (30).
It was considered appropriate to the goals of the
current study because it ‘facilitates rapport/empathy,
allows a greater flexibility of coverage, enables the
interview to enter novel areas, and tends to produce
richer data’ (30, p. 12).
The interview schedule opened with general ques-
tions about the teaching of singing, for example, ‘Howdo you achieve a good sound in singing?’ and ‘How do
you correct a problematic sound?’ No direct questions
on open throat were asked but the topic was discussed
in depth once the pedagogue initiated discussion.
Reference to a key term or to an associated term,
such as open throat, throat widening, retraction or
space at the back, provided the impetus to begin a
discussion of open throat. If the pedagogue did notmention open throat during the interview, the inter-
viewer asked the question, ‘What do terms like open
throat, throat widening, retraction and space at the
back mean to you?’
The interview schedule focused on elucidating
pedagogues’ view of open throat, using categories
and terminology derived from the literature. These
were:
1. Relationships between the four most commonly
used terms: open throat, throat widening, retrac-
tion and space at the back2. Preferred term/terms
3. Sound qualities associated with these terms
4. Required technique to achieve the sound quality
5. Relationships between terms, actions and sound
qualities
6. Physical sensations and presumed physiology
associated with the terminology
Procedure
The interview was pilot tested on an experienced
singing pedagogue and modifications were made
to increase the clarity of the questions where necessary.
The interview schedule was used as a guide; specificopen-ended questions were followed by individualized
prompts and probes (30). A lexicon of interview
prompts was used to facilitate the flow of ideas
in the interview, although care was taken not
to interrupt the participant (31). Participants were
encouraged to elaborate on ideas through the use
of agreement prompts from the interviewer. The
interviewer used clarifying prompts to refine partici-
pants’ statements as necessary. Researcher agreement
was established regarding the appropriate use
of prompts during the interview prior to commence-
ment.
Interviews were conducted either in person (N�/4)
or by phone (N�/11) at a time and place convenient
to the participant, and permission to record the
interviews was obtained before proceeding. Tapes
were marked by participant number and transcribed
with transcriptions sent back to respondents for
verification and final approval before they were
qualitatively analysed. At this stage, participants
were given the opportunity to clarify anything that
they had said.
Analysis
Transcripts were analysed using in-depth qualitative
analysis, detailed in Smith (30) to identify relation-
ships between terminology, theory and teaching prac-
tice for each pedagogue (32). Transcripts were read in
depth to generate a profile of the topic responses. They
were then similarly coded and common themes were
identified. Significant or representative statements
were highlighted and collated across participants.
Recurrent themes were examined for inter-subject
similarities and differences. For example, mentions of
new terms were compared across pedagogues with
regard to term used, sound quality achieved and
technique applied. This type of qualitative data
collection and analysis has been used extensively in
research of singing evaluation and assessment (33, 34).
A coding scheme for the transcripts was developed.
Three of the four authors reviewed the codes and
independently selected exemplars of each code. This
type of interpretative analysis is described by Smith
(30) as a way of capturing the meaning of a partici-
pant’s responses.
These qualitative data form the basis of the results
presented below. Participants’ responses to the ques-
tion topics are grouped thematically in Results, in
accordance with the stated aims of the semi-structured
interview topics. The quotes were examined by an
expert pedagogue, within the context of the transcript,
to ensure they represented instances of the stated
theme, or illustrated a significant statement unique to
an individual or sub group.
Defining open throat 169
Logoped Phoniatr Vocol 2836
RESULTS
Participants
The 15 singing pedagogues interviewed for the study
were aged between 39 and 63 years, with an average
age of 49. Ten primarily taught in Australia and fivetaught in the UK. They dedicated an average of 95%
of their professional life to singing or the teaching of
singing. They had between 4 and 28 years of
experience in teaching singing, with an average of 20
years.
For nine participants (P1, 2, 3, 6, 7, 9, 10, 11, 12), an
average of 67% of their students performed in the top
four categories (superstar, international, national/bigcity, regional/touring) of the Bunch and Chapman
taxonomy (28). Participants’ studios contained an
average of 20% full-time singing students, and for six
participants (P1, 4, 5, 8, 13, 14), ]/50% of their studio
comprised full-time students at a tertiary institution.
Influences
Table 1a and Table 1b present data in response to the
two questions ‘Do you follow any particular school of
thought or pedagogy or scientific approach?’ and
‘How does that pedagogy or writer influence your
view on open throat, throat widening, retraction or
space at the back?’ respectively. From the table, it can
be seen that the major influences were divided equallybetween Estill (7 participants), Richard Miller (7
participants) and Janice Chapman (5 participants).
Seven participants also referred to their own experi-
ence or study as influential on their pedagogy.
Participants who cited experience as their key
influence made comments of the following type: ‘I
got to the point in my development where I just had to
know what those things meant for me personally, so Ithink it was a lot of my own exploration that brought
about a deeper level of understanding’ (P15).
Factors in the achievement of good sound
In response to the questions ‘How do you achieve a
good sound in singing?’ and ‘How might you correct a
problematic sound in singing?’ 50% of participants
offered an average of four key concepts that they
considered essential to good sound (range 1�/6 terms).
Good breathing was the most important factor for ten
participants (67%) followed by balance or coordina-
tion for eight (53%). Table 2a presents all the factors
cited by participants and the frequency of citation,
and Table 2b presents references to open throat.
Samples of typical comments on good sound are:
‘A good sound really relies so much on freedom and
good coordination and connection with air so that the
right muscles are engaged with the right amount of
energy and there’s no interference’ (P8). ‘I would aim
to establish a freedom of production, so that is looking
at breath, de-constriction, a free flow, a free pathway
for the sound’ (P9). ‘I’m looking for a looseness
around the neck and for the articulators to be loose so
that there’s no musculature resistance to the emission
of sound. I’m looking for something that old peda-
gogy would call space making’ (P15).
Five teachers referred to the subjective nature of its
definition. Specifically, a good sound ‘is a sound that
is non-abusive and appropriate to genre’ (P7); ‘is
subjective to people’s tastes, so I work towards making
sure that the sound is safe, and then I work on the
aesthetics of the sound’ (P12); ‘needs good aural
awareness and good aesthetic training’ (P4).
Of the eleven participants, three referred to specific
technical instructions to achieve a good sound. They
emphasized specific components of their pedagogical
approach and named the phenomenon of open throat
technique: ‘Well the first thing I work on is posture,
and the second thing I work on is breathing and the
third thing I work on is resonance . . . and the most
important thing about resonance is the open throat’
(P1); ‘I teach all my clients to retract, that’s one of the
Table 1a. Pedagogues most regularly cited as influential in the development of teaching techniques for singing
Influence Chapman Estill Miller
Pedagogues 1, 4, 7, 9, 15 1, 2, 3, 4, 7, 9, 10, 11, 13, 15 2, 4, 8, 9, 10, 13, 15% 33 67 47
Table 1b. Pedagogical influences of singing teachers with respect to open throat techniques
Influence Estill Miller Self/eclectic Chapman Caesari Other
Pedagogues 1, 2, 3, 7, 9, 10, 11, 12, 13 1, 4, 6, 8, 10, 11, 14 1, 2, 4, 5, 6, 8, 10, 13, 15 2, 9, 10, 11, 14, 15 11, 14 4, 11% 60 47 60 40 13 13
170 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2837
first things I teach them’ (P12); ‘The larynx is sitting in
a slightly lower position . . . basically the larynx is
being pulled down by one laryngeal depressor, namely
the sternothyroid’ which ‘maximizes pharyngeal
depth’ (P10).An additional two participants mentioned primal
sound in their account of achieving good sound. This
was later defined as an action containing a component
of open throat technique. ‘I’ll start with breathing,
support, posture and the primal sound’ (P2). ‘Sound is
related to primal expression, people say it’s over the
top, I say opera starts over the top’ (P14).
Associations between the four terms
After the open-ended questions exploring general
concepts of pedagogy, participants were offered four
common terms used in the singing literature and in
singing pedagogy: open throat, throat widening,
retraction and space at the back, and were asked
‘What do the terms mean to you?’ and ‘How do you
feel these terms are related?’
Nine participants (P1, 3, 5, 6, 7, 8, 11, 12, 14) agreed
that the terms were either interrelated or interchange-
able, through universal use in singing studios. How-
ever, each participant went on to define more specific
differences between terms and chose one or more that
they advocated in teaching: ‘Well, they do [mean the
same thing]. If you’re talking about open throat,
everybody knows what you mean*/it’s a generally
accepted term’ (P5).
Three preferred the term open throat (P5, 6, 14) and
consistently used the term throughout the interview.
Five (P1, 3, 7, 11, 12) of the nine interpreted all
terms as attempts to describe the action of retraction:
‘All of those [terms] to me mean that the false folds are
retracted, that there is no interference of the false folds
in the airflow’ (P3). ‘I would say that open throat and
retraction are the same thing’ (P7).
For six participants (P2, 4, 9, 10, 13, 15), each term
was defined differently, and they described subtle
differences between these terms. Statements gave either
positive or negative associations. For example:
Positive statements: ‘Retraction I use because I find
it’s actually a very useful technique when people are
constricted, I can hear false vocal fold interaction very
clearly . . . open throat is an end product of getting
something right physiologically . . . it doesn’t have any
meaning outside of the actual physiological basis’
(P2). ‘I think retraction for me would mean false folds
opening up or open wider, so that there’s a sense of
free emission of sound . . . Space for me, also carries a
physical sensation of stillness, so that it’s not forced in
any way, there’s no laryngeal depression via the base of
the tongue, or no artificial palate lifting that involve
tension of any of the other articulators’ (P15).
Negative statement: ‘Open throat is the beginning of
a yawn, relaxed, slightly soft back of the tongue,
‘‘high-ish’’ palate, wide pharynx. Throat widening
might be slightly less relaxed and it might make me
think a little bit about retraction’ (P4).
For seven participants (P2, 4, 6, 7, 12, 13, 15), space
at the back was perceived to be a different concept to
open throat because of the required action of the soft
palate in enlarging the space. ‘Space at the back . . . to
me means wide pharynx and a high palate . . . open
throat is for me achieved by having a correct inhala-
tion process, which in my terms is a relaxed abdominal
or splat breath’ (P2). ‘Throat widening to me can be
lower than ‘at the back’ . . . [which] I perceive to be in
the soft palate region’ (P13).
Two participants (P7, 8) defined throat widening as
an external action, and related less to the internal
action of open throat. In their pedagogy, it was not
perceived as a useful term: ‘Throat widening [is not the
same as retraction]; I prefer open throat, because then
it, for me, indicates that something is being opened
Table 2a. General factors necessary to achieve a good
sound, ranked by pedagogue frequency of citation
Factors %
Breathing 67Balance/coordination 53Posture 40Support 20Energy 13Primal sound 13Resonance 13Tongue 13Airflow 7Aural awareness 7Effort 7Placement 7Positioning 7Ring/twang 7
Table 2b. Specific open throat references as a necessary
factor to achieve a good sound, ranked by pedagogue
frequency of citation
Open throat references %
Ease/freedom 40De-constriction/not constricted/tense 33Open/open throat 20Depth 13Retraction 13
Defining open throat 171
Logoped Phoniatr Vocol 2838
from the inside rather than the outside which is where
I teach people to make space’ (P7).
Two participants (P9, 10) expressed dissatisfactionwith all of the given terms because they were
potentially misleading: ‘Open throat, I think is a little
bit misleading for some people in that most singers
were raised on the old yawn sigh and the concepts of
yawning and making space’ (P9).
Preferred terms
Eight participants favoured one or more of the four
terms in their teaching. The other seven gave a
different term. Table 3 presents the use of preferredterminology.
New terms were given as (i) replacements for the
four given terms, or (ii) as better descriptors of similar
concepts; or (iii) as descriptors of new vocal pedagogy
believed to incorporate the same pedagogical goal.
Four participants (P8, 9, 11, 13) favoured the use of
the words free, freedom or freedom of tone.
‘I wouldn’t say I want an open throat . . . I wouldsay I want a free sound’ (P11). ‘A free sound, a
freedom of sound but the beginnings of what we call a
trained voice sound’ (P13).
‘Collar’ and ‘depth’ were discussed by four partici-
pants (P2, 9, 10, 15) and described as similar to, or
replacements for, open throat. For one, the concept
of collar is intrinsically related to the function of
retraction or open throat. ‘I teach [retraction] as afundamental of singing by getting the engagement of
the collar . . . the collar is both a postural and
a physiological function’ (P2).
For two participants (P10, 15), open throat was the
achievement of pharyngeal depth in classical singing.
It was also directly compared to the technique ‘collar’:
‘[The term collar], I’d call it low larynx or pharyngeal
depth’ (P10). ‘There is also a sense of laryngeallowering so that there feels like a great depth in the
vocal tract’ (P15).
Three of the participants (P2, 10, 15) would not use
a specific term in teaching but rather preferred to use
the students’ own language to describe their sensation
after teaching them the technique: ‘I would look for as
many synonyms as possible to describe the same thing,
and then I would probably ask a person to feed back
to me in their own language what we’ve just been
discussing so that we can actually have a point of
reference, so we’re both very clear as to what we’reboth talking about’ (P15).
Sound qualities associated with the terminology
Participants were asked ‘What sound quality do youassociate with these terms?’ They collectively produced
18 descriptions that they felt best characterized the
sound when open throat was used. Table 4 identifies
the most common sound terms used. ‘Balanced/
coordinated’ was the most important quality, named
by 14 of 15 teachers (93%). ‘Free’, ‘open’, ‘even/
consistent’ and ‘warm’ were each nominated by over
50% of pedagogues as the sound quality intrinsic tothe use of some form of open throat.
Fourteen participants (93%) said that using open
throat made the entire sound more balanced or
coordinated. They explained that using open throat
achieved: ‘a much better range of harmonics and
obviously from that, the formants’ (P3); ‘a resonating
spectrum [in which] frequencies balance quite beauti-
fully’ (P15).Eight pedagogues (53%) felt that the use of open
throat made the overall sound more even or consis-
tent: ‘If I told them to laugh, I was going to get a more
efficient result and a more effective sound, therefore a
rounder, bigger, more even, more secure, more stable,
more confident production of sound’ (P1).
Five pedagogues (P1, 2, 3, 7, 12) described healthy,
safe sound and then associated this with terms likeclear, clean and free: ‘The sound is clean . . . the sound
has no interference’ (P7). ‘The timbre changes when
retraction is achieved, and the voice sounds much
healthier, clearer, cleaner’ (P12).
Required technique to achieve the sound quality
In response to the questions ‘Suppose I were a new
student*/how would you tell me about some of these
terms?’ and ‘How would you teach me these techni-
ques?’ six pedagogues used some form of laugh and
five used cry or sob. A pre-yawn or start of a yawn was
mentioned by three, and focus by two. Three pedago-
Table 3. Terms favoured by each pedagogue as descriptors for open throat
Favoured term Open throat Retraction Open throat or retractioninterchangeably
New
Pedagogues 5, 6, 14 3, 7, 12 1, 2 4, 8, 9, 10, 11, 13, 15% 20 20 13 47
172 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2839
gues addressed the technique only through teaching
breathing. These are summarized in Table 5.
Nine pedagogues (P1, 2, 3, 5, 7, 8, 10, 11, 12)
believed that the action of open throat was
fundamental to the consistent production of good
sound. It was an important technique to maintain and
monitor throughout singing: ‘Otherwise you start to
use all the muscles that you shouldn’t deliberately be
using’ (P5). ‘It enables the mechanism to work
efficiently at source, so that you get a good sound
signal’ (P7). ‘It’s very much concerned with freedom
and the idea of just opening up and letting the sound
flow’ (P8). ‘Basically after [freedom], you’re adjusting
resonances’ (P11).Five pedagogues (P2, 7, 8, 9, 13) believed that open
throat needed to become automatic. They stated that
this was possible once a student had mastered
the action: ‘Yes, it’s there all the time when you’ve
got it*/it becomes part and parcel of your normal way
of singing. But it’s part of the breathing and support
function’ (P2). ‘There has to be a portion of the
concentration monitoring, until it becomes natural . . .
every time they open their mouths, they know it’s
going to happen’ (P13). ‘It is one of those things that a
student can become able to do automatically and
therefore it goes into the muscular memory and
it’s something that once you’ve experienced, you might
only monitor when you notice that it’s not happening’
(P7).
One pedagogue (P11) highlighted the problem of
actually learning to achieve the action of freedom
consistently: ‘It seems to me that you actually have to
have exercises that maintain that sensation, you can’t
just identify it and say that’s what’s happening’ (P11).
For two pedagogues (P6, 8), open throat was valued
as a tool that demanded more concentration in certain
ranges of the voice and particularly singing through
the passaggio: ‘I would say, particularly as one is going
into the passaggio . . . it helps people to have that
feeling of space at the back of the throat, and almost a
sensation of the voice going backwards’ (P6).
In the singing studio, three participants (P1, 8, 12)
said that it was a technique they regarded as important
to consider every time they taught: ‘[Retraction] is
something I teach every day. And with every student,
no matter how advanced’ (P1). ‘I use it all the time. I
think it’s like second nature . . . if I weren’t retracting,
I’d be constricting’ (P12).
Two participants (P4, 9) regarded open throat and
the teaching of it as a prescriptive tool or as a
technique to visit only when constriction was causing
a problem. They advocated addressing constriction
in other ways rather than discussing open throat:
‘[Open throat] isn’t an integral concept I consciously
and consistently consider . . . so on a needs basis.
Table 4. Sound qualities associated with open throat. Pedagogues’ individual choices indicated by I
Pedagogue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 %
Balanced/coordinated I I I I I I I I I I I I I I 93Free I I I I I I I I I I 67Open I I I I I I I I 53Even/consistent I I I I I I I I 53Warm I I I I I I I I 53With space I I I I I I I I 40Healthy/safe I I I I I 33Round I I I I I 33Overtones/formants I I I I I 33Easy/flexible I I I I I 33Clear I I I I 27Full I I I 20Efficient I I I 20With depth I I I I 20Clean I I I 20Sexy/juicy/lusty I I 13Natural voice I I 13Relaxed I 7
Table 5. Pedagogical views on techniques to achieve open throat
Technique Laugh Cry or sob Yawn (start/pre) Focus Flow of breath Inhalation
Pedagogues 1, 3, 7, 9, 12, 15 1, 2, 9, 10, 15 2, 4, 9 5, 6 13, 14 8% 40 33 20 13 13 7
Defining open throat 173
Logoped Phoniatr Vocol 2840
I don’t necessarily talk about open throat that much
. . . it’s not fundamental’ (P4). ‘[I use open throat] as
and when it matters. Some people have very little
constriction . . . often, if their breath is in place and if
that is buoyant enough, that takes care of a lot of
constriction anyway’ (P9).
Relationships between terms, actions and sound
qualities
Thirteen participants (80%) made a connection be-
tween breath or breathing-in and the concept of an
open throat. For eight participants, it was linked with
the inhalation process (P1, 2, 3, 8, 9, 10, 12, 15): ‘Open
throat . . . is for me achieved by having a correct
inhalation process . . . it’s a recoil diaphragmatic
breath’ (P2). ‘We want to maximize resonance and
freedom, maintaining that posture of inhalation’ (P8).
‘It’s a physical feeling that you get on inhalation and
that you maintain until the end of the phrase’ (P10).
‘I think space is initiated with inhalation’ (P15).
Similarly, seven participants (P1, 2, 7, 8, 9, 10, 15)
indicated that they would actively use the technique to
prepare for producing a note and linked it to the
inhalation process: ‘I would use it in association with
breath and taking breath because those two activities
are combined’ (P9). ‘You set it up with the intake of
breath.’ (P12).For four participants (P5, 7, 13, 14), it was an action
connected with the flow of breath and the continua-
tion of the breath throughout the phrase: ‘So there is
this lovely column of sound between the focus point
and the support in the voice’ (P5). ‘I talk more about
energizing the breath, energizing the voice, keeping the
tone consistent, working to the end of the phrase’
(P13). ‘It is to do with the flow of breath. If I can
perceive that the flow of breath is neither squeezed nor
dissipated . . . then generally the throat is open’ (P14).
Instructions to achieve an open throat included
actions to maximize space. Two participants (P2, 9)
used a pre-yawn or pre-yawn sigh to describe this to
students: ‘Sometimes I add instructions to maximize
the vocal tract by using, for example, the roar of a lion,
to feel how big those spaces can be and the yawn-sigh
effect, the pre-yawn sigh, I call it of allowing the sound
to come out without constricting the walls of the
pharynx’ (P2).
The technique of laugh was consistently linked to
the term retraction: ‘I use the device of laughter
because I know that that brings about the physical
action [of retraction]’ (P3). ‘[I would teach retraction]
via silent laughing, but I also use achieving a silent in
and out breath, so basically silent inspiration and
expiration . . . it’s a position and it’s a muscular
activity. What happens with the breath in terms of
airflow for instance happens as a result of it’ (P7).
For three participants (P1, 9, 15), laugh and cry orsob produced a subtle change to the sound quality:
‘a different colour’ (P1), ‘richer’ (P9) and a ‘different
harmonic readout within the tone’ (P15). ‘The other
way to open the throat is to cry in tune... on the
laughing basis, but I want to change the colour in the
sound. So I’m doing that for an aesthetic colour’ (P1).
Seven participants (P1, 2, 7, 8, 9, 10, 15) sponta-
neously commented on the value of open throat inclassical singing. It was considered (i) as a necessary
technique for making a classical sound and (ii) a
quality in the sound itself.
Classical technique: ‘Requires a degree of work
taking place in the larynx, and so therefore you need
to have the retracted position, the open throat, in
order to make it safe’ (P7). ‘The aim for classical is to
have the greatest sense of freedom involved, andfreedom with the laryngeal tract, or the vocal tract,
happens to occur with space-making.’ (P15)
Classical sound quality: ‘I think the Italian articu-
latory system sets up an open throat, and western
classical singing came from that . . . [open throat] is the
difference between a national sound and an interna-
tional sound’ (P2). ‘I think that [open throat] not only
improves but in terms of classical singing, it fines thequality that one requires if you want to be a classical
singer’ (P8). ‘It probably contributes acoustically to
the sound that is most satisfying in that particular
musical style’ (P9). ‘With depth, I will hear basically a
warmer, more classical sound’ (P10).
Physical sensations and presumed physiology associated
with the terminology
The physical action of open throat was conceptualizedby participants in two ways: in terms of anatomy or
physiology, and/or as a physical sensation.
Table 6 presents the anatomical and physiological
concepts mentioned in response to the question: ‘In
terms of physiology, which muscles would you expect
to be working, that is to say, what happens when you
use open throat?’
Eight participants consistently mentioned a pre-sumed ‘abduction’ or retraction of the false vocal folds
or ventricular folds which they linked to the term
retraction. Participants who referred to false vocal fold
movement reported seeing the action on endoscopy.
‘The false folds are retracted and there is no inter-
ference of the false folds in the airflow or indeed that
the false folds are not in any way pressing on the true
vocal folds’ (P3). ‘What I have noticed, when I look atpeople under endoscopy and stroboscopy, is that when
the ventricular bands are retracted the true vocal folds
174 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2841
have a much more symmetrical mucosal wave, and so
they behave much more efficiently and you hear that
efficiency in the purity of the tone’ (P12).
Seven participants stated that pharyngeal space was
involved in the technique they used: ‘[In retraction],
there is some lowering [of the larynx] and I think
there’s some widening of the larger spaces in the
laryngeal part of the pharynx’ (P2). ‘Open throat to
me is optimizing; perhaps all your pharyngeal space,
and that might involve a slight lowering of the larynx’
(P4). ‘I think it has more to do with pharyngeal depth
than it is to do with widening. Because as the larynx
descends, it comes slightly forward in the neck and if
you have a high, wide tongue position that’s slightly
further forward and you actually gain more pharyn-
geal depth as well as pharyngeal length’ (P10).
Six participants mentioned larynx lowering or the
use of sternothyroid as having an effect on open throat
or pharyngeal space: ‘I think that there has to be an
absence of the glossus muscles, that the muscles that
push down on the larynx from above are a hindrance
in getting an open throat. Sternothyroid, the true
depressor of the larynx, which actually functions
physiologically from below is much more valuable’
(P2). ‘I expect that there is a little bit of thyroid tilt, so
that the muscles involved in the thyroid tilting
mechanism are going to be involved [in retraction]’
(P3). ‘I would say the oropharynx or the middle
constrictor region, certainly tongue muscles, the
thyrohyoid, maybe sternothyroid as well would all
work as a unit to allow laryngeal lowering. Certainly
you are looking for a slackness in the muscles above
the larynx’ (P15).
Five participants (P2, 3, 7, 11, 12) commented that
there was not adequate research, and they could not
be completely sure what muscles were used or affected
by the action of open throat: ‘I can’t tell you that
really. I think that the research has just not been done’
(P2). ‘This is one of the things that needs more
research, we don’t absolutely know what happens
when you retract’ (P12).
Table 7 presents data on sensations associated with
the use of this technique. Participants described the
sensation they had or tried to achieve when they used
open throat. They named three locations of sensation
they associated with this technique: (i) internally
at laryngeal level; (ii) externally at laryngeal level;
and (iii) in the oropharynx and laryngopharynx. Table
8 presents the participants’ responses to the questions
of associated physical sensations when using open
throat.
Three participants described an internal sensation
at laryngeal level and specifically linked this to the
action of retraction: ‘[You feel] a widening . . . right at
the laryngeal area itself and also it feels much wider
behind the epiglottis and down in the pharynx itself’
(P1). ‘You feel it right inside the vocal mechanism. It
may just be that I’ve got a good imagination, but I feel
the giggle or the sob has got a very subtle, very internal
manoeuvre’ (P2).
For four participants, the action was felt externally
at laryngeal level. It was also visible in the movement
of the thyroid cartilage: ‘You can actually feel it
around the sides of the thyroid cartilage, because if
you’re widening on the inside there, then you do
actually get a widening [on the outside]. So you can
feel the effects of the retraction*/you can’t feel the
false vocal folds from outside’ (P7).
One participant (P10) was more explicit about this
external action, and linked it to vertical movement of
the larynx: ‘I feel it in the base of my neck . . . which is
actually sternothyroid having to kick in to pull the
larynx low enough so that you can get a thick enough
edge of vocal fold together for a slower vibratory rate’
(P10).
Six participants explained that the action of open
throat was felt throughout the oropharynx and
laryngopharynx: ‘One feels a kind of opening and a
Table 7. Pedagogues’ report of perceived sensation when they used open throat
Sensation Internally at laryngeal level Externally at laryngeal level In oropharynx and laryngopharynx
Pedagogues 1, 2, 3 3, 7, 10, 12 4, 6, 8, 9, 11, 15% 20 27 40
Table 6. Pedagogues’ perceptions on anatomical and structures involved when using open throat
Structure False vocal folds Pharynx Sterno-thyroid Larynx lowering Other structure
Pedagogues 1, 2, 3, 7, 11, 12, 13, 15 1, 2, 4, 8, 9, 10, 15 2, 10, 12, 15 2, 3, 4, 10, 15 7% 53 47 27 33 7
Defining open throat 175
Logoped Phoniatr Vocol 2842
rising feeling at the back of the throat’ (P6). ‘[The
sensation] is all about pharyngeal muscles’ (P8).
‘Mainly, the sensation is within the throat region’
(P15).
Table 8 presents responses to the questions ‘Do you
think it is possible to have voluntary control of
muscles responsible for open throat, throat widening,
retraction and space at the back?’ and ‘Do you believe
it is possible to isolate and control the muscles used in
open throat, throat widening, retraction and space at
the back?’Thirteen participants (87%) considered that it was
possible to have voluntary control of the action of
open throat: ‘Conscious control of constriction or
retraction, nothing’s new really... my experience is that
of course it is’ (P11). ‘I certainly think you have
voluntary control of open throat, in that you can
optimize space’ (P4). ‘I think there is, up to a point,
voluntary control’ (P6). ‘It can take a long time, but
yes, you can.’ (P10). ‘I think to a certain extent... I
don’t know how efficiently [the muscles] work to
sustain that. I think you can generate a response,
particularly on the point of inhalation, but I’m sure
that there are postural considerations such as head
and neck alignment that help to maintain that
positioning’ (P9).
Two participants considered open throat as part of a
greater whole of singing technique, rather than a single
action: ‘The muscles controlling the larynx work of
their own accord’ (P5). ‘If a singer thought that they
had to have voluntary control every time they went out
on stage, they’d be in a mess. No they don’t have
voluntary control, but you do it, you get to the point
where... once you put the message into the subcon-
scious, it becomes habitual and then we just do it’ (P8).Seven participants believed that is was possible to
isolate and control the action. These comments were
specific to retraction: ‘I’m primarily looking for the
isolation simply of getting the false folds retracted so
that we have a de-constricted space at vocal fold level’
(P3). ‘Yes, I believe so, it has certainly been my
experience’ (P7). ‘I feel as though you can [isolate]. I
don’t know that I can actually control that, it may be
something else that’s moving’ (P9). ‘[Isolation] defi-
nitely, yes. However, I don’t know if anyone could
feedback accurately exactly what muscles they were
working with at any given point in time’ (P15).
Five participants believed that isolation of open
throat was not possible and that the action was a
component of a larger action or gesture in singing: ‘[Awhole motion] would certainly be my approach, rather
than trying to isolate elements’ (P6). ‘I don’t think you
can isolate one little thing. It’s part of a bigger picture,
and I think sometimes, when you try to isolate bits, it
can become too compartmentalized and a student
actually ends up thinking about bits and forgets about
singing’ (P13). ‘I don’t think that one can take one
muscle and say, exercise it, for example. They all workin a greater union’ (P15).
DISCUSSION
The majority of participants interviewed for this study
agreed that open throat was essential to good singing
and more specifically to classical singing. Specific
sound qualities in good singing were identified and
some were linked directly to open throat technique.
Participants acknowledged the complex nature ofinterdependent techniques needed in good singing.
Good breathing, balance or coordination, and free-
dom or ease of sound were considered essential to the
production of a good sound in singing. These concepts
have often been linked in the literature; for example,
Miller (24) believed that freedom in sustained singing
is produced by balanced breathing.
With respect to terminology regarding open throat,open throat and retraction were preferred, from the
given terms, to describe the phenomenon. Other terms
such as freedom, collar and depth were suggested as
better terms. Individual participants preferred one of
these terms to describe their concept of open throat,
although some used terms interchangeably. Reasons
for preferences given include: clarity of meaning;
preference for a specific pedagogy’s terminology; orreference to the type of action used to achieve the
technique.
Different pedagogues preferred different terminolo-
gies: the descriptors used reflected preferences to
current and historical singing pedagogy or pedago-
gues. For example, Miller (24) referred to open throat,
Estill referred to retraction (16) and Chapman to
collar, and the use of these terms were reflected in thecomments of this sample. However, the sound quali-
ties associated with all these terms were very similar.
Table 8. Pedagogues’ view on degree to which open throat was voluntary, involuntary, could be isolated or was part of
a coordinated action
Control Voluntary Involuntary Isolated action Coordinated action
Pedagogues 1, 2, 3, 4, 6, 7, 9, 10, 11, 12, 13, 14, 15 5, 8 2, 3, 7, 8, 9, 11, 12 4, 5, 6, 13, 15% 87 13 47 33
176 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2843
Singing teachers associated open throat with both a
sound quality that was characterized by freedom,
warmth and openness and an action that produced
balance, coordination, evenness and consistency. The
terms related to vocal quality such as ‘warm’, ‘full’ or
‘round’, as well as the functional terms such as ‘easy’
or ‘clean’, were used interchangeably by participants
in this study. The use of function and quality
descriptors was addressed in Wapnick and Ekholm
(26), who found that there were differing preferences
for use of quality or function terms. Although experts
tended towards one category of descriptor, the high
correlations of all descriptors suggest that sound
quality and function are a unitary concept in evaluat-
ing good sound. The participants in this study did not
separate these types of terms and seemed comfortable
with their use in the studio.
Participants in this sample demonstrated an under-
standing and interest in anatomical and physiological
relationships between sound quality and technique,
citing presumed muscle action such as ventricular fold
abduction or pharyngeal widening and associated
physical sensations. Most considered it feasible to
have voluntary control over the action of open throat,
but did not achieve consensus over the ability to
isolate the action from good singing technique. The
techniques used by participants in this sample in-
volved laughing, crying, conscious maintenance of
inhalation postures and (pre-) yawn. Scientific inves-
tigations related to each of these vocal actions provide
interesting hypotheses for what may happen physiolo-
gically when participants teach open throat technique
(25, 35�/9).
Laughing is advocated by pedagogues such as Estill
(16) and Miller (24) as a fundamental tool in each
pedagogical style for (i) the achievement of false vocal
fold retraction and (ii) for the balance of onset and
release respectively. The majority of those who advo-
cated false vocal fold retraction (ventricular fold
abduction) said that isolation of this action was
possible. This view is consistent with Estill (16, 25)
who recommends independent control of components
of the vocal mechanism. It is important to note that
no studies have been published providing evidence of
voluntary control of movement of the ventricular folds
laterally away from the midline. Current thinking is
divided as to whether retraction (25) occurs in open
throat and if there is a response to an instruction to
‘retract your ventricular folds’ or ‘retract’ from a rest
position.
Participants who used laugh as a technique did so to
achieve retraction. Although the use of this term has
been absorbed into singing pedagogy, it lacks anato-
mical validity. The anatomical term ‘abduction’ refers
to a lateral movement away from the body midline
with the opposite term, ‘adduction’ referring to a
lateral movement towards the body midline. The term
abduction appears to be the movement that some
pedagogues referred to as retraction which is not
usually used anatomically in that sense. However, the
use of the term retraction by the participants inter-
viewed here clearly meant more than a relaxation to a
rest position from a constricted vocal pattern; it also
implied a lateral movement away from the approxima-
tion to the midline. What is not known is whether
there is any voluntary control of the lateral position of
the ventricular folds as advocated by Estill. Those
participants who used the term and technique of
retraction did so to reduce constriction or tension to
achieve a healthier sound quality as well as to achieve
a specific sound quality.
Physiologically, abduction of the ventricular folds
presents a puzzle as they are ‘incapable of becoming
tense’ (40, p. 117). Histological reports have described
a rostral extension of the thyroarytenoid muscle up to
13 mm above the glottis, which at the level of the
ventricle, 2�/3 mm above the glottis, lies lateral to it,
and at this level it is referred to as the ventricular
muscle (41). The thyroarytenoid muscle is an adductor
of the vocal folds, and will have the same action on the
ventricular folds and so cannot be responsible for
abduction of the ventricular folds. There is voluntary
control over constriction of the ventricular and vocal
folds with adduction of the ventricular folds, and
changes in the vocal vibratory waveform are observed
in strained/tense and also in pressed phonation.
Reduced tension of the larynx and a degree of
expansion in the laryngeal ventricle was reported in
‘covered singing’, a technique used when singing
through the passaggio (42).
Changes in the lateral position of the ventricular
folds may alter the shape and position of the ventricle,
which could be responsible for other resonator char-
acteristics or even the vibratory pattern of the vocal
folds. In studies of specific vocal register by Welch
et al .(38), they found that male falsettists demon-
strated a systematic increase in pharyngeal and
ventricular area as well as shape and position of the
laryngeal ventricle across a series of octaves. They
suggested that darkening or covering was achieved
while still maintaining ‘singer’s formant’, by opening
the ventricular space and lowering the larynx. Refer-
ence to the ventricle and its perceived role in singing
were made by the Bel Canto school, and interpreted in
the twentieth century by practitioners such as Manen
(19) to include direct reference to the ventricular folds,
and she concluded that they are altered in length and
width, which directly affected the ventricle of Mor-
gagni and produced the ‘specific timbres of Bel Canto’
(19, p. 34).
Defining open throat 177
Logoped Phoniatr Vocol 2844
However, it is not yet established whether the
opening of the ventricular space and lowering the
larynx is influenced by instruction and adjustments
made for the technique of open throat. Sundberg (39)
proposed that the observed distinctive high-energy
spectrum in singer’s formant is present when a specific
size relationship occurs between the diameter of the
pharynx and the opening of a narrowed laryngeal
inlet or tube (size ratio pharynx to larynx opening
areas �/6:1), which is more likely to occur when the
larynx is lowered. Participants in this study did not
specifically link the use of an open throat with ‘ring’,
terms which have previously been linked (43),
although they did suggest it facilitated a better balance
or more desirable distribution of harmonics.
Some participants in this sample advocated cry as
an instruction and discussed larynx lowering as an
action in open throat. Certain participants speculated
that the contraction of the sternothyroid muscle might
influence space created in the pharynx, larynx low-
ering or other associated movements in the vocal
mechanism. There is some support for this hypothesis.
Sternothyroid contraction is believed to influence the
thickening of the vocal folds in low pitches, but it is
uncertain as to its effect on higher pitches in trained
singers (44). The sternothyroid muscle attaches to
the thyroid cartilage, and it is possible that on short-
ening, it makes some impact on the vocal fold
configuration. Roubeau et al . (44) found that the
activity of the strap muscles, including the sternothyr-
oid, varied according to fundamental frequency, with
most activity occurring at the extremes of range. Hong
et al . (10, 45) concurred that the strap muscles did
affect fundamental frequency through the mechanism
of laryngotracheal pulling or bending. These studies
support the views of the participants’ interviewed that
contraction of the sternothyroid muscle has some
effect on the laryngeal configuration.
Vilkman et al . (46) agreed that changes to the
vertical position of the larynx had effects not only on
fundamental frequency but also may contribute to
abduction and adduction functions within the larynx.
This supports the benefits of lowered larynx as used in
classical singing. Shipp (47) determined that thyro-
hyoid and sternothyroid muscles were responsible for
subjects’ vertical laryngeal positioning while singing,
which matches the participants’ beliefs of laryngeal
control and manipulation.
Pedagogical methods linking breath, or the intake
of breath, was deemed by pedagogues in this sample to
be vital in the production of open throat, and achieved
using a variety of images or gestures in order to locate
a sound quality and sense of space or freedom (5).
Sonninen’s (48) pioneering study into external laryn-
geal frame musculature supported this pedagogical
theory, and identified anterior and inferior displace-
ment of the thyroid cartilage by sternothyroid and
tracheal exertion via inhalation, which may account
for vibratory changes in the vocal folds and acoustic
changes in fundamental frequency.
Many participants in this sample talked about
widening of the pharynx or a sensation of space
within the pharynx. They also associated this with
breath, and an action initiated through inhalation.
Anatomically, there is evidence that the pharynx can
widen or lengthen during singing. The pharynx is
comprised of three constrictors so, anatomically, only
lengthening of the vocal tract, or protrusion of the
tongue, should enlarge the pharyngeal resonator. The
yawn/yawn-sigh has been used to encourage open
throat, but has been questioned as a tool for achieving
it, as it may add unnecessary tensions (24). Pre-yawn
or yawn-sigh was named in this study as other
pedagogical devices to achieve open throat. Boone
and McFarlane (49) studied the yawn-sigh under naso-
endoscopy and identified characteristic lowered larynx
and widened pharynx across subjects, as well as
retracted elevation of the tongue which supports the
participants’ application of yawn or pre-yawn to
enlarge the pharyngeal resonator. Acoustically, this
yawn-sigh produced lower second and third formants.
The notion of pharyngeal size and its impact on the
increased intensity in sound was reported by the
participants interviewed in this study. Sonninen et al .
(50) investigated the size of the pharynx during open
and covered singing, in loud and quiet examples, and
proposed that the lower pharynx was in fact narrower
during the loud covered condition than in the open
condition. However, it is qualified by its restriction to
the low level of pharynx area studied. Although
pedagogues here referred to wide pharynx, they also
quantified depth via laryngeal lowering as important
to open throat. The narrowing of epiglottal area, and
its role in the production of singer’s formant is
investigated by Titze (9), who proposed the benefits
of adopting a wide pharynx, to achieve a lower first
formant to enhance the sound, and increase the
singer’s formant with a narrowed laryngeal inlet or
tube (39).
Recent studies (10, 36) investigated various voice
qualities including the ‘yawny’ voice quality and
concluded that this vocal tract manipulation resulted
in a distinctive increase in vocal tract volume, through
lengthening the vocal tract and widening the oral
cavity, which resulted acoustically in a closer grouping
of the first two formants. These findings confirm
pedagogical statements in this study related to the
perceived impact of the use of the sob/cry/yawn
manipulations to achieve greater space and lower the
larynx, but do not investigate the concept of tension.
178 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2845
Studies of control of different sound qualities,
focusing on different vocal genres have identified
different vocal ‘postures’ in singing (35, 37). Physio-logical movements were different across singing styles
and were believed to account for changes to the sound
quality. When the vocal tract was compared across a
single singer, operatic technique versus belting techni-
que elicited a lower larynx, wider larynx tube and
more separated ventricular folds. Interestingly, the
instruction to change the sound quality produced the
physiological variations.
CONCLUSIONS AND RECOMMENDATIONS
This study assessed the use of language about sound
quality associated with the technique of open throat.
Although the 15 participants offered 18 terms todescribe sound quality, clear associations or simila-
rities in their usage and application were apparent.
This study advances previous work in clarifying
terminology related to open throat (51, 52). Commu-
nication of techniques in singing pedagogy can be
improved by attempts to inform the use of terminol-
ogy. Consensus of pedagogues is displayed in the
evaluations of singing performance (33), although theway in which pedagogues interpret a sound is often
difficult to define verbally.
It also expands Wapnick and Ekholms’ (26)
attempts at matching pedagogical evaluation with
perceptual descriptors. Singing pedagogues rely on
their perceptions of sound qualities to determine the
physiological processes at work in the production of
the sound quality. The human ear must complementacoustical study in integrating the complex dimensions
of the human voice. Language in singing pedagogy,
despite Seashore’s (2) ideals, is destined to remain
subjective. However, in order to spread information
more effectively, we would make the following recom-
mendations:
1. Singing pedagogy and voice research need to
become more integrated. De-constriction is a
more anatomically correct term than retraction.
This would help to convey meaning more effec-
tively between science and art.2. Qualitative research is a useful tool for the
generation of research questions and elucidation
of study related to the voice, in particular, to the
sound quality.
3. Subsequent research needs to establish the asso-
ciated characteristics of open throat and the
perceptual quality accompanying this technique
(26, 27).4. In this study, open throat was considered impor-
tant in current pedagogy. Pedagogues were aware
of the technique’s value as well as the need to
tailor their terminology and instructions in the
singing studio to each student’s vocal needs andlearning styles.
REFERENCES
1. Jorgenson D. A History of Conflict. NATS Bulletin1980; 36: 31�/5.
2. Seashore CE. Psychology of Music. New York andLondon: McGraw-Hill Book Company; 1938.
3. Bartholomew W. A physical definition of ‘‘good voicequality’’ in the male voice. J Acoust Soc Am 1934; 6:25�/33.
4. Tosi PF. Opinioni de’cantori antichi, e moderni, o sienoOsservazioni sopra il canto figurato 2nd edn. London:JWilcox: 1743
5. Puritz E. The teaching of Elisabeth Schumann. London:Methuen; 1956.
6. Herbert-Caesari EF. The voice of the mind. London:Robert Hale; 1951.
7. Agricola JF. In: Julianne C Baird (ed.). Introduction tothe art of singing; translated. Cambridge: CambridgeUniversity Press; 1995.
8. Vennard W. Singing: the mechanism and the technic, 5thedn. New York: Fischer; 1968.
9. Titze I. Voice Research: The Wide Pharynx. Journal ofSinging 1998; 55: 27�/8.
10. Titze IR, Story BH. Acoustic interactions of the voicesource with the lower vocal tract. J Acoust Soc Am1997; 101: 2234�/43.
11. Joiner JR. Charles Amable Battaille: pioneer in vocalscience and the teaching of singing. Lanham (Md):Scarecrow Press; 1998.
12. Ware C. Basics of vocal pedagogy: the foundations andprocess of singing. New York: McGraw-Hill; 1998.
13. Marafioti PM. Caruso’s Method of Voice Production.New York: Dover Publications; 1981.
14. McKinney JC. The Diagnosis and Correction of VocalFaults. Nashville, Tennessee: Broadman Press; 1982.
15. Reid CL. A dictionary of vocal terminology: an analysis.New York: Joseph Patelson Music House; 1983.
16. Estill J. Primer of Basic Figures, 2nd edn. Santa Rosa:Estill Voice Training Systems; 1996.
17. Miller R. On the art of singing. New York: OxfordUniversity Press; 1996.
18. Hemsley T. Singing and imagination: a human approachto a great musical tradition. Oxford and New York:Oxford University Press; 1998.
19. Manen L. Bel canto: the teaching of the classical Italiansong-schools its decline and restoration. Oxford: OxfordUniversity Press; 1987.
20. Potter J. Vocal authority: singing style and ideology.New York: Cambridge University Press; 1998.
21. Garcia M. Hints on singing, New and revised edition.London: Ascherberg, Hopwood and Crew; 1894.
22. Shakespeare W. The art of singing: based on theprinciples of the old Italian singing masters, and dealingwith breath-control and production of the voice,together with exercises, Revised edition. Boston: OliverDitson; 1921.
23. Lehmann L. How to sing, New revised and supplemen-ted edition translated by Clara Willenbucher. New York:Dover Publications; 1993.
Defining open throat 179
Logoped Phoniatr Vocol 2846
24. Miller R. The structure of singing: system and art invocal technique. New York and London: SchirmerBooks; 1996.
25. Citardi MJ, Yanagisawa E, Estill J. Videoendoscopicanalysis of laryngeal function during laughter. Ann OtolRhinol Laryngol 1996; 105: 545�/9.
26. Wapnick J, Ekholm E. Expert consensus in solo voiceperformance evaluation. J Voice 1997; 11: 429�/36.
27. Ekholm E, Papagiannis GC, Chagnon FP. Relatingobjective measurements to expert evaluation of voicequality in Western classical singing: critical perceptualparameters. J Voice 1998; 12: 182�/96.
28. Bunch M, Chapman J. Taxonomy of singers used assubjects in scientific research. J Voice 2000; 14: 363�/9.
29. Patton MQ. Qualitative evaluation and research meth-ods, 2nd edn. Newbury Park, California: Sage Publica-tions; 1990.
30. Smith JA. Semi-Structured Interviewing and QualitativeAnalysis. In: Langenhove LV (ed.). Rethinking Methodsin Psychology. London:SagePublications;1995. pp.9�/26.
31. Mathieson CM, Barrie CM. Probing the Prime Narra-tive: Illness, Interviewing, and Identity. Qual Health Res1998; 8: 581�/601.
32. Miles MB, Huberman MA. Qualitative Data Analysis:An Expanded Sourcebook. California: Sage Publica-tions Ltd; 1994.
33. Davidson JW, Coimbra D. Investigating performanceevaluation by assessors of singers in a music collegesetting. Musicae Scientiae 2001; 5: 33�/53.
34. Kokotsaki D, Davidson JW, Coimbra D. Investigatingthe Assessment of Singers in a Music College Setting:The Students’ Perspective. Research Studies in MusicEducation 2001; 16: 15�/32.
35. Lovetri J, Lesh S, Woo P. Preliminary Study on the Abilityof Trained Singers to Control the Intrinsic and ExtrinsicLaryngeal Musculature. J Voice 1999; 13: 219�/26.
36. Story BH, Titze IR, Hoffman EA. The relationship ofvocal tract shape to three voice qualities. J Acoust SocAm 2001; 109: 1651�/67.
37. Sundberg J, Gramming P, Lovetri J. Comparisons ofPharynx, source, formant, and Pressure Characteristicsin Operatic and Musical Theatre Singing. J Voice 1993;7: 301�/10.
38. Welch GF, Sergeant DC, MacCurtain F. Some PhysicalCharacteristics of the Male Falsetto Voice. J Voice 1989;2: 151�/63.
39. Sundberg J. Articulatory interpretation of the ‘‘singingformant’’. J Acoust Soc Am 1974; 55: 838�/44.
40. Zemlin WR. Speech and Hearing Science: Anatomy andPhysiology, 4th edn. Boston: Allyn and Bacon; 1998.
41. Hirano MS. K. Histological Color Atlas of the HumanLarynx. San Diego: Singular Publishing Group; 1993.
42. Hertegard S, Gauffin J, Sundberg J. Open and CoveredSinging as Studied by Means of Fiberoptics, InverseFiltering, and Spectral Analysis. J Voice 1990; 4: 220�/
30.43. Acker BF. Vocal Tract Adjustments for the Projected
Voice. J Voice 1987; 1: 77�/82.44. Roubeau B, Chevrie-Muller C, Lacau Saint Guily J.
Electromyographic activity of strap and cricothyroidmuscles in pitch change. Acta Otolaryngol 1997; 117:459�/64.
45. Hong KH, Ye M, Kim YM, Kevorkian KF, Berke GS.The role of strap muscles in phonation*/in vivo caninelaryngeal model. J Voice 1997; 11: 23�/32.
46. Vilkman E, Sonninen A, Hurme P, Korkko P. Externallaryngeal frame function in voice production revisited: areview. J Voice 1996; 10: 78�/92.
47. Shipp T. Vertical laryngeal position during continuousand discrete vocal frequency change. J Speech Hear Res1975; 18: 707�/18.
48. Sonninen A. The external frame function in the controlof pitch in the human voice. Ann NY Acad Sci 1968;155: 68�/90.
49. Boone DR, McFarlane SC. A critical view of the yawn-sigh as a voice therapy technique. J Voice 1993; 7: 75�/80.
50. Sonninen A, Hurme P, Laukkanen AM. The externalframe function in the control of pitch, register, andsinging mode: radiographic observations of a femalesinger. J Voice 1999; 13: 319�/40.
51. Burgin JC. Teaching singing. Metuchen (NJ): ScarecrowPress; 1973.
52. Monahan BJ. The art of singing: a compendium ofthoughts on singing published between 1777 and 1927.Metuchen (NJ): Scarecrow Press; 1978.
180 H. F. Mitchell et al.
Logoped Phoniatr Vocol 2847
2.1 Conclusion
The similarities in pedagogues’ description of laryngeal gestures to achieve the
technique or sound quality and their consensus regarding its value, particularly to
classical singing, generated the hypotheses for subsequent acoustic studies. Pedagogues’
subjective descriptions of sound qualities indicated the difficulty in communication of
vocal pedagogy.
To date, research into the singing voice has described acoustic properties of voice and
its visual representation with few or no links to pedagogy. However, there are two
empirical studies which link acoustic measures with perceptual judgements and have
provided a basis for the projects in this thesis. One, by Wapnick and Ekholm (1997)
established 12 generally accepted perceptual criteria applied to the assessment of voice
quality in classical singing, and the other, by Ekholm, Papagiannis and Chagnon (1998)
applied four of these criteria (“appropriate vibrato”, “resonance/ring”, “color/warmth”,
and “clarity/focus”) and related them to objective measurements taken from acoustic
analysis of the voice signal. This thesis extends this work by investigating an individual
singing technique.
Hypotheses on the effects of open throat were generated from pedagogical beliefs and
practices derived in the first study (Mitchell, Kenny, Ryan, & Davis, 2003). These were
tested acoustically and perceptually in the second study and compared with factors of
optimal singing quality which have been identified in the literature and defined
acoustically. As open throat was considered an important component of good singing,
the following studies compare singing with and without application of open throat
48
technique to factors of good singing quality identified in previous voice science
literature. One such factor is vibrato.
49
3 Vibrato
There is now both a vast literature on vocal acoustic properties and the visual
representation of voice with many unanswered questions about what conclusions can be
drawn from such analyses. Current singing pedagogy texts absorb these findings into its
descriptions of voice (Callaghan, 2000; Callaghan & Wilson, 2004; Miller, 1996; Nair,
1999; Sundberg, 1988; Thurman & Welch, 2000), but these in turn lead to many
unanswered questions about what conclusions can be drawn from the scientific voice
literature (Callaghan, 2000; Miller, 1998).
Empirical singing studies define ideal vibrato parameters of rate, extent and onset by
singing genre, singing style and decade of study. Such studies explore the optimal
parameters for good vibrato, but most of these produce means based on multiple
singers’ data. Each parameter varied significantly across studies and results were related
to the decade of study, the style of singing and the musical stimuli performed. Studies in
vibrato rate (Horii, 1989) or extent (Seidner et al., 1995) in individual singers have been
unable to define consistent intra- or inter-singer vibrato parameters in professional
singers. Prame’s (1994) detailed model of vibrato identified the unique factors of
vibrato rate: intra-tone (features of vibrato cycles within an individual tone), intertone
(comparisons of vibrato rates in the 25 tones analysed for each artist) as well as inter-
artist (variations of mean rates across artists). Loudness or differences in SPL have been
associated with changes in vibrato parameters (Michel & Myers, 1991; Titze et al.,
1999), yet pedagogy perceives an evenness in vibrato across dynamic changes (Miller,
1996).
50
Few studies compare the same singers’ vibrato parameters in different tasks. While
vibrato is believed to be intrinsic to a singer’s sound, it does change across musical
styles (Easley, 1932; Hakes, Shipp, & Doherty, 1987). Differing emotional instructions
have affected vibrato parameters (Rothman & Arroyo, 1987; Sundberg, 1997), but to
date, no study has assessed the impact of a specific singing technique on vibrato, rather
than respective vibrato parameters of a group of singers. A singer’s technical intention
may play as great a role in the production of vibrato as any musical or stylistic changes
they make to their singing.
Six advanced singing students sang in two conditions: “optimal” (O), using maximal
open throat, “sub-optimal” (SO), using reduced open throat and “loud sub-optimal”
(LSO) to control for the effect of loudness. Studies that compare different vocal
instructions have identified changes in overall sound pressure (SPL) or power levels as
a direct result of an experimental condition (Foulds-Elliott et al., 2000; Rossing,
Sundberg, & Ternstrom, 1986, 1987; Thorpe et al., 2001). Therefore, the LSO condition
was included, in accordance with Foulds-Elliot et al. (2000) to take into account
possible effects of loudness, which may impact on vocal timbre.
The first acoustic study analyses changes to vibrato parameters of rate, extent and onset
using maximum open throat and reduced open throat technique. Analyses revealed
significant reductions in vibrato extent and increases in onset time, but no change to
vibrato rate in SO compared to O. This study was the first to link changes in vibrato
parameters with the use of a technical instruction rather than a particular singing genre
or musical era. Results of this study were presented to an international music
51
conference, Music and Gesture [Mitchell, H. F., & Kenny, D. T., Evaluation of open
throat technique on classical singing, International conference presentation at Music and
Gesture, August 2003, East Anglia, UK].
52
PAPER 2
Mitchell, H. F., & Kenny, D. T. (2004). The impact of open throat technique on vibrato rate, extent and onset in classical singing.
Logopedics Phoniatrics Vocology, 29(3), 99-118.
53
The impact of ‘open throat’ technique on vibrato rate,extent and onset in classical singing
Helen F. Mitchell and Dianna T. Kenny
From the Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music, TheUniversity of Sydney, New South Wales, Australia
Received 8 September 2003. Accepted 3 August 2004.
Logoped Phoniatr Vocol 2004; 29: 171�/182
Mitchell, Kenny et al . (2003) identified ‘open throat’ as integral to the production of an even and consistent sound in classicalsinging. In this study, we compared vibrato rate, extent and onset of six advanced singing students under three conditions:‘optimal’ (O), representing maximal open throat; ‘sub-optimal’ (SO), using reduced open throat; and loud sub-optimal (LSO),using reduced open throat but controlling for the effect of loudness. Fifteen expert judges correctly identified the soundproduced when singers used open throat with 85% accuracy. Having verified the technique perceptually, we used a series ofunivariate repeated measures ANOVAs with planned orthogonal contrasts to test the hypotheses that frequency modulationsassociated with vibrato rate, extent and onset would vary outside acceptable or desirable parameters for SO and LSO.Hypotheses were confirmed for vibrato extent and onset but not for rate. There were no significant differences between SO andLSO on any of the vibrato parameters. As vibrato is considered a key indicator of good singing, these findings suggest thatopen throat is important to the production of a good sound in classical singing.
Key words: open throat, singing technique, vibrato, vocal pedagogy.
Helen Mitchell, Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music, TheUniversity of Sydney, New South Wales, Australia 2006. Tel.: �/61 2 9351 9644. Fax: �/61 2 9351 9540. E-mail:[email protected]
INTRODUCTION
The principal goal of classical and operatic singing
pedagogy is the production of a voice of quality that is
achieved through instruction in specific vocal techni-
ques and exercises to achieve the desired overall
sound. Instruction in the use of ‘open throat’ techni-
que to achieve a good sound is widespread (1�/4).
Historically, the term ‘open throat’ can be traced
throughout the vocal pedagogy literature (5�/7) as a
way of describing freedom or lack of tension in the
area of the throat, resulting in a lack of constriction
and a better tone quality. As early as 1935, Bartholo-
mew commented: ‘If the various tricks of the trade
that voice teachers use to improve quality are ana-
lyzed, most or all of them will be found to be a device
for directly or indirectly enlarging the throat.’ (8)
According to the singing literature, ‘open throat’ is a
complex maneuver that involves a pedagogical in-
struction and a perceived sensation or action that
results in a specific sound quality. Previous pedagogi-
cal strategies have linked ‘open throat’ to an indirect
action in the preparation to sing, on inhalation in the
imposto (9, 10), appoggio or as part of ‘breath
management’ (3, 11), or in the surprise breath or
smelling the rose (3, 9, 12). Pedagogues also used
imagery though visualization of space within the
throat, through an ‘air ball’ or ‘soap bubble’ to achieve
the posture of ‘open throat’. The action of yawn (13,
14) or yawn-sigh (2) was recommended, but pedago-
gues cautioned against yawning to the point of distor-
tion of sound through unnecessary tension (2, 3).
Perceptually, open throat produces a tone ‘free of
constrictor tensions’ ((2), p. 83). The sound quality
attributed to open throat is perceived in resonance
(3, 15), roundness (16), freedom (17), purity (18),
richness and warmth (19). Use of the technique
facilitates technical dexterity and better use of other
techniques, such as ‘ring’ (1, 2).
There are differing pedagogical perspectives of the
vocal mechanics involved in ‘open throat’ technique.
Pedagogues who explain their pedagogy in physiolo-
gical terms link their preferred strategy to achieving
open throat to lowered larynx, pharyngeal width and
raised soft palate (3, 11, 15, 20). Singing pedagogy also
hypothesized that ‘open throat’ counters constriction
�ORIGINAL ARTICLE �
# 2004 Taylor & Francis. ISSN 1401-5439 Logoped Phoniatr Vocol 29
DOI: 10.1080/14015430410001033 54
which may result in a reduction of pharyngeal space
(2). Pedagogues generally understand that open throat
is not necessarily an ‘enlarged pharynx’ or action that
may excessively depress the hyoid bone, tongue (2) or
larynx (3) as that may cause tension, and limit
resonance in the sound quality.
A recent study examined current understanding and
use of open throat in the classical singing studio.
Mitchell et al . (4) interviewed 15 singing pedagogues
and results indicated that all 15 pedagogues described
‘open throat’ technique as fundamental to singing
training. Although some pedagogues preferred differ-
ent terms, such as ‘retraction’, ‘freedom’ or ‘collar’,
they were in fact describing the same underlying
phenomenon of ‘open throat’. They recognized it in
the final sound as balanced and coordinated, free,
even or consistent, warm and open. Expert pedago-
gues in this study believed that open throat maximizes
pharyngeal space and/or abducts the ventricular folds
(4). While some vowels, for example [a], cannot be
produced without pharyngeal constriction (21) and
hence pharyngeal space cannot be maximized to the
same degree on every vowel, in the bel canto (Western
classical singing) tradition, the ideal is to achieve the
same warm, open voice quality on every note (10).
Story et al . (22) using MRI (magnetic resonance
imaging) techniques have confirmed that different
vocal postures are associated with different vocal
qualities such as ‘yawny’ and ‘twangy’. ‘Yawny’
involves widening of the oral cavity and lengthening
of the vocal tract while ‘twangy’ is produced by slight
constriction of the oral cavity and shortening the vocal
tract. Open throat is related to the ‘yawny’ quality
(2, 11) but pedagogically the yawn is considered too
exaggerated for classical singing. A pre-yawn or cry/
sob instruction is preferred by most pedagogues (2, 3,
23, 24). Story et al . (22) analyzed these physiologically
established vocal qualities acoustically but not percep-
tually, but recommended that perceptual assessment
was necessary to complement the acoustic and phy-
siological findings.
It may be possible to understand good classical
singing through acoustic features associated with
subjective descriptions of acceptability or beauty in
singing. Vibrato, for example, is widely accepted as a
component of singing, but despite acoustical, physio-
logical and perceptual studies (15, 25�/28), it has been
difficult to define its most desirable parameters. It is,
however, possible to trace different preferences over
the century (29) and also to assess its defects or
undesirable qualities (3, 15). However, acoustic studies
of undesirable qualities of vibrato have not achieved
consensus in their definitions (30, 31).
In this study, we investigated the technique of open
throat by identifying acoustic differences from the
same singers when they consciously manipulated the
technique in their singing. As pedagogues in a
previous study by these authors (4) reported thatsingers normally use open throat in good singing, and
were conscious of using the technique, we believed that
experienced singers could manipulate their use of this
technique. The aim of this study was 1) to test the
perceptual distinctiveness of the sound produced by
singers using open throat and 2) to determine the
measurable differences between a sound quality using
maximal open throat and a sound that used a lesserdegree of open throat, while holding other aspects of
singing technique or sound quality constant.
METHOD
Participants
Six female singers (three sopranos and three mezzo-
sopranos) volunteered to participate in this study.
They were advanced students with excellent techniqueof an experienced singing pedagogue, who is a
Lecturer in Vocal Studies and Opera at a state
Conservatorium of Music in Australia. It is the
premier institution for musical education in the
country and has produced singers of international
repute. The pedagogue had 30 years’ teaching experi-
ence, a private studio consisting of international and
national singers (32) and is considered to be one of thetop five singing teachers in Australia. Criteria for
participant selection included, through this pedago-
gue’s assessment, singers who: 1) had a good classical
singing technique for their level of training and
experience; and 2) understood and demonstrated
skillful control of ‘open throat’ or ‘retraction’ techni-
ques in their singing. The institutional human ethics
committee approved the study.Prior to the voice recording, participants completed
a questionnaire seeking information on age, years of
singing study, number of years of study with each
singing teacher and highest qualifications attained or
currently undertaken in music and/or singing, and
singer type (soprano or mezzo-soprano). The partici-
pants were also asked to classify the genres of singing
they performed in public (opera, classical, choral,music theatre and contemporary) in accordance with
the Bunch and Chapman taxonomy (32) and to
estimate the percentage each style played in their total
performing career.
The demographic information of the participants is
presented in Table 1. Participants were aged between
23 and 30 years, with a mean of 26 years. All had
studied singing for at least 7 years (average 9.8 years)and had spent an average of 5 years studying with
their present singing teacher. Each singer held a
172 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2955
qualification in singing or music (four had Bachelor of
Music degrees and two had diplomas, in music and/or
singing) and five of six were currently undertaking asecond degree in singing (three postgraduate Diploma
of Opera and two Bachelor of Music degrees). All
defined the majority of their singing as operatic
(�/50%), with the second most common style classical
(�/20%), in accordance with the Bunch and Chapman
taxonomy (32) of singing voices. All reported that they
were in good health and able to perform the tasks.
Procedure
Singers were sent information about the project and
were invited to take part in an acoustic and perceptual
study of singing technique. They were required to
attend a single recording session lasting approximately
one hour. They were advised that the object of the
study was to investigate the acoustical and perceptual
features of the use of open throat in singing and toinvestigate the sound qualities associated with the use
of open throat.
Experimental protocol for assessing sound qualities
produced by open throat
A protocol was developed to assess the effect of ‘open
throat’ technique in singing. Two musical tasks werechosen to investigate the use of the technique in two
song excerpts. Prior to singing, each singer selected the
sequence of their tasks before commencing the experi-
ment by selecting a blank card, the reverse side of
which represented one of the tasks, to reduce the
possible effects of task-order.
Singers sang each of the two song excerpts under
three conditions: optimal, sub-optimal and loud sub-optimal. ‘Optimal’ (O) condition was necessary to
provide a sound quality with the best technique that
they could. This involved the maximal use of open
throat. ‘Sub-optimal’ (SO) condition involved the use
of a reduced (open throat) technique but still with an
acceptable singing technique and without consciously
altering any other aspect of their technique. It was
hypothesized, from interviews with pedagogues (4),that the SO condition would result in a reduction of
sound pressure level (SPL), so a third condition ‘loud
sub-optimal’ (LSO) involved the same instruction as
the SO condition, but with the added instruction that
the singer should try to achieve a louder dynamic than
in SO. This addressed SPL as an additional variable to
the SO condition and was similar to technical condi-
tions established by Foulds-Elliot et al . (33) when SPLwas a variable of degrees of emotional connection in
operatic singers.Tab
le1.
Su
mm
ary
of
sub
ject
cha
ract
eris
tics
.D
escr
ipti
on
of
sub
ject
nu
mb
er,
age,
the
nu
mb
ero
fyea
rsea
chsu
bje
cth
as
tak
ensi
ng
ing
less
on
s,th
en
um
ber
of
tea
cher
sea
chsu
bje
cth
as
ha
d,
the
nu
mb
ero
fyea
rsea
chsu
bje
ctst
ud
ied
wit
hth
eir
thre
em
ost
rece
nt
tea
cher
s(
T1� /T
3)
,th
eh
igh
est
leve
lo
fed
uca
tio
nco
mp
lete
dby
each
sub
ject
,th
ecu
rren
tleve
lo
fed
uca
tio
nu
nd
erta
ken
,vo
ice
typ
e,p
rop
ort
ion
of
each
sub
ject
’sca
reer
spen
tsi
ng
ing
op
era
,cl
ass
ica
l,ch
ora
l,m
usi
cth
eatr
e,
con
tem
po
rary
or
oth
erst
yle
so
fm
usi
c.
Su
bje
ctA
ge
(yea
rs)
Yea
rso
fle
sso
ns
No
.o
fte
ach
ers
T1
T2
T3
Hig
hes
ted
uca
tio
nC
urr
ent
edu
cati
on
Vo
ice
typ
eO
per
aC
lass
ica
lC
ho
ral
Mu
sic
thea
tre
Co
nte
mp
Oth
er
12
68
41
13
Ba
chel
or
Ba
chel
or
Mez
zo8
0%
20
%2
25
73
1.5
34
.5B
ach
elo
rP
ost
gra
du
ate
So
pra
no
80
%2
0%
33
01
22
10
2D
iplo
ma
N/A
So
pra
no
97
%3
%4
23
10
24
5.5
Ba
chel
or
Po
stg
rad
uat
eS
op
ran
o5
0%
45
%5
%5
25
10
27
.52
.5D
iplo
ma
Po
stg
rad
uat
eM
ezzo
73
%2
5%
2%
62
81
26
51
.51
.5B
ach
elo
rB
ach
elo
rM
ezzo
50
%3
5%
5%
10
%
Evaluating open throat 173
Logoped Phoniatr Vocol 2956
Each task was performed twice in the O and SO
conditions while the LSO condition was performed
only once. In total, each singer performed eachmusical task five times.
Instructions
A pedagogue was present during the recording ses-
sions to provide accompaniment for warm-ups and
practice of the tasks where necessary and to instruct
singers to achieve the required vocal postures for eachexperimental condition. For example, she instructed
them to pay attention to producing the most open
sound in their throat in the O condition, and a lesser
degree in the SO and LSO conditions. Some singers
asked how to produce the LSO condition and the
pedagogue instructed them to ‘use more twang’, as
taught in their lessons (23).
Reliability check
Prior to the commencement of the study, the partici-
pating students and pedagogue had several
practice sessions in which the singer was instructed
to use either O or SO at random. The pedagogue
indicated which technique she thought the singer
had used. Practice with each singer continued until
both the singer and the pedagogue reached 100%agreement on the occasions that the technique was
applied.
The musical tasks
Musical tasks were chosen in order to test different
demands of good singing, but were not musically
difficult. They were designed to test the use of openthroat, and contained musical features derived from a
previous qualitative study on the use of the technique,
where use or lack of the technique was deemed to be
particularly valuable or noticeable (4). These features
were: high tessitura, sustained or legato singing,
dynamic range control, and vocal agility.
The Mozart song Ridente la Calma , K 152, bars 1�/
27 (Fig. 1a) was selected as it is a nominally simplesong in the Italian language (34) with a mixture of
common musical statements involving repeated legato
lines as well as the initial stylized leaps of a major 4,
and short scale figures. All six singers sang this aria in
the same key (F-major).
The third verse of the Schubert lied, Du bist die Ruh
D. 776 (Op. 59, No. 3) (Fig. 1b), bars 54�/80 was
chosen for its demanding vocal control, sustainedmusical line and high climactic tessitura. The three
sopranos and three mezzo-sopranos sang this in an
appropriate key depending on soprano or mezzo-
soprano voices (E-flat, D-flat and C-major).
Recording
Participants were given time to warm up in the singing
studio and become familiar with the room before
recording. Recording levels for each singer were set
during this time. The acoustic signal was recorded
digitally (Behringer Ultragain preamplifier/Marantz
CDR 630) via a high-quality microphone (AKG C-
477) positioned on a head boom a constant 7 cm
distance from the subject’s lips. This ensured that
direct energy of the performers’ voices was recorded
rather than room reflections, enabling us to use a
studio environment with low ambient noise rather
than an anechoic studio (35).
Fig. 1a. The Mozart task, Ridente la Calma , K 152,
bars 1�/27 with the numbering of 37 notes selected for
analysis.
Fig. 1b. The Schubert task, Du bist die Ruh D. 776
(Op. 59, No. 3), bars 54�/80 with the numbering of 20
notes selected for analysis.
174 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2957
Calibration was carried out in each recording by
using a recording of two pink noise samples on
audiotape played immediately following each record-ing session at the same recording gain used for
recording the singer’s voice. Pink noise is a broadband
steady state signal which enables calibration of an
audio playback system across the frequency range for
tasks such as subjective testing and, if the analysis is
just done on computer, the noise is at least as good as
any other steady-state known signal. For calibration of
absolute sound pressure levels (SPL), the microphoneof a sound level meter (Rion NL-06 SPL) was held
adjacent to the AKG microphone from a speaker
(Bose Lifestyle) from which the pink noise was played.
The SPL (in dB at 7 cm) recorded by the sound level
meter was noted for the two noise signals and used
later for calibration.
The CDs of the recordings were marked by
subject number. The audio recordings were acquiredto computer at 16 kHz with a Loughborough Sound
Images PC/C32 board using Phog Version 2.0
and analyzed using Soundswell Version 4.00 (Hitech,
Sweden) software to produce nine channels of raw
and calculated data. A real-time digital SPL meter
was used to set a recording gain so that the pink
noise calibration level recorded in Phog was the
same (�/ 0.5 dB) as the SPL level noted duringrecording.
Perceptual test of open throat
Because open throat is a pedagogical concept that has
not yet been verified empirically, it was necessary to
determine whether expert pedagogues could reliably
distinguish the sound quality associated with this
instructional technique. Accordingly, a perceptual
test was designed to determine the degree to whichexpert pedagogues agreed that a sound was indicative
of the application of open throat. Judges who assessed
singers were not the singers’ pedagogues and did not
know any of their identities.
Listeners assessed 30 tracks (6 singers�/2 condi-
tions (O and SO)�/2 musical tasks) in random order,
including six repeats, also selected at random. (A pilot
test found that LSO did not produce a sufficientlydiscernible voice quality to be correctly distinguished
from SO. As LSO was created as acoustic confirma-
tion of SO, it was considered unreasonable to expect
the human ear to detect so small a difference in timbre.
The LSO condition was therefore not included in the
perceptual study.)
The perceptual test was conducted in a quiet
environment and samples were played from a SonyCD Walkman (DEJ885W) via circum-aural closed-
back stereo monitoring headphones (Sennheiser HD
270). Prior to presentation of stimuli, participants
were given information on the two singing conditions,
O and SO, and were presented with the musical scoreof each musical task. They were asked to judge
whether or not the singer was using open throat
technique in each sample.
Measurement of sound pressure level (SPL)
To calculate mean SPL, the SPL channel of the
Soundswell program was calibrated with the two
known SPL levels of the pink noise samples for eachsinger. The histogram function was applied to the SPL
channel in the Soundswell program for each complete
rendition of musical task and generated the mean,
standard deviations (SD), and percentile values of
SPL in dB.
Vibrato measurement
Vibrato cycles were measured manually from the
spectrogram feature in the Soundswell program using
a fine resolution Hamming analysis window, FFT
length 1024, with a bandwidth of 30 Hz. A cursor
point logged the time and frequency of each vibrato
peak and trough and the data exported to spread-
sheets. Frequency was measured on a high harmonic
across all subjects, as resolution increases with thepartial number (26). Fundamental frequency of each
logged point was calculated by dividing it by its partial
number. Data were manipulated within spreadsheets
(Excel, Microsoft 2000). Thirty-seven notes were
measured in the Mozart task and 20 in the Schubert.
The vibrato onset (VO) in seconds (s) was measured
from the initiation of phonation on the particular note
until the first conclusive peak of the vibrato cycle.Onset was only considered following the start of a
word or after a break in phonation; 18 VO occurrences
were studied in the Schubert task and 23 in the Mozart
task.
The vibrato rate (VR) in hertz (Hz) and vibrato
extent (VE) in semitones (ST) were measured from the
first identifiable peak until the last trough of the note.
The VR of each cycle was measured as the timedifference between adjacent vibrato peaks and calcu-
lated in Hz. The VR for each note was calculated as
the mean of all successive VR cycles of the note. If
vibrato peaks and troughs could not be discerned by
visual inspection of the spectrograph, a nil result was
recorded.
Vibrato extent in semitones (ST) was calculated by
dividing the peak-to-trough ST values by 2 (28). Afrequency point in Hz of a peak or trough was
calculated in ST using the formula:
Evaluating open throat 175
Logoped Phoniatr Vocol 2958
F0 in ST�12(log10(F0) � log10 16:35)
log10 2[1]
The semitone difference between adjacent cycle
peaks and troughs was calculated as:
Peak-to-trough(ST)
�12(log10(peak F0) � log10(trough F0))
log10 2[2]
In this study, VE in ST represents the distance from
peak to the mean F0 of the vibrato cycle or from the
mean F0 of the cycle to the next trough (28).
Frequency values from successive vibrato cycles wereaveraged for each note studied.
Study 1: Perceptual verification of open throat. The
total number and percentages of correct responses foreach O and SO sample for each listener and for the
group as a whole were calculated. Intra-rater reliability
was also assessed by calculating the percentage of each
repeated pair that was correctly identified by each
judge.
Study 2: Vibrato parameters of open throat. The
study design was a repeated measures (three depen-
dent measures: vibrato rate (VR); vibrato extent (VE);
vibrato onset (VO)) randomized complete block with a
2 (task (Mozart versus Schubert))�/3 (condition
(optimal, sub-optimal and loud sub-optimal)) factor-
ial structure with planned orthogonal contrasts oneach of the three dependent variables.
Data were analyzed firstly through a series of fixed
and mixed effect univariate ANOVAs with three
factors (task, condition, subject). In the first series,
all three factors were entered as fixed effects; in the
second, subject was entered as a random effect.
Finally, the data were subjected to analysis using the
general linear model (GLM) with planned orthogonalcontrasts. Since all analyses yielded the same out-
comes, only the results of the contrasts are presented.
Main effects for task and condition and interaction
effects (contrasts) were calculated for each dependent
measure (VR, VE and VO). In the first contrast, O was
compared to SO and LSO and in the second contrast
SO was compared to LSO.
Hypotheses
The following hypotheses were generated:
1) Study 1: Expert judges would reliably identify the
use of open throat in samples of classical singing.
2) Study 2: SPL would decrease from O to SO.3) Study 2: Frequency modulations associated with
vibrato rate, extent and onset would vary outside
acceptable or desirable parameters established in
singers’ O singing for SO and LSO.
RESULTS
Hypothesis 1: Perceptual verification of open throat
Expert listeners correctly identified 85.3% O and
80.0% SO. Intra-judge reliability using the six repeated
samples was 83.33% (that is to say, judges, on average,
rated five out of six singing sample pairs correctly).
Hypothesis 2: SPL would decrease from the O to SO
condition
Hypothesis 2 stated that SPL would decrease from the
O to SO condition. Hypothesis 2 was not confirmed
from O to SO F(1,5)�/0.010, p�/0.924). The mean SPL
for O was 97.38 dB (SD 1.86) in the Mozart task and97.65 dB (SD 2.61) in the Schubert task. In the SO
condition, the mean SPL was lower (96.18 dB in
Mozart and Schubert). In LSO, each singer’s SPL was
associated with SPL levels that were either similar to
the O condition (6 examples B/2 dB different) and in 9
of the 12 LSO, samples increased SPL from the level
of the original O in the LSO condition (LSO means:
99.18 dB, 98.92 dB).
Vibrato
Data screening. The distributions for each depen-
dent measure (VR, VE, and VO) were assessed for
normality and outliers. Examination of skewness and
kurtosis statistics indicated that the distributions were
relatively normally distributed. Univariate tests for
homogeneity of variance (Levene’s Test of Equality of
Error Variances) for each of the dependent measures
(VR: F�/5.7, p B/0.001; VE: F�/2.65, p B/0.001; VO:F�/9.699, p B/0.001) indicated that this assumption
was violated for all three measures. Univariate F-tests
(with Bonferroni adjustment for three tests: p B/0.017)
indicated that task, condition and subject were all
significant, hence a very conservative p-value of p B/
0.005 was set for all subsequent analyses.
Descriptives. The complete vibrato data are pre-
sented in Table 2, showing the overall means from
averaging all measured notes sung by each singer, in
each condition, for vibrato rate (VR), vibrato extent
(VE) and vibrato onset (VO) in each musical task
(Mozart and Schubert).Differences in VR, VE and VO were considered
separately for effects of condition and task.
176 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2959
Hypothesis 3: Frequency modulations associated with
vibrato rate, extent and onset would vary outside
acceptable or desirable parameters established in
singers’ O singing for SO and LSO
Hypothesis 3a: For vibrato extent, singers’ F0 would
show greater variation from the mean in SO/LSO
compared to O
Hypothesis 3a stated that VE would show greaterincreases or decreases from mean fundamental fre-
quency in SO and LSO compared to O. Hypothesis 3a
was confirmed as overall means of VE by condition
indicated a significant reduction from O to SO/LSO in
both musical tasks. For VE, there was a main effect for
contrast 1 (F(1,5)�/20.821, p�/0.006) but not for
contrast 2 (F(1,5)�/1.448, p�/0.283). There was no
effect for musical task (F(1,5)�/0.032, p�/0.865). Fig. 2presents these data. There was also a significant
interaction between musical task and contrast 1
(F(1,5)�/12.136, p�/0.018), but not for musical task
and contrast 2 (see Fig. 2); that is, VE was wider for
both Mozart and Schubert in the O condition, but the
same for both musical tasks in the SO and LSO
conditions. Table 2 shows that all subjects’ mean VE
decreased in both the SO and LSO conditions by 0.09
ST to 0.92 ST, with an average reduction of 0.50 ST.
Inspection of the narrow standard deviations con-
firmed that singers’ reductions of VE in SO and LSO
were similar for all subjects. Although VE in semitones
changed overall in each condition, the singers’ varia-
bility remained slight (SD5/0.25) for notes within a
condition. The time plots in Fig. 3 plot the consecutive
VE of the notes measured for vibrato cycles in each
musical task (37 in Mozart, 20 in Schubert) in the O
and SO conditions for subjects 4 and 5. These present
data from subject 4, who showed a typical decrease
from O to SO, and subject 5, who showed a smaller
reduction in VE. In some SO vibrato cycles, there was
no measurable VE. Subjects 3 and 4 maintained the
Table 2. Mean vibrato rates in Hz, mean vibrato extent in semitones and mean vibrato onset in seconds, for the
Mozart and Schubert musical tasks in each condition, optimal (O), sub-optimal (SO) and loud sub-optimal (LSO);
standard deviations (SD) are given in parentheses.
Rate Extent Onset
Condition Condition Condition
Task Subject O SO LSO O SO LSO O SO LSO
Mozart 1 5.96 5.93 5.92 1.16 0.68 0.63 0.09 0.22 0.21(0.22) (0.28) (0.40) (0.15) (0.14) (0.16) (0.07) (0.17) (0.17)
2 5.63 5.32 5.36 1.26 0.77 0.73 0.15 0.16 0.21(0.28) (0.40) (0.29) (0.14) (0.16) (0.15) (0.17) (0.17) (0.09)
3 5.58 5.58 5.58 1.63 0.81 0.81 0.19 0.24 0.24(0.40) (0.29) (0.27) (0.16) (0.15) (0.21) (0.17) (0.09) (0.12)
4 5.65 5.60 5.51 1.48 0.60 0.60 0.12 0.28 0.27(0.29) (0.27) (0.31) (0.15) (0.21) (0.17) (0.09) (0.12) (0.10)
5 6.80 6.78 5.51 1.00 0.72 0.60 0.09 0.22 0.27(0.27) (0.31) (0.24) (0.21) (0.17) (0.20) (0.12) (0.10) (0.07)
6 7.27 6.28 6.63 0.85 0.52 0.67 0.10 0.27 0.25(0.31) (0.24) (0.36) (0.17) (0.20) (0.24) (0.10) (0.07) (0.11)
Mean 6.15 5.93 5.97 1.23 0.68 0.67 0.12 0.23 0.24(0.71) (0.53) (0.61) (0.29) (0.10) (0.08) (0.04) (0.04) (0.03)
Schubert 1 5.79 5.57 5.63 1.00 0.58 0.61 0.12 0.56 0.63(0.31) (0.33) (0.28) (0.20) (0.15) (0.15) (0.13) (0.43) (0.47)
2 5.23 5.06 5.18 1.32 1.19 0.88 0.22 0.26 0.42(0.32) (0.14) (0.14) (0.14) (0.21) (0.25) (0.10) (0.12) (0.16)
3 5.68 5.67 5.63 1.32 0.60 0.51 0.13 0.32 0.46(0.23) (0.26) (0.44) (0.25) (0.19) (0.15) (0.09) (0.27) (0.44)
4 5.66 5.70 5.65 1.44 0.61 0.58 0.13 0.22 0.10(0.23) (0.49) (0.37) (0.24) (0.19) (0.16) (0.10) (0.17) (0.07)
5 6.70 6.77 6.68 1.07 0.85 0.71 0.18 0.18 0.23(0.25) (0.26) (0.51) (0.19) (0.18) (0.12) (0.08) (0.13) (0.16)
6 5.96 5.50 5.88 0.90 0.77 0.81 0.18 0.20 0.15(0.73) (0.52) (0.75) (0.20) (0.14) (0.19) (0.10) (0.12) (0.11)
Mean 5.84 5.71 5.78 1.18 0.77 0.68 0.16 0.29 0.33(0.49) (0.57) (0.50) (0.22) (0.23) (0.14) (0.04) (0.14) (0.20)
Evaluating open throat 177
Logoped Phoniatr Vocol 2960
two highest mean VEs overall, but also the greatest
variance (SD 0.25 and 0.24 ST respectively) but only inthe Schubert task (Table 2). The greatest VE seems to
be in the Mozart task, not the Schubert task (Table 2).
Hypothesis 3b: Vibrato rate would show greater
variations in SO/LSO compared to O
For VR, the hypothesis that vibrato would show
greater variations in SO and LSO compared to O
was not confirmed for task or condition. Overall
means of VR were not significantly different in each
task (F(1,5)�/2.415, p�/0.181). There was no effect of
condition for contrast 1 (F(1,5)�/2.828, p�/0.153) or
contrast 2 (F(1,5)�/0.629, p�/0.464), that is, VR in O,
SO and LSO conditions was not significantly different.
There were no interaction effects for task and condi-
tion. Fig. 4 presents the results of this analysis
graphically.
The mean VR across all subjects, conditions and
tasks was 5.92 Hz (SD 0.11). Four of the subjects
displayed a relatively similar mean VR across all
conditions and tasks (5.3�/5.8 Hz), whereas two
Condition
LSOSOO
Ext
ent i
n se
mito
nes
1.4
1.2
1.0
0.8
0.6
Mozart
Schubert
Fig. 2. Vibrato extent. Main effect of condition incontrast 1, that is to say, between optimal (O) and sub-
optimal (SO)/loud sub-optimal (LSO). Interaction
effect of musical task (Mozart and Schubert) for
contrast 1, that is, between optimal (O) and sub-
optimal (SO)/loud sub-optimal (LSO).
Condition
LSOSOO
Rat
e in
Hz
6.5
6.0
5.5
5.0
Mozart
Schubert
Fig. 4. Vibrato rate: no statistical effects of task
(Mozart and Schubert) or condition (optimal (O)
and sub-optimal (SO)/loud sub-optimal (LSO)).
0.
1.1.2.
0
050
0 5 10 15 20 25 30 35
0.0
1.01.52.0
0 5 10 15 20 25 30 35
0 5 10 15 200.00.5
0.5
1.1.2.
050
0 5 10 15 200.
1.1.2.
0
050
Mozart Note
Mozart Note
Schubert Note
Schubert Note
Exte
nt
in S
T
Exte
nt
in S
TE
xte
nt
in S
T
Exte
nt
in S
T
Optimal Sub-optimal
(a)
(b)
0.5
0.5
Fig. 3. Time plots of mean vibrato extent in semitones in the Mozart task, notes 1�/37 and the Schubert task in
the optimal and sub-optimal conditions: a�/subject 4; b�/subject 5.
178 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2961
subjects each showed a higher mean VR of 6.8 Hz as
shown in Table 2. Subject 6 was an outlier.
Hypothesis 3c: Vibrato onset would be longer in SO/
LSO compared to O
Hypothesis 3c stated that VO would be longer in SO/
LSO compared to O. Hypothesis 3c was confirmed as
overall means indicated a significant increase from O
to SO/LSO in both musical tasks. For VO, there was a
main effect for contrast 1 only (F(1,5)�/13.118, p�/
0.015), that is, VO was significantly less in the O
condition than in either SO or LSO. There were no
differences in VO between SO and LSO (F(1,5)�/0.905,
p�/0.385) and there was no effect for musical task
(F(1,5)�/1.559, p�/0.267) There were no interaction
effects between task and contrasts, and VO was
affected by condition in the same way in each task.
Fig. 5 presents the results of this analysis graphically.
A visual example of the changes to vibrato para-
meters by condition is given in Fig. 6. Each panel of
Fig. 6 represents the same singer and the same 3-
second portion of ‘Von deinem Glanz’ in the Schubert
musical task across each condition (O, SO and LSO).
The vibrato cycles are less regular or consistent in each
of the SO and LSO conditions, and in the LSO in
particular, the cycles are erratic and unstable.
There was no vibrato on 75 of a total of 684 notes
chosen for measurement. In the Mozart condition, 73
notes had no vibrato and 46 of these occurred in SO.
An example of this is shown in Fig. 6c.
DISCUSSION
We hypothesized that open throat technique would
produce audible differences in vocal quality and this
was confirmed by the perceptual study. Listeners were
reliable in their judgements and consistently identified
the technique in repeated samples. Singers’ vocal
quality produced discernable sound quality differences
in 85.3% of O and 80.0% of SO samples, which
listeners attributed to the pedagogical instruction of
‘open throat’. This study confirmed pedagogues’
recognition of a fundamental technique in classical
singing previously defined by expert pedagogues by a
specific sound quality (4). Further analysis of the
perceptual qualities of this technique is the subject of
ongoing research by these authors.
This study had a number of methodological
strengths. For example, we considered it important
Condition
LSOSOO
Ons
et in
sec
onds
0.4
0.3
0.2
0.1
Mozart
Schubert
Fig. 5. Vibrato onset: main effect of condition in
contrast 1, that is to say, between optimal (O) and sub-
optimal (SO)/loud sub-optimal (LSO).
Fig. 6a�/c. Examples of vibrato from subject 3 singing
‘Von deinem Glanz’ in each of the three conditions,
optimal, sub-optimal and loud sub-optimal (O, SO,
LSO) respectively.
Evaluating open throat 179
Logoped Phoniatr Vocol 2962
to demonstrate that voice quality produced during SO
was not simply the result of changes (that is to say,
reductions) in SPL. The inclusion of a second
comparison group, LSO, indicated that SPL did not
account for differences between O and SO. Foulds-
Elliot (33) and Rossing (36) have used similar study
designs. The hypothesized reduction in SPL did not in
fact occur; however, there were differences in the
energy distributions of each condition that will be
explored further in future studies on this technique. A
second methodological strength was the use of a
homogenous group of singers of similar skill and
training who could consistently apply the technique.
While other studies have used the same sex subjects
(37, 38), singers in previous studies have ranged from
novice to advanced students. Studies assessing voice
qualities in which subjects vary in amount of singing
training or experience associate certain qualities with
levels of training, and find differences in vibrato or
energy which could be attributed to training. In some
cases, singing studies are unable to correlate length of
training with improved sound quality or any measur-
able acoustic feature of voice (39), suggesting it is not
length of training per se but quality or type of training
that is the critical factor.
Having established that the overall sound quality of
open throat technique is perceptually identifiable, we
subsequently assessed vibrato, a major defining fea-
ture of vocal quality. Although vibrato is taken for
granted as a feature of the classical singing voice, as a
component of tone quality, ‘richness’ (15), ‘vibrancy’
(3) or ‘resonance’ (40), it is widely accepted in the
singing literature that vibrato outside acceptable rate
and extent parameters is indicative of poor technique
(3, 15) and inferior sound quality. Vibrato rate and
extent are important ways to communicate emotion in
singing. For example, deep, slow vibrato is perceived
as sad, while fast, irregular vibrato communicates fear
(41). Vibrato may vary for other reasons; for example,
for dramatic purposes (42); in response to stylistic
requirements of particular genres (43, 44); or as a
result of changes to SPL in messa di voce (45); or for
dynamic variation (25) towards the ends of notes.
However, consistent vibrato within specified para-
meters has been linked to beautiful sound (38) and
to listeners’ overall preference (37). In the classical
singing literature, a steady and even vibrato is
universally promoted (3).
In this study, we demonstrated reliable differences in
vibrato parameters as a result of varying the degree to
which singers applied the technique of open throat.
Reduction of open throat technique for these singers
produced a significant decrease in VE in SO/LSO and
increase in VO in both SO/LSO. Singers spanned the
conventional VR parameters, from 5 Hz to �/6 Hz,
and although visual inspection showed that VR was
less even in SO/LSO, this was not statistically sig-
nificant. While singers vary vibrato for emotional
reasons, or suitability in different musical genres, these
six singers’ vibrato changed as a result of using
optimal or insufficient open throat technique. Since
vibrato parameters largely define good singing tech-
nique in the literature, open throat would appear to be
an essential element of sound vocal pedagogy.
Different genres require differing vibrato extent. For
example, singers use a wider vibrato extent in operatic
singing compared to concert singing (43) and in
classical singing compared to baroque or early music
style (44). Informal studies have found that the same
singers increased vibrato extent in dramatic lied or
Verdi operatic arias in comparison to Schubert’s ‘Ave
Maria’ (26). Changes to vibrato extent in an individual
singer are normally associated with emotional intent
(for example, agitated and peaceful in (42)). In this
study, singers reduced VE by, on average, more than
40% from O to SO/LSO. Decreased vibrato extent
associated with reduced open throat suggests that
instruction and technique are also relevant to the
production of classical voice quality, which was
perceptually equated with poor singing technique.
These findings suggest that open throat is indeed a
specific technique used to achieve a classical or
operatic quality. Our understanding of what produces
changes in vibrato extent may need to be extended to
include not only the demands of genre and emotional
intention, and to a lesser extent, the production of
particular vowels, but also its deliberate cultivation
through instruction in particular pedagogical techni-
ques such as open throat.
A singer’s vibrato rate has been identified as an
intrinsic component of his or her sound quality (28)
and is the most consistent parameter (46, 47). The
vocal system has a natural tendency to oscillate in the
5�/6 Hz frequency range. Results from this study
support this finding. Singers’ rate did not vary outside
of their O range, regardless of condition. Changes to
VR have been identified in different musical genres
(43), but even in these, VR was the most consistent
factor when the same singers sang opera or concert
songs. Our findings confirm previous data on vibrato
rate. Visual inspection of the spectrographs shows that
reduction of open throat in the SO and LSO condi-
tions was associated with greater irregularity of the
vibrato pattern compared to O, although these
differences did not reach statistical significance when
mean rates were compared across conditions. This is
probably due to the fact that mean VR could not take
into account occasions where no cycle was measured
because singers did not produce vibrato. Such cycles
occurred only in SO and LSO. Additionally, there were
180 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2963
large inter-subject differences of up to 1 Hz across
different tasks and conditions.
In pedagogical terms, if vibrato onset was delayedor absent in sustained pitches or during rapid syllables,
then the voice was not ‘natural’ and this would
indicate a faulty technique. This view concurs with
Ekholm et al . (37) who associated late onset of vibrato
(�/0.5 seconds) with the overall perception of poor
vibrato and this also was highly correlated with lower
ratings of the overall voice quality. In terms of
measurement, Prame (25) did not directly addressVO, although he rejected the first vibrato cycle in the
mean VR or VE, because the pattern had not
stabilized. In our study, both SO and LSO conditions
revealed a marked increase in VO (Table 2 and Fig.
6b�/c), which is associated with poorer overall vibrato
and inferior application of technique.
Pedagogical implications
The challenge for vocal pedagogues is to assess each
individual voice and to devise a program of technical
work to improve the basic sound. In Western classical
singing, good technique is integral to vocal quality,
and poor technique is identified by faults perceived
within the overall sound (48, 49). Research into
singing, by contrast, has hitherto focused on identify-ing acoustic parameters that characterize a ‘good’
voice by reference to findings that have been generated
by averaging data from a number of voices. The
dilemma and paradox in this approach is that the
averaged findings may not adequately represent any of
the individual voices used to generate the desirable
acoustic parameters that become the benchmark of
vocal quality. Pedagogy focuses on individual techni-ques to achieve the complete sound and evaluating
teaching techniques individually is necessary to verify
their effectiveness and reliability as teaching tools.
This study extends previous work in its attempts to
assess and verify the efficacy of specific teaching tools
by reference to their acoustic and perceptual impact
on vocal quality. While replication of these findings is
required prior to confident recommendations thatopen throat technique be applied in singing studios,
this study has demonstrated that individual teaching
techniques can be empirically assessed and their effects
verified both perceptually and acoustically.
CONCLUSION
Pedagogical and perceptual studies agree that sub-
standard vibrato is immediately identifiable and
indicates inadequate technique or poor vocal quality.Whether reduced vibrato extent and delayed vibrato
onset are consequences of reduced ‘open throat’
technique specifically, or the primary factors perceived
as ‘open throat’ or indeed any technique used in good
singing needs further investigation. Vibrato alone doesnot necessarily carry the acoustic cues of ‘vocal
quality’ to listeners (50), yet perceptually, in this study,
the best quality was associated with open throat
technique. Further acoustic analysis of O and SO
sound qualities is necessary to identify which para-
meters most closely concur with the holistic percep-
tions of expert pedagogues.
Auditory-perceptual aspects of a voice represent apsychological reality for both the singer and the
listener. Linking an acoustic parameter such as vibrato
to a commonly used instruction, such as open throat,
has the potential to provide a common terminology
for ease of communication between singers, their
teachers, and clinicians. Future research may also
define ‘open throat’ as a set of physiological para-
meters.
ACKNOWLEDGEMENTS
We thank Ms Maree Ryan for her pedagogical advice
and support, Dr Densil Cabrera and Peter Thomas for
their advice on acoustical matters and the six singers
for their willing participation in the project.
REFERENCES
1. Reid CL. Voice: Psyche and Soma. New York: JosephPatelson Music House; 1975.
2. Reid CL. A dictionary of vocal terminology. New York:Joseph Patelson Music House; 1983.
3. Miller R. The structure of singing: system and art invocal technique. New York; London: Schirmer Books;1996.
4. Mitchell HF, Kenny DT, Ryan M, Davis PJ. Definingopen throat through content analysis of experts’ peda-gogical practices. Logoped Phoniatr Vocol 2003; 28:167�/80.
5. Fields VA. Training the Singing Voice: an analysis of theworking concepts contained in recent contributions tovocal pedagogy. New York: King’s Crown Press; 1947.
6. Burgin JC. Teaching singing. Metuchen, N.J.: ScarecrowPress; 1973.
7. Monahan BJ. The art of singing: a compendium ofthoughts on singing published between 1777 and 1927.Metuchen, N.J.: Scarecrow Press; 1927.
8. Bartholomew WT. The Role of Imagery in VoiceTeaching. In: Music Teachers National AssociationProceedings for 1935; Oberlin, Ohio; 1935. p. 78.
9. Hemsley T. Singing and imagination: a human approachto a great musical tradition. Oxford; New York: OxfordUniversity Press; 1998.
10. Manen L. Bel canto: the teaching of the classical Italiansong-schools its decline and restoration. Oxford: OxfordUniversity Press; 1987.
Evaluating open throat 181
Logoped Phoniatr Vocol 2964
11. Miller R. The Open Throat (La gola aperta). In: MillerR, editor. On the Art of Singing. New York: OxfordUniversity Press; 1997. p. 17�/9.
12. Puritz E. The teaching of Elisabeth Schumann. London:Methuen; 1956.
13. Shakespeare W. The Art of Singing. Ditson: New York;1921.
14. Garcia M. Hints on Singing. Translated from the Frenchby Beata Garcia. London: E. Ascherberg; 1894.
15. Vennard W. Singing: the mechanism and the technic. 5thed. New York: Fischer; 1968.
16. Joiner JR. Charles Amable Battaille: pioneer in vocalscience and the teaching of singing. Lanham, Md.:Scarecrow Press; 1998.
17. Ware C. Basics of vocal pedagogy: the foundations andprocess of singing. New York: McGraw-Hill; 1998.
18. Marafioti PM. Caruso’s Method of Voice Production.New York: Dover Publications; 1981.
19. McKinney JC. The Diagnosis and Correction of VocalFaults. Nashville, Tennessee: Broadman Press; 1982.
20. Herbert-Caesari EF. The voice of the mind. London:Robert Hale; 1951.
21. Baer T, Gore JC, Gracco LC, Nye PW. Analysis of vocaltract shape and dimensions using magnetic resonanceimaging: vowels. J Acoust Soc Am 1991; 90(2 Pt 1):799�/828.
22. Story BH, Titze IR, Hoffman EA. The relationship ofvocal tract shape to three voice qualities. J Acoust SocAm 2001; 109: 1651�/67.
23. Estill J. Primer of Basic Figures. 2nd ed. Santa Rosa:Estill Voice Training Systems; 1996.
24. Chapman J, Davis PJ. Primal Singing Australian Voice1998; 4: 9�/11.
25. Prame E. Measurement of the vibrato rate of ten singers.J Acoust Soc Am 1994; 96: 1979�/84.
26. Prame E. Vibrato extent and intonation in professionalWestern lyric singing. J Acoust Soc Am 1997; 102: 616�/
21.27. Shipp T, Sundberg J, Hadlund S. A Model of Frequency
Vibrato. In: Van Lawrence L, editor. Transcripts of theThirteenth Symposium: Care of the Professional Voice;1984; New York: The Voice Foundation; 1984. p. 116�/
17.28. Sundberg J. Acoustic and Psychoacoustic Aspects of
Vocal Vibrato. In: Dejonckere PH, Hirano M, SundbergJ, editors. Vibrato. San Diego: Singular PublishingGroup, Inc.; 1995. p. 35�/62.
29. Rothman HB, Diaz JA, Vincent KE. Comparinghistorical and contemporary opera singers with histor-ical and contemporary Jewish cantors. J Voice 2000; 14:205�/14.
30. Keidar A, Titze IR, Timberlake C. Vibrato Character-istics of Tenors Singing High Cs. In: Van Lawrence L,editor. Proceedings of the Thirteenth Symposium: Careof the Professional Voice; 1984; New York City: TheVoice Foundation; 1984.
31. Seidner W, Nawka T, Cebulla M. Dependence of theVibrato on Pitch, Musical Intensity, and Vowel inDifferent Voice Classes. In: Dejonckere PH, HiranoM, Sundberg J, editors. Vibrato. San Diego: SingularPublishing Group, Inc.; 1995. p. 63�/82.
32. Bunch M, Chapman J. Taxonomy of singers used assubjects in scientific research. J Voice 2000; 14: 363�/9.
33. Foulds-Elliott S, Thorpe C, Cala S, Davis P. Respiratoryfunction in operatic singing: effects of emotional con-nection. Logoped Phoniatr Vocol 2000; 25: 151�/68.
34. Wapnick J, Ekholm E. Expert consensus in solo voiceperformance evaluation. J Voice 1997; 11: 429�/36.
35. Cabrera D, Davis P, Barnes J, Jacobs M, Bell D.Recording the Operatic Voice for Acoustic Analysis.Acoustics Australia 2002; 30: 103�/8.
36. Rossing TD, Sundberg J, Ternstrom S. Acoustic com-parison of soprano solo and choir singing. J Acoust SocAm 1987; 82: 830�/6.
37. Ekholm E, Papagiannis GC, Chagnon FP. Relatingobjective measurements to expert evaluation of voicequality in Western classical singing: critical perceptualparameters. J Voice 1998; 12: 182�/96.
38. Robison CW, Bounous B, Bailey R. Vocal Beauty: Astudy proposing its acoustical definition and relevantcauses in classical baritones and female belt singers.Journal of Singing 1994; 51: 19�/30.
39. Vurma A, Ross Y. Priorities in voice training: carryingpower or tone quality. Musicae Scientiae 2000; 4: 75�/93.
40. Callaghan J. Singing and voice science. San Diego,California: Singular Publishing. Group; 2000.
41. Gabrielsson A, Juslin PN. Emotional Expression inMusic Performance: Between the Performer’s Intentionand the Listener’s Experience. Psychology of Music1996; 24: 68�/91.
42. Sundberg J, Iwarsson J, Hagegard H. A Singer’sExpression of Emotions in Sung Performance. In:Fujimura O, Hirano M, editors. Vocal Fold Physiology:Voice Quality Control. San Diego, California: SingularPublishing Group, Inc.; 1995. p. 217�/31.
43. Easley E. A Comparison of the Vibrato in Concert andOpera Singing. In: Seashore CE, editor. The Vibrato.Iowa: University of Iowa Press; 1932. p. 269�/75.
44. Hakes J, Shipp T, Doherty ET. Acoustic Properties ofStraight Tone, Vibrato, Trill and Trillo. J Voice 1987; 1:148�/56.
45. Titze IR, Long R, Shirley GI, Stathopoulos E, RamigLO, Carroll LM, et al. Messa di voce: an investigation ofthe symmetry of crescendo and decrescendo in a singingexercise. J Acoust Soc Am 1999; 105: 2933�/40.
46. Titze IR, Solomon NP, Luschei ES, Hirano M. Inter-ference between normal vibrato and artificial stimula-tion of laryngeal muscles at near-vibrato rates. J Voice1994; 8: 215�/23.
47. Titze IR, Story B, Smith M, Long R. A reflex resonancemodel of vocal vibrato. J Acoust Soc Am 2002; 111(5 Pt1): 2272�/82.
48. Davidson JW, Da Costa Coimbra D. Investigatingperformance evaluation by assessors of singers in amusic college setting. Musicae Scientiae 2001; 5: 33�/53.
49. Stanley M, Brooker R, Gilbert R. Examiner Perceptionsof Using Criteria in Music Performance Assessment.Research Studies in Music Education 2002; 18: 43�/52.
50. Rothman HB, Rullman FJ, Arroyo AA. Inter- andIntrasubject Changes in Vibrato: Perceptual and Acous-tic Aspects. J Voice 1990; 4: 309�/16.
182 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2965
3.1 Conclusion
Vibrato is an accepted tone quality of the classical singing voice (Miller, 1996;
Vennard, 1968). It is widely accepted in the singing literature that poor vibrato is
indicative of poor technique (Miller, 1996; Vennard, 1968) and inferior sound quality or
timbre. Previous studies have not considered the application of vocal techniques to
refine vocal features such as vibrato in singing. These results indicate that for these
singers, open throat was an integral technical component of their production of a good
sound.
From the results of the vibrato study, it was hypothesized that the energy distribution in
these singers’ voices in each condition would also reflect the timbral changes associated
with open throat. Long-term average spectra have been used to represent vocal genres
and also vocal qualities, in particular, the carrying power identified in operatic and
classical voices. This methodology was used to further assess the differences in quality
achieved by open throat.
66
4 Long term average spectra
In singing literature, particular LTAS models or exemplars have become synonymous
with vocal qualities and voices of quality. Singing analyses present exemplars of voice
types (Sundberg, 1974) or singing genres (Borch & Sundberg, 2002; Cleveland et al.,
2001). LTAS curves have been used to make obvious differentiations between males
and females (Mendoza, Valencia, Munoz, & Trujillo, 1996), singers and speakers
(Barrichelo, Heuer, Dean, & Sataloff, 2001), solo voice and choral voice (Rossing et al.,
1987) and pop or country from opera singers (Borch & Sundberg, 2002; Cleveland et
al., 2001). Research designs to compare multiple spectra indicate that it is important that
only one factor be varied in any given comparison (Barnes et al., 2004; Rossing et al.,
1987; Thorpe et al., 2001). Visual inspection of LTAS in each case confirmed
differences between pairs of samples.
Early studies using LTAS present visual data for analyses (Jansson & Sundberg, 1975;
Rossing et al., 1987) and later studies seek to quantify the data contained in the LTAS
to a single meaningful number (Barnes et al., 2004; Novak & Vokral, 1995; Omori et
al., 1996; Thorpe et al., 2001). These have been used to predict quality in voice
professionals (Novak & Vokral, 1995), to compare voices and also as a measure of
goodness or a voice of quality by level of professional singing (Barnes et al., 2004;
Thorpe et al., 2001). While exemplars illustrate the most common or expected features
by singer or singing genre, they carry little statistical power in the identification or
report of vocal quality.
67
More recent research linked features of these exemplars with performers of the highest
regard in singing performing circles (Barnes et al., 2004; Thorpe et al., 2001). LTAS
evaluation methods have followed these trends identified through visual inspection of
singers’ spectra and assess vocal quality through measures of central tendency or
variability between singers (Novak & Vokral, 1995; Sundberg, 2001).
Current practice in singing research accepts LTAS as representative of timbre. The
analysis has been used to track gross differences between singers and speakers
(Barrichelo et al., 2001) and inherent similarities between stylised country singing and
speech (Cleveland et al., 2001). It follows that LTAS should identify timbral differences
in the same group of subjects performing under different conditions.
LTAS are used to assess the acoustic output and timbre in extended singing. In this next
paper, we assessed the impact of open throat on visual displays of LTAS as well as on
conventional analysis methods of singing power ratio (SPR) (Omori et al., 1996) and
energy ratio (ER) (Thorpe et al., 2001). We measured the spectral distribution of
acoustic energy in the same group of singers in their singing of a classical song and a
romantic lied. Our aim was to confirm timbral differences identified from changes to
vibrato from O to SO.
68
PAPER 3
Mitchell, H. F., & Kenny, D. T. (2004). The effects of open throat technique on long term average spectra (LTAS) of female classical
voices. Logopedics Phoniatrics Vocology, 29(4), 171-182.
69
The effects of open throat technique on long termaverage spectra (LTAS) of female classical voices
Helen F. Mitchell and Dianna T. Kenny
From the Australian Centre for Applied Research in Music Performance (ACARMP), Sydney Conservatorium of Music,The University of Sydney, New South Wales 2006, Australia
Received 8 October 2003. Accepted 13 May 2004.
Logoped Phoniatr Vocol 2004; 29: 99�/118
In the third of a series of studies on open throat technique, we compared long term average spectra (LTAS) of six advancedsinging students under three conditions: ‘optimal’ (O), representing maximal open throat, ‘sub-optimal’ (SO), using reducedopen throat, and loud sub-optimal (LSO) to control for the effect of loudness. Using a series of univariate repeated measuresANOVAs with planned orthogonal contrasts, we tested the hypotheses that sound pressure level (SPL) and the ratio of spectralenergy in peaks and areas between 0�/2 kHz and 2�/4 kHz would be reduced in SO and LSO compared to O. There weresignificant differences between SO and LSO but hypotheses were not confirmed for O. These findings do not accord withdifferences in vibrato extent and onset between O and SO/LSO (Mitchell and Kenny, in press). These results suggest that whileLTAS provides information on energy distribution, measuring spectral energy areas appears to be the most sensitive measureof energy distribution between conditions. Plotting the differences between O and SO/LSO pairs of LTAS clearly indicates theareas of spectral change. The findings from this study also indicate that LTAS are not sufficiently sensitive to measure vocaltimbre as they were not consistent with perceptual or other acoustic studies of the same samples.
Key words: long-term average spectra, open throat, singing technique, vocal pedagogy.
Helen Mitchell, Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music, TheUniversity of Sydney, New South Wales, Australia 2006. Tel.: 61 2 9351 9644. Fax: 61 2 9351 9540. E-mail: [email protected]
INTRODUCTION
In the singing literature and current pedagogy, open
throat technique is considered important to the
achievement of a good sound, in particular, as a
necessary component of resonance (1�/3). In a recent
study, pedagogues agreed that the technique of open
throat produces an audible quality in the sound,
equating its use with an even and consistent, balanced
and coordinated, free, warm and open quality in the
overall sound (4). They described the technique as a
way of maximizing pharyngeal space and/or achieving
abduction of the ventricular folds that could be
achieved by laugh, sob, correct inhalation or main-
taining the posture of inhalation (4).
Despite the routine use of ‘open throat’ technique in
the singing studio, there has to date been only one
scientific study that has addressed the vocal quality
believed to be produced by this technique (5). Hy-
potheses generated from pedagogical beliefs about
open throat were tested acoustically by Mitchell and
Kenny (5). In this study, we compared vibrato rate,
extent and onset of six advanced singing students
under three conditions: ‘optimal’ (O), representing
maximal open throat, ‘sub-optimal’ (SO), using re-
duced open throat, and loud sub-optimal (LSO), using
reduced open throat but controlling for the effect of
loudness. Fifteen expert judges correctly identified the
sound produced when singers used open throat with
87% accuracy. Having verified the technique percep-
tually, we used a series of univariate repeated measures
ANOVAs with planned orthogonal contrasts to test
the hypotheses that frequency modulations associated
with vibrato rate, extent and onset would vary outside
acceptable or desirable parameters for SO and LSO.
Hypotheses were confirmed for vibrato extent and
onset but not for rate. There were no significant
differences between SO and LSO on any of the vibrato
parameters. As vibrato is considered a key indicator of
good singing, these findings suggested that open
throat is important to the production of a good sound
in classical singing.
In this study, we examined the spectral distribution
of acoustic energy in the same six singers’ voices in
�ORIGINAL ARTICLE �
# 2004 Taylor & Francis. ISSN 1401-5439 Logoped Phoniatr Vocol 29
DOI: 10.1080/14015430410015722 70
each condition to identify the timbral changes in SO
and LSO. Positive descriptions of spectral energy such
as ‘resonance/ring’ (6), the singer’s formant (7) have
been linked to production of a beautiful sound and
with an efficient and aesthetically pleasing distribution
of energy or harmonics and loudness in classical
singing (4).
Long-term average spectra (LTAS) have been used
to analyse musical timbre (8) and also vocal features,
in both speech (8, 9) and singing (10, 11). An LTAS
gives an overall impression of an entire excerpt (12),
identifies certain consistent features contained in the
sound over time, and averages out short-term varia-
tions in phonetic structure (13). After 20�/30 seconds,
the LTAS represents the individual’s sound spectra
(12, 14). The LTAS, when taken over a sufficiently
long production, suppresses the short-term variations
that are due to notes and lyrics, while retaining such
long-term invariance in the spectrum as can be
expected to reflect a particular phonatory and/or
articulatory strategy. The LTAS plot is a measure of
the dB level of the time-average of the power of the
acoustic signal at each frequency.Rossing et al. (15) used LTAS to track differences in
the same singers and found that soprano singers’
LTAS plots were different depending on whether the
singing was solo or choral, and soft or loud. Jansson
and Sundberg (12) noted that the nature of the musical
task, the key, the instrument on which it was
performed, and timbral differences on the same
instrument all influenced the LTAS plot. These data
indicate that in research designs comparing LTAS, it is
important that only one factor be varied in any given
comparison. This might involve the same singer sing-
ing and speaking the same task (11), singing the same
task in different styles (15), different singers singing
the same task (16) or instruments playing in different
musical keys or producing the sound in different ways
or with different musical expression (12).
Long-term average spectra, particularly in singing,
have been accepted as analogues for vocal quality.
Vocal projection has also been conceptualized in this
way. Vocal projection and the ability to be audible
above an orchestra are vital to the classical singing
voice (17, 18). Classical and operatic singers demon-
strate a characteristic spectral energy above 2 kHz, to
enable amplification over an orchestra, which does not
show amplification in this frequency band (12, 16, 19).
This carrying power was identified by Bartholomew
(20) as a spectral reinforcement around 2800 Hz,
produced by male and low female singers. Sundberg
classified this as the ‘singer’s formant’ (21). The
existence of this energy band has been widely debated
as an element in the soprano voice (20, 22�/24). Some
studies have identified greater energy in sopranos of
international standard than other classically trained
sopranos (15, 23) and concluded that energy above 2
kHz may be as important for sopranos as for other
voice types. However, in some cases, an increased
energy above 2 kHz was perceived as detrimental to
vocal quality when assessed perceptually (18). It is
therefore important to consider carrying power as
energy distribution and not simply as an indicator of
quality.
The typical method of assessing vocal quality
through LTAS has been through an investigation of
differences in measures of central tendency and
variability between singers (10, 14). There is a need
to explore intra-singer differences under different
performance requirements or conditions because in-
dividual spectra may be unique. Averaged data may
produce curves that do not adequately represent the
unique qualities of individual singing voices and may
indeed produce a curve that does not represent any
voice (11, 14, 16, 25). The individuality of spectra is
evident when the same singer singing portions of the
same aria produces different spectral plots and both
plots are considered in making a mean trend of
spectral changes (25). Studies using the same singers
but different musical tasks (e.g. country singers singing
an anthem and a song (11)) show marked differences
in spectra between the two conditions. This demon-
strates the importance of using the same musical
sample or stimuli when examining LTAS derived
under different experimental conditions.
A conventional way of reducing the information in
the LTAS to a single meaningful number is to compute
the ratio of energies in a low and a high frequency
band. Often 2 kHz is chosen as the delimiting
frequency (25, 26). In LTAS, particularly in clinical
practice, the slope between these peaks is calculated
(13). In singing, measures of spectra compare energy
peak height (singing power ratio) and peak area
(energy ratio) between 0�/2 kHz and 2�/4 kHz. The
difference between the height of the major peaks
between 0�/2 kHz and 2�/4 kHz (26) quantifies the
relative energy between 2�/4 kHz although it does not
account for the shape of these energy peaks. Thorpe et
al. assessed the area under the LTAS curve at 0�/2 kHz
and 2�/4 kHz, rather than the highest peaks of energy,
to identify the areas in which the energy is reinforced
or reduced when singers sang with varying levels of
emotional connection (25). It also enables effective
comparison between individual singers singing in
different conditions and between groups of singers.
Comparing pairs of LTAS, where the data have been
normalized to remove the sound pressure level (SPL)
variables (11), produced little difference between
LTAS of country singers’ speech and singing.
100 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2971
In this study, we assessed whether open throat
technique is a necessary component of good classical
vocal quality and measured the spectral distribution ofacoustic energy with the same group of singers who
consciously manipulated the technique in their singing
of a classical song and a romantic lied. Our aim was to
use LTAS to discern measurable differences in spectral
energy between an optimal sound quality, using
maximal open throat, and a sound which used a lesser
degree of open throat, produced without consciously
altering other components of their singing techniqueor sound quality. This approach represents a major
extension on current work that describes vocal fea-
tures at only one point in time.
METHOD
Participants
Six female singers (three sopranos, three mezzo-
sopranos) volunteered to participate in this project.They were advanced students with excellent technique
of an experienced singing pedagogue, who was a
Lecturer in Vocal Studies and Opera at a state
Conservatorium of Music. Criteria for participant
selection included, through this pedagogue’s assess-
ment, singers who: 1) had a good classical singing
technique for their level of training and experience;
and 2) understood and demonstrated skilful control of‘open throat’ or ‘retraction’ techniques in their sing-
ing. The institutional human ethics committee ap-
proved the study.
Prior to the voice recording, participants completed
a questionnaire seeking information on age, years of
singing study, number of years of study with each
singing teacher, highest qualifications attained or
currently undertaken in music and/or singing, andsinger type (soprano or mezzo soprano). The partici-
pants were also asked to classify the genres of singing
they performed in public (opera, classical, choral,
music theatre and contemporary) and to estimate the
percentage each style played in their total performing
career.
The demographic information of the participants is
presented in Table 1. Participants were aged between23 and 30 years, with a mean of 26 years. All had
studied singing for at least 7 years (average 9.8 years)
and had spent an average of 5 years studying with
their present singing teacher. Each singer held a
qualification in singing or music (four had Bachelor
of Music degrees and two had diplomas, in music and/
or singing) and five of six were currently undertaking a
second degree in singing (three postgraduate Diplomaof Opera and two Bachelor of Music degrees).
All defined the majority of their singing as operatic
(�/50%), with the second most common style classical
(�/20%), in accordance with the Bunch and Chapman
(27) taxonomy of singing voices. All reported that they
were in good health and able to perform the tasks.
Procedure
Singers were sent information about the project and
were invited to take part in an acoustic and perceptual
study of singing technique. They were required to
attend a single recording session lasting no more than
an hour and were told that the object of the study was
to investigate acoustical and perceptual features of the
use of open throat in singing and to discover the sound
qualities associated when a singer uses some form of
open throat technique compared with when a singer
does not use open throat.
Instruments
Protocol. A protocol was developed to assess the
effect of ‘open throat’ technique in singing. Two
musical tasks were developed to investigate the use
of the technique in two song excerpts. Musical tasks
were chosen in order to test different demands of good
singing, but were not musically difficult. They were
specifically designed to test the use of open throat, and
contained musical features derived from a previous
qualitative study on the use of the technique (4), where
use or lack of the technique was deemed to be
particularly valuable or noticeable. These features
were: high tessitura, sustained or legato singing,
dynamic range control, and vocal agility. Each musical
task took an average of 74 seconds to perform.
The Mozart song Ridente la Calma , K 152, bars
1�/27 (Figure 1a) was selected as it is a nominally
simple song in the Italian language (28) with a mixture
of common musical statements involving repeated
legato lines as well as the initial stylized leaps of a
major 4, and short scale figures. All six singers sang
this aria in the same key (F major).
The third verse of the Schubert lied, Du bist die Ruh
D. 776 (Op. 59, No. 3), bars 54�/80 (Figure 1b) was
chosen for its demanding vocal control, sustained
musical line and high climactic tessitura. The three
sopranos and three mezzo-sopranos sang this in an
appropriate key depending on soprano or mezzo-
soprano voices (E-flat, D-flat and C major).
Prior to singing, each singer selected the sequence of
their tasks before commencing the experiment by
selecting a blank card, the reverse side of which
represented one of the tasks, to reduce the possibility
that task order may influence the results.
Effects of open throat technique on LTAS 101
Logoped Phoniatr Vocol 2972
Recording
Participants were given time to warm up in the singing
studio and become familiar with the room before
recording. Recording levels for each singer were set
during this time. The acoustic signal was recorded
digitally (Behringer Ultragain preamplifier/Marantz
CDR 630) to CD via a high-quality microphone
(AKG C-477) positioned on a head boom a constant
7-cm distance from the singer’s lips. This ensured that
direct energy of the performers’ voices was recorded
rather than room reflections, enabling us to use a
studio environment with low ambient noise rather
than an anechoic studio (29).Calibration was carried out in each recording by
using a recording of two pink noise samples on
audiotape played immediately following each record-
ing session at the same recording gain used for
recording the singer’s voice. Pink noise is a broadband
steady state signal of known sound pressure level. Pink
noise enables calibration of an audio playback system
across the frequency range for tasks such as subjective
testing and if the analysis is just done on computer, the
Tab
le1.
Su
mm
ary
of
pa
rtic
ipa
nt
cha
ract
eris
tics
.D
escr
ipti
on
of
sin
ger
nu
mb
er,
age,
the
nu
mb
ero
fyea
rsea
chsi
nger
ha
sta
ken
sin
gin
gle
sso
ns,
the
nu
mb
ero
f
tea
cher
sea
chsi
nger
ha
sh
ad
,th
en
um
ber
of
yea
rsea
chsi
nger
stu
die
dw
ith
thei
rth
ree
mo
stre
cen
tte
ach
ers
(T
1� /T
3)
,th
eh
igh
est
leve
lo
fed
uca
tio
nco
mp
lete
dby
each
sin
ger
,th
ecu
rren
tleve
lo
fed
uca
tio
nu
nd
erta
ken
,vo
ice
typ
e,p
rop
ort
ion
of
each
sin
ger
’sca
reer
spen
tsi
ng
ing
op
era
,cl
ass
ica
l,ch
ora
l,m
usi
cth
eatr
e,
con
tem
po
rary
or
oth
erst
yle
so
fm
usi
c
Sin
ger
Ag
e(y
ears
)Y
ears
of
less
on
sN
o.
of
tea
cher
sT
1T
2T
3H
igh
est
edu
cati
on
Cu
rren
ted
uca
tio
nV
oic
ety
pe
Op
era
Cla
ssic
al
Ch
ora
lM
usi
cth
eatr
eC
on
tem
pO
ther
12
68
41
13
Ba
chel
or
Ba
chel
or
Mez
zo8
0%
20
%2
25
73
1.5
34
.5B
ach
elo
rP
ost
gra
du
ate
So
pra
no
80
%2
0%
33
01
22
10
2D
iplo
ma
N/A
So
pra
no
97
%3
%4
23
10
24
5.5
Ba
chel
or
Po
stg
rad
uat
eS
op
ran
o5
0%
45
%5
%5
25
10
27
.52
.5D
iplo
ma
Po
stg
rad
uat
eM
ezzo
73
%2
5%
2%
62
81
26
51
.51
.5B
ach
elo
rB
ach
elo
rM
ezzo
50
%3
5%
5%
10
%
Fig. 1a. The Mozart task musical score.
Fig. 1b. The Schubert task musical score in the
soprano key.
102 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2973
noise is at least as good as any other steady-state
known signal. In addition, a tone has a disadvantage
over noise by being more susceptible to interference in
the environment, and hence would give less repeatable
results. For calibration of absolute sound pressure
levels (SPL), the microphone of a sound level meter
(Rion NL-06 SPL) was held adjacent to the AKG
microphone from a speaker (Bose Lifestyle) from
which the pink noise was played. The SPL (in dB at
7 cm) recorded by the sound level meter was noted for
the two noise signals and used later for calibration.
Experimental conditions
Singers sang each of the two song excerpts under three
conditions: optimal, sub-optimal and loud sub-opti-
mal. ‘Optimal’ (O) condition was necessary to provide
a sound quality with the best technique that they
could. This involved the maximal use of open throat.
‘Sub-optimal’ (SO) condition involved the use of a
reduced (open throat) technique but still with an
acceptable singing technique and without consciously
altering any other aspect of their technique. It was
hypothesized that the SO condition would result in a
reduction of SPL, so a third condition ‘loud sub-
optimal’ (LSO) involved the same instruction as the
SO condition, but with the added instruction that the
singer should try to achieve a louder dynamic than in
SO. This addressed the possibility of SPL as the only
variable responsible for the differences between SO
and O. This was similar to technical conditions
established by Foulds-Elliot et al. (30) when SPL
was a variable of degrees of emotional connection in
operatic singers.
Each task was performed twice in the O and SO
conditions while the LSO condition was performed
only once because of the slight possibility of adverse
effects on vocal health. In total, each singer performed
each musical task five times.
Instructions
A pedagogue was present during the recording ses-
sions to provide accompaniment for warm-ups and
practice of the tasks where necessary and also selected
the terminology and instructions necessary for each
singer to achieve the experimental conditions. For
example, she instructed them to pay attention to
producing the most open sound in their throat in the
O condition, and a lesser degree in the SO and LSO
conditions. Some singers asked how to produce the
LSO condition and the pedagogue instructed them to
‘use more twang’, as taught in their lessons (31).
Analysis
The audio recordings were digitized at 16 kHz using
Phog Version 2.0 (Hitech Development, Sweden)
software and reopened in Soundswell Version 4.0
(Hitech Development, Sweden) to produce nine chan-
nels of raw and calculated data (including SPL and
F0). A real-time digital SPL meter was used to set a
recording gain so that the pink noise calibration level
recorded in Phog was the same (�/0.5 dB) as the SPLlevel noted during recording.
Long-term average spectra
Long-term average spectra (LTAS) analyses (band-
width 300 Hz) (32, 33) of the 36 song samples
demonstrated the differences in spectral energy for
the singers, conditions and musical tasks. LTAS
contained only voiced data (13, 32). A 300 Hz LTAS
bandwidth is less sensitive than a narrow-band analy-sis to movement in the partials when the singer is at
higher fundamental frequencies. A complication when
using LTAS is to account for loudness, which is
problematic as frequencies above 2 kHz increase faster
than those below 2 kHz as sound pressure level (SPL)
increases (34).
In order to relate these LTAS plots to the known dB
of the calibration tone, an LTAS was performed onsteady five-second portions of the pink noise sample
for each singer, using the same 300 Hz filter. The Sect
tool in Soundswell does not compute the overall
equivalent level (Leq) for an LTAS. Therefore, to
calculate the absolute SPL of each calibration tone, it
was necessary to find the mean SPL of the pink noise
LTAS. Each point of the LTAS curve was linearized
(y�/10 (x/10)), then all data points were summed, andthis total converted back to dB (x�/10 �/log(y)). The
mean SPL of pink noise (in decibels down from full
scale) was subtracted from the known SPL of the
calibration noise measured during recording to pro-
duce the calibration offset. This calibration offset (in
dB) was applied to each LTAS. In earlier research
using LTAS methodology, White (32) arrived at a
‘calibration offset’ by calculating the mean SPL of asung /i/ vowel by adding the peak level (dB) of the
LTAS with the known level. Using LTAS area to
calculate the mean SPL of pink noise was used in
preference to calculating the LTAS peak height and
was considered a valid method for evaluating the
overall level (dB) of the pink noise at 7 cm.
LTAS analysis
Acoustic analysis focused on two frequency areas: 0�/2kHz and 2�/4 kHz. From the LTAS plots, the highest
peaks in the 0�/2 kHz and 2�/4 kHz regions were
Effects of open throat technique on LTAS 103
Logoped Phoniatr Vocol 2974
labelled P1 and P2 respectively. Peak levels (in dB) and
peak centre frequencies (in Hz) were calculated for
each singer in each task and condition.Three measures were used to quantify these LTAS
data. Linear regression was used to quantify the
relationship between the shapes of each singer’s
LTAS by condition (O, SO, LSO), in each musical
task. In the regression coefficients, a slope of 1 and an
intercept of 0 would imply a perfect correlation, thus
deviations demonstrate differences observed between
condition results. This methodology draws compar-isons between the changes in the same task with
different overall qualities (11, 16).
The singing power ratio (SPR), described by Omori
et al. (26) compares the peak levels of P1 and P2. The
SPR is the difference between the level of P1 and P2 in
dB (LP1�/LP2) and is a measure of spectral slope
between 0�/2 kHz and 2�/4 kHz. When the energy is
focused in the 0�/2 kHz region, it results in higher SPRresults and there is typically less energy reinforcement
in the 2�/4 kHz region. A low SPR indicates a stronger
energy peak �/2 kHz. Although SPR does not prove
the presence of a singer’s formant, it enables inter- and
intra-singer comparison of spectral energy in the voice
(35, 36).
The energy ratio (ER), described by Thorpe et al.
(25) compares overall energy between the area between0�/2 kHz (A1) and the area between 2�/4 kHz (A2). It is
calculated by taking the difference between average
energy values for the two frequency areas (A1�/A2). A
low ER represents a greater reinforcement in the 2�/4
kHz region whereas a high ER represents a smaller
energy boost �/2 kHz. Areas below the LTAS curve
were calculated in the same way as the calibration
tone. When there is less reinforcement in the 2�/4 kHzregion, ER results follow SPR. As ER assesses the
area of the LTAS curve, the energy differences between
singers’ O/SO, O/LSO and SO/LSO LTAS data were
plotted graphically to compare the energy distribution
at each frequency point.
Other factors that may influence the LTAS plot,
such as fundamental frequency (F0) distribution, and
the pitch range of each musical task was considered(12). Histograms of the F0 were computed in Sounds-
well for each musical task.
Study design
The study design was a repeated measures (three
dependent measures: SPL; P1; P2) randomized com-
plete block with 2 (composer (Mozart versus
Schubert))�/3 (condition (optimal, sub-optimal and
loud sub-optimal)) factorial structure with plannedorthogonal contrasts on the three dependent variables.
This design was chosen because the study hypotheses
were mainly concerned with the effects of experimental
condition on SPL, P1 and P2 and not between-singer
variability. The chosen design absorbs singer varia-
bility within the error term to reveal the main effects
and interactions of interest.
Data were analysed firstly through a series of fixed
and mixed effect univariate ANOVAs with three
factors (task, condition, singer). In the first series, all
three factors were entered as fixed effects; in the
second, singer was entered as a random effect. Finally,
the data were subjected to analysis using the general
linear model (GLM) with planned orthogonal con-
trasts. Since all analyses yielded the same outcomes,
only the results of the GLM are presented here. Main
effects for task and condition and interaction effects
(contrasts) were calculated for each dependent mea-
sure (SPL, P1, P2). In the first contrast, O was
compared to SO and LSO and in the second contrast
SO was compared to LSO.
Hypotheses
Hypotheses were generated regarding the spectral
distribution of acoustic energy and sound pressure
levels produced by singers using open throat compared
to a reduced open throat condition. The loud sub-
optimal (LSO) condition was used as a control for
loudness between O and SO. Therefore we compared
O with SO/LSO. Our specific hypotheses were as
follows:
1. Overall sound pressure level (SPL) will decrease
from O to SO and from O to LSO. Expert
pedagogues stated that the use of open throat
would increase their perception of ‘loudness’ (4).2. Following from hypothesis 1, long-term average
spectral distribution, that is, peak levels P1 and P2
(in dB), will decrease in SO/LSO. Voice acoustics
predict that the magnitude of the level difference
between P1 and P2 will become smaller with
increasing SPL, except at extremes of effort and
vowel but we wanted to test whether this relation-
ship also pertains during the use of a perceptuallydistinct singing technique (4, 5).
3. There will be a difference in energy distribution
shown in the shape of the LTAS curve between O
and SO/LSO, based on expert pedagogical de-
scriptions and perceptual judgements (4, 5).
4. SPR measurements will increase in SO/LSO
compared to O. Lower SPR is associated with
higher vocal quality and greater length of training.The pedagogical literature (1�/3, 37) and our
previous study (4) have indicated that length of
training and vocal quality are also associated with
104 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2975
open throat. We therefore test the hypothesis that
open throat will be associated with lower SPR.
5. ER measurements will increase in SO/LSO com-pared to O. In singing, ER results tend to follow
SPR. That is, lower SPR and ER are associated
with higher vocal quality (36).
6. Changes observed in each condition will be
consistent with reductions in vibrato extent and
increases in vibrato onset in SO/LSO compared to
O (5). Given the statistical confirmation that SO/
LSO were different to O (5), we investigate howthese findings affect the LTAS energy distribution.
RESULTS
Descriptives
Figure 2a and b present the LTAS of each singer in
each condition for the Mozart and Schubert tasks
respectively. The complete SPL and spectral peak data
are presented in Table 2, showing the overall means of
SPL and levels of P1 (0�/2 kHz) and P2 (2�/4 kHz)
with corresponding centre frequency (Hz) by eachsinger, in each condition and in each musical task
(Mozart and Schubert).
Differences in SPL, P1 and P2 were considered
separately for effects of condition and task.
Hypothesis 1: overall sound pressure level (SPL) will
decrease from O to SO and from O to LSO
Hypothesis 1 stated that SPL will be lower in the SO
condition than in O and from O to LSO. Hypothesis 1
was confirmed between SO and LSO. For SPL, there
was a main effect for contrast 2 only (F(1,5)�/18.948,
p�/0.007), that is, SPL was significantly higher in theLSO condition than SO. Overall means indicated no
significant difference between SPL of O and SO/LSO
(F(1,5)�/0.010, p�/0.924). There was no effect of task
(F(1,5)�/0.001, p�/0.974), and SPL was affected by
condition in the same way in each task. Figure 3
presents these data graphically.
The mean SPL for the O condition over the six
singers was 97.4 dB (SD 1.9) in the Mozart task and97.7 dB (SD 2.6) in the Schubert task. In the SO
condition, the mean SPL was significantly lower,
particularly in the Mozart task. In both the SO and
LSO condition, each singer’s SPL was significantly
different to O (Table 2). The LSO condition was
associated with SPL levels that were either similar to
the O condition (six examplesB/2 dB different) and in
nine examples, increased SPL from the level of theoriginal O in the LSO condition (LSO means: 99.2 dB,
98.9 dB).
Hypothesis 2: peak levels P1 and P2 (in dB) will
decrease in SO/LSO
Hypothesis 2 was confirmed for P1. For P1, there was
a main effect for contrast 1 only (F(1,5)�/17.359,
p�/0.009), that is, the level of P1 was significantly
greater in the O condition than SO/LSO. There was no
significant difference between SO and LSO (F(1,5)�/
4.911, p�/0.078). There was no effect of task (F(1,5)�/
1.392, p�/0.291). There were no interactions between
musical task and condition. These data are presented
in Fig. 4. In the SO condition P1 was associated with a
considerable decrease (�/3 dB) in 9 of the 12 (6
singers�/2 musical tasks) SO cases compared to the
O condition. In the LSO condition P1 was reduced
�/3 dB in half of 12 LSO cases, but was greater in one
take compared to the O condition.
Hypothesis 2 was not confirmed for P2. For P2,
there was a main effect for contrast 2 only (F(1,5)�/
15.330, p�/0.011) that is, levels of P2 were significantly
higher in LSO than in SO but SO/LSO were not
significantly different to O (F(1,5)�/1.112, p�/0.340).
Fig. 5 presents these data. There was no effect of
musical task (F(1,5)�/1.643, p�/0.256) and there were
no interactions between task and condition for P2.
Changes in the P2 by condition were more difficult to
quantify. In the SO condition, P2 was associated with
a considerable decrease (�/3 dB) in 7 of the 12 SO
cases compared to the O condition. Some were
markedly reduced, such as 8�/10 or 20 dB lower (Table
2). One SO sample was associated with a greater SPL
(�/3 dB) compared to the O condition.
Hypothesis 3: there will be a difference in energy
distribution shown in the shape of the LTAS curve
between O and SO/LSO
Hypothesis 3 stated that there will be a difference in
energy distribution shown in the shape of the LTAS
curve between O and SO/LSO, based on expert
pedagogical descriptions and perceptual judgements.
This hypothesis was tested using linear regression
where all singers’ LTAS data were normalized, with
P1 at 0 dB and the LTAS distribution compared
between O and SO/LSO and between SO and LSO in
each musical task, for each singer (after Cleveland
et al . (11)). Hypothesis 3 was not confirmed. Correla-
tions between individual singers’ LTAS across the
three conditions are presented in Table 3. All correla-
tions were greater than 0.90. Singer 5 was an outlier
between O/LSO. Slopes were all close to 1 and
approaching unity, and intercepts vary from close to
zero in the Schubert task, to �/3.92�/�/3.05 in the
Mozart.
Effects of open throat technique on LTAS 105
Logoped Phoniatr Vocol 2976
Fig. 2a. The Mozart task: LTAS for the six singers in the Mozart musical task, showing mean sound pressure
levels (dB at 7 cm) on the ordinate at frequencies from 0�/6000Hz on the abscissa. The bold, thin and dashed lines
show optimal, sub-optimal and loud sub-optimal conditions respectively. Each LTAS is marked by subject
number 1�/6.
106 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2977
Hypothesis 4: SPR measurements will increase in SO/
LSO compared to O
Hypothesis 4 stated that SPR will increase in SO/LSO
compared to O, that is, the difference between peak
levels between 0�/2 kHz and 2�/4 kHz would indicate
differences in carrying power (energy in 2�/4 kHz
region). Hypothesis 4 was not confirmed for O and
SO/LSO. In testing the differences (in dB) of SPR, the
effect of task approached significance (F(1,5)�/8.928,
p�/0.031), that is, the difference between P1 and P2 in
the Schubert task was slightly greater than in Mozart,
but both musical tasks showed the same overall
pattern across conditions. There was an effect for
contrast 2 only, (F(1,5)�/45.152, p�/0.001), that is, the
Fig. 2b. The Schubert task.
Effects of open throat technique on LTAS 107
Logoped Phoniatr Vocol 2978
Table 2. Measurements of sound pressure level (SPL) at 7 cm, peak levels in dB (P1, P2) from the 0�/2 and 2�/4 kHz regions of a long term average spectra and
centre frequencies in Hz for P1 and P2 for the Mozart and Schubert musical tasks in each condition, optimal (O), sub-optimal (SO) and loud sub-optimal
(LSO)
SPL P1 P2
Condition Condition Condition
O SO LSO O SO LSO O SO LSO
Task Singer dB (Hz) Database (Hz) dB (Hz) dB (Hz) dB (Hz) dB (Hz)
Mozart1 99.3 97.0 98.3 97.4 (750) 90.9 (750) 91.6 (750) 77.2 (3125) 76.0 (3000) 79.0 (3000)2 95.3 92.8 97.1 98.2 (750) 90.6 (750) 97.6 (750) 73.7 (3125) 65.9 (3000) 77.3 (3000)3 98.6 97.6 98.6 98.5 (750) 92.2 (750) 95.3 (750) 84.4 (3000) 74.8 (3125) 80.2 (3125)4 97.9 98.0 98.4 96.8 (750) 93.8 (750) 92.5 (750) 74.8 (3125) 71.8 (3125) 73.5 (3125)5 98.3 95.6 102.8 99.4 (750) 91.5 (750) 94.8 (750) 79.2 (3000) 73.3 (3000) 80.4 (3000)6 94.8 96.0 99.9 94.9 (750) 94.2 (750) 98.6 (750) 72.7 (3500) 78.3 (3125) 83.7 (3250)Mean (SD) 97.4 (1.9) 96.2 (1.9) 99.2 (2.0) 97.5 (1.6) 92.2 (1.5) 95.1 (2.8) 77.0 (4.3) 73.4 (4.3) 79.0 (3.4)
Schubert1 98.0 96.0 98.7 93.6 (625) 89.3 (625) 90.9 (625) 75.9 (3125) 69.9 (2875) 76.6 (2875)2 98.4 90.9 96.3 103.1 (625) 91.0 (750) 97.0 (750) 78.3 (3125) 58.3 (3125) 72.9 (3250)3 100.6 99.1 100.9 100.8 (750) 96.9 (750) 98.7 (750) 79.7 (3000) 73.7 (3000) 80.6 (3125)4 99.7 99.4 101.4 99.6 (750) 99.3 (750) 99.6 (750) 74.9 (3250) 72.6 (3250) 75.1 (3375)5 93.7 94.7 96.9 97.2 (500) 92.3 (625) 91.6 (625) 76.0 (3375) 76.4 (3000) 77.9 (3375)6 95.5 97.0 98.8 99.1 (625) 96.8 (625) 96.1 (625) 78.9 (3500) 77.7 (3375) 82.8 (3375)Mean (SD) 97.6 (2.6) 96.2 (3.1) 98.8 (2.1) 98.9 (3.3) 94.3 (3.9) 95.7 (3.7) 77.3 (1.9) 71.4 (7.0) 77.6 (3.6)
10
8H
.F
.M
itchell
an
dD
.T
.K
enny
Lo
gop
edP
ho
nia
trV
oco
l2
9
79
difference between P1 and P2 was significantly lower
in LSO than in SO. There was no significant difference
between O and SO/LSO (F(1,5)�/5.101, p�/0.133).
There was no interaction effect between task and
condition. These data are presented in Fig. 6.
The SPR of each task and condition was calculated
for each singer to compare the relationship of energypeaks between 0�/2 kHz and 2�/4 kHz for inter-subject
differences in task and intra-subject effects of condi-
tion. Individual SPR results are presented in Table 4.
Lower SPR results, indicating more energy between
2�/4 kHz, from O to SO were seen in three Mozart SO
and two Schubert SO cases. The SPR was lower for all
singers in LSO compared with SO.
Hypothesis 5: ER measurements will increase in SO/
LSO compared to O
Hypothesis 5 stated that ER will increase in SO/LSO
compared to O. The ER was calculated for each
musical task in each condition for each subject.
Individual subjects’ ER by task and condition are
presented in Table 5. The changes to ER results by
singer corresponded to SPR results in the Mozart and
Schubert tasks from SO/LSO compared to O, that is,ER decreased with condition across singers. In the
Schubert task, subject 5 was an outlier.
Hypothesis 5 was not confirmed for O and SO/LSO.
There was a weak trend showing the effect of task on
ER (F(1,5)�/5.278, p�/0.070), that is, ER in the
Schubert task was slightly greater than in Mozart,
but both musical tasks showed the same overall
pattern across conditions. There was an effect forcontrast 2 only, (F(1,5)�/36.913, p�/0.002), that is, ER
was significantly less in LSO than in SO. There was no
Fig. 3. Main effect for SPL (in dB at 7 cm) between
sub-optimal and loud sub-optimal. Means for two
musical tasks (Mozart and Schubert) in each of the
three conditions, optimal, sub-optimal and loud sub-optimal.
Fig. 4. Main effect for P1 between optimal and sub-
optimal/loud sub-optimal. Means for the two musicaltasks (Mozart and Schubert) in each of the three
conditions, optimal, sub-optimal and loud sub-opti-
mal.
Fig. 5. Main effect for P2 between sub-optimal and
loud sub-optimal. Mean differences between the two
musical tasks (Mozart and Schubert) in each of the
three conditions, optimal, sub-optimal and loud sub-
optimal.
Effects of open throat technique on LTAS 109
Logoped Phoniatr Vocol 2980
significant difference between O and SO/LSO (F(1,5)�/
3.531, p�/0.119). There was no interaction effect
between task and condition. These data are presented
in Fig. 7.
The energy distribution differences in the LTAS of
each singer were compared by subtracting SO from O
(illustrated in Fig. 8), LSO from O and LSO from SO.
The energy differences for all six singers between O/
SO, O/LSO and SO/LSO in the Mozart and Schubert
tasks are presented in Fig. 9a�/c. Data points above
zero represent an increase in energy at the particular
frequency whereas data points below zero indicate a
decrease in energy between conditions.
At peak level P1 (between 500 Hz and 1 kHz), Fig.
9a shows that all singers reduced acoustic energy
between O and SO. Similarly, in Fig. 9b, singers
showed a reduction in the energy of P1 in LSO
compared to O. Conversely, there was an increase for
all singers between 0�/2 kHz between LSO and SO.
Changes around peak level P2 were more complex.
The differences between O and SO, and O and LSO
were characterized by a series of peaks and troughs at
points between 2�/4 kHz rather than a consistent
reduction or increase in energy at all data points.
Between SO and O, these differences are associated
with a reduction of SPL (Fig. 9a), and between LSO
and O, the changes represent a combination of
increases and decreases in SPL. Fig. 9c shows the
subtracted differences between LSO and SO where the
majority of these data points show an increase in
acoustic energy. This indicates that for these singers
the shape of the SO and LSO LTAS plots was similar,
but with greater overall energy in LSO.
Fig. 9a and b show a prominent increase of energy
at 2 kHz. In Fig. 2a and b, the SO and LSO conditions
showed a narrower P2, predominantly in the Mozart
musical task. The variation in energy around 2 kHz in
Fig. 9a and b shows that, rather than a consistent
increase of energy, the shape of the peaks had
changed. The shape of the P2 peak differs for each
singer depending on condition, illustrated in the thin
and dashed lines in Fig. 2a and b.
Table 3. Correlations (r), intercepts and slope of linear regressions of LTAS values calculated in the optimal and
sub-optimal, and optimal and loud sub-optimal conditions of the Mozart and Schubert Tasks
O-SO O-LSO SO-LSO
Task Singer r Intercept Slope r Intercept Slope r Intercept Slope
Mozart1 0.97 �/0.48 0.94 0.95 0.36 0.80 0.99 0.79 0.852 0.93 0.68 1.10 0.95 �/1.72 0.83 0.94 �/3.92 0.693 0.96 �/0.78 0.87 0.98 �/1.26 0.83 0.97 �/0.46 0.914 0.96 �/2.48 0.91 0.93 �/3.07 0.78 0.98 �/0.65 0.875 0.91 1.61 0.88 0.86 3.05 0.80 0.98 2.29 0.956 0.98 �/0.44 0.91 0.95 0.88 0.80 0.96 1.08 0.79
Schubert1 0.97 �/0.62 0.84 0.95 �/0.44 0.77 0.99 0.28 0.942 0.96 �/0.94 0.94 0.97 �/1.38 0.80 0.97 �/2.90 0.793 0.96 �/1.42 1.00 0.99 �/0.43 1.00 0.98 �/0.05 0.954 0.92 �/1.51 0.83 0.92 �/0.87 0.71 0.99 0.32 0.855 0.96 �/0.87 0.85 0.96 �/0.05 0.72 0.98 0.31 0.836 0.95 0.81 0.81 0.93 �/1.95 0.72 0.96 �/0.46 0.88
Fig. 6. Main effect for singing power ratio (SPR)
differences between P1 and P2 between sub-optimal
and loud sub-optimal. Means of difference between P1and P2 in the two musical tasks (Mozart and
Schubert) in each of the three conditions, optimal,
sub-optimal and loud sub-optimal.
110 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2981
Hypothesis 6: Changes observed in each condition will
be consistent with reductions in vibrato extent and
increases in vibrato onset in SO/LSO compared to O
Hypothesis 6 stated that changes in vibrato observed
in each condition will be reflected in the correspond-
ing LTAS plots for each singer. Specifically, the
reduction in vibrato extent and onset would affect
each singer’s LTAS curve in SO/LSO compared to O.
In each musical task, fundamental frequency distribu-
tion (F0) was affected by an absence or decrease of
vibrato. Both a reduction of extent and an increase inthe delay of onset meant that for the time period of
each musical task, fewer pitches were sung. Mean F0
in each condition and in each task varied little across
conditions (B/20 Hz) in 23 of 24 SO and LSO takes
compared to O. The changes to the distribution of F0
across tasks are presented in Fig. 10 for each task in
each condition. These panels, of O, SO and LSO F0
distribution respectively, show a decrease in the spreadin F0 covered during each task. In Fig. 10a, there is
one mean distribution, for the same key used by all
singers, but in Fig. 10b, there are three separate plots
representing the respective keys used by sopranos and
mezzos.
Fig. 11 presents normalized LTAS of singer 3’s
Mozart task with inset panels of F0 distribution. It
presents an exemplar of the effects seen because ofvibrato extent changes to the F0 and the correspond-
ing shape changes to the LTAS across conditions.
DISCUSSION
Long-term average spectra of singers in this study
showed that there were significant differences between
SO and LSO but not between O and SO/LSO in
spectral distribution of energy peaks and area between
0�/2 kHz and 2�/4 kHz and in overall SPL. For the
individual peaks P1 and P2, there was a significant
difference in the level of P1 in O compared with SO/
Table 4. Measurements of singing power ratios long
term average spectra analysed for the Mozart and
Schubert musical tasks in each condition, optimal
(O), sub-optimal (SO) and loud sub-optimal (LSO)
SPR
Condition
Task Subject O SO LSO
Mozart1 20.2 14.9 12.62 24.4 24.7 20.33 14.0 17.4 15.04 22.0 22.0 19.05 20.2 18.2 14.56 22.2 15.8 14.9
Schubert1 17.8 19.5 14.32 24.8 32.7 24.13 21.1 23.2 18.14 24.7 26.6 24.55 21.1 15.9 13.76 20.1 19.0 13.4
Table 5. Measurements of energy ratios over the long
term average spectra analysed for the Mozart and
Schubert musical tasks in each condition, optimal
(O), sub-optimal (SO) and loud sub-optimal (LSO)
ER
Condition
Task Subject O SO LSO
Mozart1 17.0 14.9 13.12 19.6 19.2 15.13 12.1 16.8 14.24 19.1 17.3 14.15 17.4 15.2 12.06 17.4 13.1 11.1
Schubert1 16.1 17.9 13.62 22.2 29.4 20.33 18.5 20.2 15.84 20.1 20.9 18.15 16.2 14.1 12.66 17.2 16.1 10.2
Fig. 7. Main effect for energy ratios (ER) between the
sub-optimal and loud sub-optimal conditions. Meansof the difference between ER in the two musical tasks
(Mozart and Schubert) in each of the three conditions,
optimal, sub-optimal and loud sub-optimal.
Effects of open throat technique on LTAS 111
Logoped Phoniatr Vocol 2982
LSO, but for P2, there was a difference between SO
and LSO. Based on the regression plots, there were
highly statistically significant correlations between the
shapes of singers’ LTAS across all conditions, indicat-
ing the lack of sensitivity of such measures in detecting
timbral differences.
SPL has an effect on LTAS slope, such that an
increase in SPL typically increases high frequency
energy at a faster rate than lower frequency energies
(34, 36). Therefore, it was important to generate more
than one experimental condition to test the impact of
open throat technique in singing. In addition, we
chose to use average SPL as a separate measure from
power levels in these analyses to detect differences
between conditions (36). The inclusion of a second
comparison group, LSO, indicated that SPL did not
account for differences between O and SO. Foulds-
Elliot (30) and Rossing (15) have used similar study
designs. The hypothesized reduction in SO SPL did
not in fact occur; however, there were differences in the
energy distributions of each condition. While SPL
would be expected to be higher than both P1 and P2,
the SPL was the average of the sound pressure (or
power) level, rather than the levels of the average
powers that were accumulated into the LTAS in the
respective frequency regions for P1 and P2. Because
the distribution of power in typical sounds is heavily
skewed toward low powers, with brief excursions to
very high powers, the average of the power level will
tend to be lower than the level of the average power.
The average SPL more accurately reflects the typical
sound level than the level of the average power.There were statistically significant correlations be-
tween conditions in the shape of singers’ normalized
LTAS. The variations in positive and negative inter-
cepts are indicative of spectral differences greater than
4 kHz. When Cleveland et al. (11) compared country
singers’ speech and singing they found a slope close to
1 and an intercept close to zero indicating similarity
between speech and singing plots in country singers.
While strong similarities might be expected in country
singing and speaking, a far greater difference would be
expected between optimal and sub-optimal attempts
in classical singing, but this was not the case. LTAS is
used in clinical contexts as an indicator of the natural
voice sound (8), and therefore should have been ideal
for assessing the changes in the timbre associated with
each condition.
As the LTAS curves were so similar, detailed
analyses of spectral distribution on the calibrated
data were essential. The increased P1 in O confirms
that ‘open throat’ is often characterized by a more
dominant fundamental partial. Subjectively, this
would correspond to a rounder and less pressed
sound. Calibrating these LTAS enabled precise infor-
mation on the relative or absolute level of the peaks
(15, 32), which would have been lost in normalized
LTAS curves (11).
Measures of SPR and ER reduce LTAS information
to a single number, which can be used to compare
intra- and inter-singer energy. The difference in energy
between 0�/2 kHz and 2�/4 kHz is indicative of the
carrying power of the voice. A smaller difference
between P1 and P2, or SPR, identified LSO as the
most effective for vocal projection. In LTAS, a smaller
difference between peak levels P1 and P2 implies
improved carrying power and a better classical quality
(10, 18). This is the case for male singers in production
of the singer’s formant (21). For SO, there was a
reduction of P2 compared to LSO, and therefore
reduction in carrying power. In the case of LSO,
Vurma and Ross (18) have previously reported that a
small difference between peaks was not necessarily
associated perceptually with an improved vocal qual-
ity.
Measuring the difference between peaks provides a
representation of energy rather than of quality. As a
form of measurement, Lundy et al. (35) also found
SPR (26) problematic in assessing energy differences
between students’ sung and spoken vowels. They
found no differences between the SPR of each. In
addition to the level of the peak, the centre frequency
of P2 has been used to define an energy boost in the
Fig. 8. Example of singer 1’s LTAS plots in the
optimal and sub-optimal condition for the Mozart
task. Below is an example of the calculation of energy
difference between the optimal and sub-optimal con-ditions.
112 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2983
singing voice. Centre frequencies in O, SO and LSO
ranged from 3000�/3500 Hz in all conditions which
was in keeping with Dmitriev and Kiselevs’ (38)
findings for female singers: 2700 and 3500Hz
(mezzo-soprano to high soprano). Although Weiss et
al. (22) have questioned the legitimacy of the singer’s
Fig. 9a�/c . (a) Subtracted energy between the optimal and sub-optimal conditions, for each singer in the Mozarttask, shown in the left column, and Schubert, in the right column. (b) Subtracted energy between the optimal and
loud sub-optimal conditions. (c) Subtracted energy between the loud sub-optimal and sub-optimal conditions.
Effects of open throat technique on LTAS 113
Logoped Phoniatr Vocol 2984
Fig. 10a�/b. (a) The Mozart task: mean fundamental frequency distribution for all six singers showing the
histogram distribution on the ordinate from 0�/0.2 occurrences at frequencies from 200�/900Hz on the abscissa in
optimal, sub-optimal and loud sub-optimal conditions. (b) The Schubert Task, in the three musical keys used by
the six singers.
114 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2985
formant in sopranos, they also found stronger spectral
peaks around 3 kHz. These female singers’ spectral
reinforcement was significantly less than that of male
singers.
Using conventional measures to analyse the differ-
ences in LTAS by condition did not correspond to the
previous study, where SO/LSO were found not to
indicate acceptable vibrato. However, visual inspection
of Fig. 2a and b shows that, while the difference
between peaks was much closer in the LSO condition,
the shape of P2 and energy above 2 kHz was in
complete contrast to O.
Measuring ER obtained specific information point
by point in LTAS to describe the differences in LTAS
by condition. Overall ER results followed SPR.
Exploring these ER results, through the subtracted
data (after Thorpe et al . (25)) effectively demonstrated
the changes to singers’ LTAS plots at every point.
Thorpe et al. found increases at all data points
between spectra in normal singing and using vocal
projection, but with considerable energy increases
above 2 kHz. Fig. 9a�/c clearly indicates that the
overall energy distribution fluctuates between condi-
tions; specifically, the shape of a singer’s LTAS
changed by condition. Subtracted ER shows that
while levels of P2 (in dB) are regularly matched in
LSO compared to O where the centre frequency is
close to or above zero, the level of points to either side
decrease below zero. This clearly indicates the change
in LTAS shape between O/LSO. Between O/SO, the
results are similar, but P2 in SO is lower than O, so the
dB level fluctuations occur below zero, illustrating a
decrease in overall energy distribution.
Subtracted data between SO/LSO produced results
more like Thorpe et al. (25) as all subjects demon-
strated increases in energy or spectral density over 2
kHz. This implies that there was not a difference in
LTAS shape between SO/LSO, but rather an increase
in energy level at every frequency point. As LSO was
devised as a control for loudness, specifically to
remove SPL as sole variable for the SO condition
(30), it is interesting to note that despite the differences
in overall SPL, LSO is a louder replica of SO.
Explanations for unusual peaks or troughs between
O/SO and O/LSO can be traced in Fig. 2. For example,
greater energy around 2 kHz does not represent a new
peak, but rather, highlights a change in energy
distribution above and below this frequency. The
variability of increase or decrease is far greater in the
O/SO and O/LSO comparisons, and if peaks cancel
out troughs, and so on, this may explain an equalizing
in overall ER results or no statistical difference
between O and SO/LSO. Computing ER as a single
meaningful number across the entire 0�/2 kHz and 2�/4
kHz regions may not be helpful in identifying changes
in timbre or energy. However, it appears to be a more
sensitive measure compared to regression plots and
peak height as a way of visualizing changes across
LTAS.
Fig. 11. Normalized examples of singer 3’s LTAS
plots in optimal, sub-optimal and loud sub-optimal
conditions for the Mozart task. Insets represent the
histogram of fundamental frequency for the same
rendition of the musical task.
Effects of open throat technique on LTAS 115
Logoped Phoniatr Vocol 2986
Calculating a mean of all singers’ energy increases
may not represent the energy increase of any singer
(25). The individuality of change in the subtracted
data for these singers suggests unique changes to their
vocal quality, and to their energy spectra (35). A mean
of all subjects’ performance of both musical tasks
would have obscured rather than illuminated the
unique qualities of each voice. The mean line through
ER differences illustrates the trends of change by
condition, but does not represent the changes to any
particular voice.
Changes to F0 distribution (illustrated in Fig. 10)
were accounted for by a reduction in vibrato (5), as the
musical tasks were identical in O, SO and LSO. These
singers’ O vibrato had a wider pitch range compared
to SO and LSO. This in turn may be reflected in the
spectral distribution of the LTAS plot showing peaks
in the 2�/4 kHz region of SO and LSO, where they are
distinctly narrower or include multiple peaks. Inter-
estingly, Sundberg (39) linked vibrato to the distribu-
tion of partials in the sound spectrum. Increased
vibrato extent has also been linked specifically to
spectral changes in sopranos’ solo singing compared
to choral singing, where it increased the level (dB) of
P2, and changed its overall distribution and the shape
of the peak (15). Although it was hypothesized that
sopranos produced this energy above 2 kHz by
increasing vibrato extent rather than clustering higher
formants as in male singer’s formant, results in this
study show increases in P2 by around 3 dB and not the
15 dB identified by Rossing et al. (15). Further
investigation of the effect of vibrato extent on spectra
of individual notes is needed. The changes of vibrato
parameters were not linked to such a significant
reduction in P2, suggesting that singers still increased
LSO energy above 2 kHz in other ways.
The broader P2 in O is possibly but not necessarily
due to vibrato. If open throat involves a larger mouth
opening, for example, this would reduce the Q factor
of the formant resonances, which would broaden but
also lower the peaks. P2 could still be higher in O than
in SO if the total output is higher in O. In any case, the
fact that partials are moving around vigorously due to
F0 modulation in vibrato is in itself not sufficient to
explain a broader peak in the spectrum envelope.
Rather, concomitant motions of the pharynx walls can
be expected to perturb the formant frequencies, which
would smear the peaks. Further study of these possible
explanations for the shapes of the peaks is needed.The instruction to achieve the unusual condition of
LSO by using more ‘twang’ may explain the difference
in singers’ P2 shape from O to LSO. Titze (40)
compared twang to ring quality and although he
found them similar spectrally, defined twang differ-
ently as a narrowing of the pharynx. This pedagogical
instruction has been associated with production of
resonance in the classical voice (21, 25, 40). However,
for these singers, LSO produced a similar peak levelbut different spectral distributions over the whole peak
compared to O.
Pedagogical implications
Singing pedagogues rely on a series of singing
techniques to produce the best possible sound from
each student. Pedagogues evaluate each voice on an
individual basis and internally assess what combina-
tion of instructions is necessary for the particular
voice to achieve its full potential. As researchers strive
to define a ‘good sound’ in singing through acoustic
measurement, it is important to assess a voice as apedagogue would, using the overall sound to devise
appropriate vocal exercises to achieve optimal quality
for that voice.
Teachers have a mental image of the particular
beautiful sound individual students are capable of
producing and use verbal explanation, vocal modelling
and feedback to teach this. The acoustic information
currently available has limited application in thisprocess. Mitchell and Kenny confirmed that the use
of open throat technique is useful in producing a good
classical sound and demonstrated its impact on the
regularity and evenness of singers’ vibrato rate, extent
and onset when the technique is reduced (5). Pedago-
gues did not explicitly attribute the singer’s formant or
increased carrying power to the use of open throat
technique but they did suggest its use facilitated abetter balance or more desirable distribution of
harmonics.
Poor vocal quality could not be defined in terms of
its spectral distribution in a way that we would have
expected. As our previous investigations confirmed
that SO and LSO did not reach the standards of good
classical singing, the purported link between vibrato
and overall quality illustrated through LTAS appearsto occur in the narrow or multiple peaks above 2 kHz.
Acoustic implications
LTAS is an ideal representation of a singer’s sound ina natural setting, that is, an entire performance rather
than single notes. However, despite�/60 seconds of
musical stimuli, the effect of task approached signifi-
cance for measures of SPR and ER. Jansson and
Sundberg (12) investigated LTAS by task and condi-
tion and in musical task noted different LTAS shapes
when the same instrument played the same scale in a
different musical key or played a triad figure comparedto a chromatic scale, which produced narrower and
more rounded peaks above 2 kHz. In light of these
116 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2987
findings, LTAS should be examined in the context of
specific tasks and conditions.
If LTAS does change depending on the group ofpitches analysed (12, 41), it is important in future to
address one variable in any experiment: the same
singer singing in two ways or two different singers
singing the same musical task. In this study, we
considered each singer’s spectra individually. The
mean line of subtracted ER does not represent any
one singer, and virtually cancels out the effect of the
individual singers. Previous templates of LTAS curveshave been produced from averages generated from
groups of singers (14). These averages may, in fact, not
represent any one voice and may be an artefact of
clustered measurement.
Assessment of LTAS is complex. Visual inspection
identifies obvious differences between the LTAS of
each condition; yet performing a variety of conven-
tional analyses could not identify statistically signifi-cant intra- or inter-singer differences to support the
vibrato findings, where SO/LSO represented poor
singing. Future research needs to evaluate in greater
detail factors that contribute to the shape of singers’
LTAS, both acoustically and perceptually.
CONCLUSION
Lessening the amount of open throat used in singing
produced few statistically significant differences to
these singers’ LTAS in each condition. Our perceptual
and acoustic work using open throat technique as the
dependent variable in assessment has shown that
conventional acoustic measures of voice quality donot correspond to perceptual assessments of vocal
beauty or mastery. A science of the singing voice
cannot progress without addressing the problem that
standard acoustic measures may not be sufficiently
sensitive to track perceived changes. We propose to
classify acoustic descriptors of these voices through
perceptual ratings of expert listeners to better evaluate
LTAS in singing and explore the role of LTAS as arepresentation of timbre in singing as well as energy.
This will complete the initial assessment of a single
technique and its impact on overall classical quality as
taught in the singing studio.
ACKNOWLEDGEMENTS
We thank the six singers for their willing participation
in the project and Ms Maree Ryan for her expert
pedagogical advice. The authors also thank ProfessorPamela Davis for early ideas on the research question
and study design. We are grateful for the helpful
comments of the reviewers in the revision of this
paper.
REFERENCES
1. Hemsley T. Singing and imagination: a human approachto a great musical tradition. Oxford, New York: OxfordUniversity Press; 1998.
2. Miller R. The structure of singing: system and art invocal technique. New York, London: Schirmer Books;1996.
3. Vennard W. Singing: the mechanism and the technic, 5thed. New York: Fischer; 1968.
4. Mitchell HF, Kenny DT, Ryan M, Davis PJ. Definingopen throat through content analysis of experts’ peda-gogical practices. Logoped Phoniatr Vocol 2003; 28:167�/80.
5. Mitchell HF, Kenny DT. The impact of ‘open throat’technique on vibrato rate, extent and onset in classicalsinging. Logoped Phoniatr Vocol. In press.
6. Ekholm E, Papagiannis GC, Chagnon FP. Relatingobjective measurements to expert evaluation of voicequality in Western classical singing: critical perceptualparameters. J Voice 1998; 12: 182�/96.
7. Robison CW, Bounous B, Bailey R. Vocal Beauty: Astudy proposing its acoustical definition and relevantcauses in classical baritones and female belt singers.Journal of Singing 1994; 51: 19�/30.
8. Lofqvist A. The long-time-average spectrum as a tool invoice research. Journal of Phonetics 1986; 14: 471�/5.
9. Mendoza E, Munoz J, Naranjo N. The Long-TermAverage Spectrum as a Measure of Voice Stability. FoliaPhoniatr Logop 1996; 48: 57�/64.
10. Novak A, Vokral J. Acoustic Parameters for theEvaluation of Voice of Future Voice Professionals. FoliaPhoniatr Logop 1995; 47: 279�/85.
11. Cleveland TF, Sundberg J, Stone RE. Long-Term-Average-Spectrum Characteristics of Country SingersDuring Speaking and Singing. J Voice 2001; 15: 54�/60.
12. Jansson EV, Sundberg J. Long-Time-Average-SpectraApplied to Analysis of Music. Part I: Method andGeneral Applications. Acustica 1975; 34: 15�/9.
13. Lofqvist A, Mandersson B. Long-Time Average Spec-trum of Speech and Voice Analysis. Folia PhoniatrLogop 1987; 39: 221�/9.
14. Sundberg J. Level and center frequency of the singer’sformant. J Voice 2001; 15: 176�/86.
15. Rossing TD, Sundberg J, Ternstrom S. Acoustic com-parison of soprano solo and choir singing. J Acoust SocAm 1987; 82: 830�/6.
16. Borch DZ, Sundberg J. Spectral distribution of solovoice and accompaniment in pop music. LogopedPhoniatr Vocol 2002; 27: 37�/41.
17. Murbe D, Sundberg J, Iwarsson J, Friedmann P, Hof-man G. Longitudinal study of solo singer educationeffects on maximum SPL and level in the singers’formant range. Logoped Phoniatr Vocol 1999; 24:178�/86.
18. Vurma A, Ross Y. Priorities in voice training: carryingpower or tone quality. Musicae Scientiae 2000; 4: 75�/93.
19. Sundberg J. The Science of the Singing Voice. DesKalb,Illinois: Northern Illinois University Press; 1988.
Effects of open throat technique on LTAS 117
Logoped Phoniatr Vocol 2988
20. Bartholomew W. A physical definition of ‘good voicequality’ in the male voice. J Acoust Soc Am 1934; 6: 25�/
33.21. Sundberg J. Articulatory interpretation of the ‘singing
formant’. J Acoust Soc Am 1974; 55: 838�/44.22. Weiss R, Brown WS Jr, Morris J. Singer’s formant in
sopranos: fact or fiction? J Voice 2001; 15: 457�/68.23. Sundberg J. The singer’s formant revisited. Voice 1995;
4: 106�/19.24. Sundberg J. The Acoustics of the Singing Voice.
Scientific American 1977; March: 82�/91.25. Thorpe C, Cala S, Chapman J, Davis P. Patterns of
breath support in projection of the singing voice. J Voice2001; 15: 86�/104.
26. Omori K, Kacker A, Carroll LM, Riley WD, BlaugrundSM. Singing power ratio: quantitative evaluation ofsinging voice quality. J Voice 1996; 10: 228�/35.
27. Bunch M, Chapman J. Taxonomy of singers used assubjects in scientific research. J Voice 2000; 14: 363�/9.
28. Wapnick J, Ekholm E. Expert consensus in solo voiceperformance evaluation. J Voice 1997; 11: 429�/36.
29. Cabrera D, Davis P, Barnes J, Jacobs M, Bell D.Recording the Operatic Voice for Acoustic Analysis.Acoustics Australia 2002; 30: 103�/8.
30. Foulds-Elliott S, Thorpe C, Cala S, Davis P. Respiratoryfunction in operatic singing: effects of emotional con-nection. Logoped Phoniatr Vocol 2000; 25: 151�/68.
31. Estill J. Primer of Basic Figures, 2nd ed. Santa Rosa:Estill Voice Training Systems; 1996.
32. White P. A study of the effects of vocal intensityvariation on children’s voices using long-term averagespectrum (LTAS) analysis. Logoped Phoniatr Vocol1998; 23: 111�/20.
33. White P. Formant Frequency Analysis of Children’sSpoken and Sung Vowels Using Sweeping FundamentalFrequency Production. J Voice 1999; 13: 570�/82.
34. Nordenberg M, Sundberg J. Effect on LTAS of vocalloudness variation. TMH-QPSR, KTH 2003; 45: 93�/100.
35. Lundy DS, Roy S, Casiano RR, Xue JW, Evans J.Acoustic Analysis of the Singing and Speaking Voice inSinging Students. J Voice 2000; 14: 490�/3.
36. Barnes JJ, Davis PJ, Oates J, Chapman J. The relation-ship between professional operatic soprano voice andhigh range spectral energy, J Acoust Soc Am 2004. Inpress.
37. Miller R. The Open Throat (La gola aperta). In: MillerR, editor. On the Art of Singing. New York: OxfordUniversity Press; 1997. p. 17�/9.
38. Dmitriev L, Kiselev A. Relationship between theformant structure of different types of singing voicesand the dimensions of supraglottic cavities. FoliaPhoniatrica 1979; 31: 238�/41.
39. Sundberg J. Perceptual Aspects of Singing. J Voice 1994;8: 106�/22.
40. Titze I. Acoustic Interpretation of Resonant Voice. JVoice 2001; 15: 519�/28.
41. Howard DM, Szymanski J, Welch GF. Listeners’Perception of English Girl and Boy Choristers. MusicPerception 2002; 20: 35�/49.
118 H. F. Mitchell and D. T. Kenny
Logoped Phoniatr Vocol 2989
4.1 Conclusion
Visual inspection of long-term average spectra (LTAS) confirmed differences between
O and SO in this study, but results showed that conventional acoustic measures on
LTAS did not produce statistically significant differences between the experimental
conditions. These results were not consistent with the vibrato findings and suggest that
measures of SPR [the difference between the peak height (Omori et al., 1996)] and ER
[the difference between peak areas (Thorpe et al., 2001)] are not sufficiently sensitive to
evaluate LTAS.
These findings represent a major challenge to current wisdom that acoustic parameters
of voice can identify changes in vocal timbre. Subsequent studies compare acoustic
output to perceptual assessments. The fourth study addressed the perceptual significance
of open throat technique and its impact on overall singing quality. The fifth study
considered the effectiveness of acoustic analysis in assessing vocal quality.
90
5 Perceptual verification
Evaluation of singing in tertiary institutions revealed that the assessment of singing was
based initially on the whole performance, before appraisal of specific technical and
performance components (Davidson & Da Costa Coimbra, 2001; Stanley, Brooker, &
Gilbert, 2002). Music performance studies, particularly in singing, find that judges do
not explicitly discuss an overall ‘good sound’, but rather, focus on aspects of technical
control (interaction between the voice and body, the role of diction on control, register
control, and adequate support for the sound) (Davidson & Da Costa Coimbra, 2001).
Lower marks were associated with a greater number of suggestions for technical
improvements.
Formal criteria are designed to complement rather than replace subjective responses to
performance assessment (Johnson, 1997). Research studies that use specific rating
scales (Boyle, 1992) have reported high correlations of all objective criteria with the
overall judgement (Ekholm et al., 1998; Geringer & Madsen, 1998; Saunders &
Holahan, 1997). As the individual dimensions are attributed to the final mark awarded,
it may not be possible to isolate the separate features that constitute overall quality.
Although some examiners believe criteria-specific evaluation interfered with the holistic
assessment of performance, they reported that use of criteria focussed their evaluation
on elements of performance, and was useful for student feedback. Perceptual studies in
music must address the global approach of listeners before focusing listeners to a single
91
technical component (Stanley et al., 2002). This enables judges to justify their response
and provide specific feedback for students.
Our aim was to determine the measurable differences between an optimal sound quality
a singer can produce involving the use of open throat, and a sound which uses a lesser
degree of the open throat technique, without altering other components of their singing
technique or sound quality. In Western classical singing, the requirements for good
vocal training and final assessment focus on the overall perception of a good sound.
Technical considerations focus on any indicators of poor quality (Davidson & Da Costa
Coimbra, 2001) rather than a series of specific technical attributes. However,
pedagogical literature focuses on a wide variety of techniques which contribute to this
final sound.
Fifteen experienced singing pedagogues and adjudicators assessed 30 samples generated
from the data collected for studies 2 and 3, and rated each sample as O or SO. Listeners
were presented with the definitions of each experimental condition, and of the technique
of open throat. The results of this perceptual study confirmed that listeners could
reliably identify the presence of open throat technique and were consistent in their
judgements of repeated samples.
92
PAPER 4
Mitchell, H. F., & Kenny, D. T. (in press). Can experts identify open throat technique as a perceptual phenomenon? Musicae Scientiae.
93
Perceptual identification of open throat 1Musicae Scientiae in press
Can experts identify “open throat” technique as a perceptual phenomenon?
Helen F. Mitchell and Dianna T. Kenny
From the Australian Centre for Applied Research in Music Performance (ACARMP), Sydney Conservatorium of Music, The University of Sydney, New South Wales 2006, Australia
Received 25 July 2004. Accepted 7 February 2005.
Musicae Scientiae; in press
Singing pedagogues have identified ‘open throat’ as a widely used pedagogical tool in the singing studio and a critical technique to achieve good classical vocal quality. This study is the first to assess the perceptual validity of open throat. Fifteen expert singing pedagogues assessed 48 messa di voce and 24 song samples with six repeats of six advanced singing students under two conditions: ‘optimal’ (O), representing use of maximal open throat technique and ‘subop-timal’ (SO), representing reduced open throat. Correctly identified responses were counted by condition (O/SO), by judge and by singer. Data were analyzed using Cohen’s Kappa. Hypotheses that correct identification would be greater than chance were confirmed for both messa di voce and the song samples, with thirteen of the fifteen judges correctly identifying 82.7% of song samples as O or SO. Singers’ self-ratings attributed their best singing to maximal use of open throat technique. These results indicate that listeners are consistent in making a dichotomous choice in identify-ing use of a singing technique through auditory-perceptual evaluation. The findings from this study suggest that there is a sound quality associated with the use of open throat technique, which is a perceptual reality to singing pedagogues and singers and that the specific vocal quality in classical singing that it produces can be reliably identified by expert listeners. If singers and expert listeners’ of singing accord, focussing future research on specific singing techniques could enhance singing pedagogy.
Helen Mitchell, Australian Centre for Applied Research in Music Performance (ACARMP), The Conservatorium of Music, The Uni-versity of Sydney, New South Wales, Australia 2006 Phone: 61 2 9351 9644; Fax: 61 2 9351 9540 Email: [email protected]
INTRODUCTION
In this paper, we report the first study to assess the perceptual validity of a specific singing technique – open throat (OT) – as taught in today’s singing studio. Previous perceptual studies (Ekholm, Pa-pagiannis, & Chagnon, 1998; Mendes, Rothman, Sapienza, & Brown, 2003; Robison, Bounous, & Bailey, 1994; Vurma & Ross, 2000; Wapnick & Ekholm, 1997) have demonstrated that singing improves over time and through training but have not identified specific components of the training that produce improvements in vocal quality and thus add little to the singing pedagogical literature. In order to evaluate the contribution of one sing-ing technique (OT), we have assessed the acoustic aspects of voice quality produced with and with-out this technique and the perceptual impressions of listeners, firstly their capacity to identify voices that used the technique and then to describe the vocal quality produced.
Open Throat Technique (OT)
The term “open throat” occurs frequently in the
vocal pedagogy literature (Burgin, 1973; Fields, 1947; Monahan, 1978). It is defined as a complex process that is both a pedagogical instruction and a perceived sensation or action that results in a specific sound quality. Current support for the use of OT in singing technique is widespread (Miller, 1996; Reid, 1975, 1983). When OT is used, peda-gogues perceive the sound as resonant (Miller, 1996; Vennard, 1968), round, (Joiner, 1998), free (Ware, 1998), pure, (Marafioti, 1981) rich and warm (McKinney, 1982). The sound quality is attributed to freedom from ‘constrictor tensions’ (Reid, 1983, p. 83). In the pedagogical literature, the sound quality attributed to open throat is linked to the preparation to sing, on inhalation (Hemsley, 1998; Manèn, 1987; Miller, 1997a, 1997b) through the surprise breath or smelling the rose imagery (Hemsley, 1998; Miller, 1996; Puritz, 1956) and visualizing space within the throat, through an ‘air-ball’ or ‘soap bubble’ (Herbert-Caesari, 1951; Manèn, 1987) to achieve the posture of OT.
The technique of OT is a pedagogical concept transmitted through the oral tradition of singing. Mitchell, Kenny, Ryan & Davis (2003) assessed
94
Perceptual identification of open throat 2Musicae Scientiae in press
current pedagogical practices and perceptions of OT. They noted that expert singing pedagogues identified OT as a critical technique in vocal pro-duction of classical singing. Despite the use of dif-ferent terminology to describe the technique (such as freedom, collar and retraction), OT was defined by these pedagogues as a technique to maximize pharyngeal space and/or abduct the ventricular folds. There was consistency in the vocal instruc-tions to achieve it. Pedagogues teach conscious control of OT using laugh, sob, correct inhalation or maintaining the posture of inhalation. They con-sistently reported a perceptually distinctive sound quality as a result of mastery of this technique and described their perception of the final sound qual-ity as balanced, coordinated, free, even, consist-ent, warm and open.
Hypotheses generated from pedagogical beliefs and practices derived in the first study on OT were subsequently tested acoustically by Mitchell and Kenny (2004b) who compared the use of open throat technique (optimal, O) to a reduction of the technique (sub-optimal, SO) using female classi-cal singers. Reduced OT resulted in significant re-ductions in vibrato extent and increases in vibrato onset time but no change to vibrato rate between the two conditions. These singers relied on OT to produce acceptable vibrato parameters. However, according to the literature, inappropriate vibrato is indicative of poor singing in general; therefore further acoustic tests were required to test the dif-ferences in timbre. Visual inspection of long term average spectra (LTAS) confirmed differences be-tween O and SO for each singer, and the O con-dition produced a rounder peak between 0-2 kHz indicating a warmer sound quality compared to SO. Despite these findings, LTAS did not demon-strate significant differences between SO and O in conventional measures performed on long term average spectra energy peak height [singing pow-er ratio (Omori, Kacker, Carroll, Riley, & Blau-grund, 1996)] and energy peak area [energy ratio (Thorpe, Cala, Chapman, & Davis, 2001)].
Technology to measure voice acoustically has become more sophisticated and is used increas-ingly in experimental voice research, the findings of which are now being incorporated into texts on singing (Miller, 1996; Nair, 1999; Sundberg, 1977; Thurman & Welch, 2000). To date, research into the singing voice has described acoustic prop-erties of voice and its visual representation with few or no links to pedagogical descriptors or per-ceptual judgments. The implications of acoustic studies of the singing voice within singing lit-
erature and pedagogy are difficult to ascertain, since most studies do not assess acoustically or perceptually the vocal strategies used in the sing-ing studio (Foulds-Elliott, Thorpe, Cala, & Davis, 2000; Thorpe et al., 2001). Indeed, few studies link acoustic findings with perceptual judgments (Ekholm et al., 1998) or discuss their pedagogical implications (Callaghan, 2000; Miller, 1998).
This study seeks to provide stronger links be-tween empirical research and singing pedagogy, and recognises the importance of singing peda-gogues in this enterprise.
Perceptual Studies
Perceptual evaluation of musical performance is important for a number of reasons. Opera singers, their teachers and audiences are primarily inter-ested in how the voice sounds and use the sensory information derived from listening as the prima-ry basis for determining the quality of the sound (Kitch & Oates, 1994). Despite recent advances in the physiological and acoustic measurement of voice, “…voice quality is fundamentally per-ceptual in nature” (Kreiman, Gerratt, Kempster, Erman, & Berke, 1993, p. 21). The auditory-per-ceptual aspects of a voice represent a psychologi-cal reality for both the singer and the listener and have the potential to provide a common terminol-ogy for ease of communication between singers, their teachers, and clinicians. Development of a reliable and valid method of communicating one’s perception has significant implications. For ex-ample, audience appreciation ensures continuing engagement of performers; expert listeners adju-dicate examinations, auditions and other competi-tions, acts which may seal the fate of performing artists. Because such evaluations can have signifi-cant impacts on musical careers, it is important to establish whether listeners are reliable in their judgments, both in comparison to other listeners and across time.
Research indicates that listeners show some de-gree of reliability and consistency in their percep-tual judgments of timbre, between vocal genres (from opera to music theatre) (Sundberg, Gram-ming, & Lovetri, 1993), and between good and poor vocal and instrumental performance (Ekholm et al., 1998; Geringer & Madsen, 1998; Saunders & Holahan, 1997; Wapnick, Flowers, Alegant, & Jasinskas, 1993), in the assessment of excellence in overall voice quality (Wapnick & Ekholm, 1997) and in rankings and ratings of performers in competitive situations (Davidson & Da Costa Coimbra, 2001).
95
Perceptual identification of open throat 3Musicae Scientiae in press
Recent evidence indicates that judges initially assess the performance as a whole before ‘justi-fying’ their ranking by assessing specific criteria (Stanley, Brooker, & Gilbert, 2002). Judges do acknowledge, however, that providing them with specific criteria focuses their assessment on certain elements of performance and that this provides the basis for useful, specific feedback to students.
Perceptual studies of vocal quality generally re-quire listeners to focus on specific vocal dimen-sions, rather than, or as well as, making an overall judgement. Focusing listeners by using a number of rating criteria improves the consistency of judges’ responses (Wapnick et al., 1993). However, analy-sis of available rating scales that request listeners to make several judgments of vocal quality reveal that these specific dimensions may all be tapping into a single underlying construct, that of vocal quality, thereby rendering individual assessments on each dimension at least partially redundant. For example, a number of researchers (Ekholm et al., 1998; Robison et al., 1994; Wapnick & Ekholm, 1997) have found very high correlations between all dimensions of voice quality studied, (for ex-ample, “appropriate vibrato”, “resonance/ring”, “color/warmth”, and “clarity/focus”). All of these dimensions were found to converge with the over-all judgement of vocal quality.
There are, of course, other limits to the accu-racy of perceptual judgements. Perceptual stud-ies produce an unnatural assessment setting for teachers and assessors of singing. Listeners have commented on their difficulty in assessing sing-ing when the duration of the sample is insufficient (Ekholm et al., 1998). In forced choice situations in which listeners are asked to make dichotomous choices, listeners can reliably make gross judg-ments such as identifying girl from boy singers (Howard, Szymanski, & Welch, 2002). However, listeners could not identify singers from speakers during speaking rather than singing tasks (Brown, Rothman, Morris, & Sapienza, 2001; Brown, Rothman, & Sapienza, 2000), or the same sing-ers singing at different pitches throughout their singing range (Erickson, 2003; Erickson & Perry, 2003). This lack of identification may have been due to the variability in the stimuli rather than a lack of sensitivity in the listeners, and these issues are addressed in the design of this study.
Perceptual identification of open throat
To date, there has been no research that assesses the sound qualities associated with particular vo-cal techniques. In Western classical singing, the
focus remains on the overall good sound quality (Davidson & Da Costa Coimbra, 2001) whereas pedagogical literature focuses on mastery of a wide variety of techniques to achieve this final sound (Miller, 1998; Stark, 1999).
In this study, we investigated the perceptual sig-nificance of ‘open throat’ technique by compar-ing the same female classical singers when they consciously used maximum open throat technique and reduced open throat technique in their singing of messa di voce (a crescendo-diminuendo on a single note of long duration), a classical song and a romantic lied. The aim of the study was to assess expert pedagogues’ perception of OT technique in singing, to explore its acoustic correlates, and to assess the degree of agreement between percep-tual and acoustic factors.
METHOD
Participants
ListenersListeners were 15 experienced singing peda-
gogues, 12 females and 3 males aged between 37 and 76 with a mean of 54 years. Six partici-pants had a postgraduate qualification in sing-ing, five had a diploma of music or singing, and three a bachelor degree in music. One cited exten-sive international performing experience as their qualification in singing. All had taught singing for 4 to 40 years, with an average for 20 years. Thirteen of fifteen participants taught singers in a Conservatorium of Music. Overall, participants’ singing studios comprised an average of 39.5% of operatic students and 36.7% classical students. For 11 pedagogues, the majority of their studio comprised these two genres. Six pedagogues also taught ≥20% of musical theatre students in their studio. Eleven taught a proportion of international and national level singers, nine at big city or re-gional/touring and eight at local community level. Participants were either known to the researchers via affiliations with key music centres in Australia or volunteered in response to an advertisement in a national singing organization newsletter. Partici-pants were sent information about the project and were invited to take part in a perceptual study of singing technique. They were required to partici-pate in a single listening session at a time and lo-cation convenient to them.
Prior to commencement, participants completed a questionnaire regarding their musical and teach-
96
Perceptual identification of open throat 4Musicae Scientiae in press
ing experience and current singing studio. Each participant was asked for demographic informa-tion and to classify their singing studio according to their students’ performing levels and singing styles (Bunch & Chapman, 2000).
SingersSix female singers, three sopranos and three
mezzo-sopranos volunteered to participate in this project. They were advanced students with excel-lent technique studying with an experienced singing pedagogue, a Lecturer in Vocal Studies and Opera at a state Conservatorium of Music in Australia. This institution is considered the premier institu-tion for musical education in the country, having produced several singers of international repute. Criteria for participant selection included, through this pedagogue’s assessment, singers who: (1) had a good classical singing technique for their level of training and experience; and (2) understood and demonstrated skilful control of ‘open throat’ or ‘retraction’ techniques in their singing. Singers were sent information about the project and were required to attend a single recording session. They were told that the object of the study was to inves-tigate acoustical and perceptual features of the use of open throat in singing.
Prior to the voice recording, participants com-pleted a questionnaire. Singer participants were aged between 23 and 30, with a mean of 26 years. All had studied singing for at least 7 years (aver-age 9.8 years) and had spent an average of 5 years studying with their present singing teacher. Each singer held a qualification in singing or music (four had Bachelor of Music degrees and two had diplomas, in music and/or singing) and 5 of 6 were currently undertaking a second degree in singing (3 postgraduate Diploma of Opera and 2 Bachelor of Music degrees). All defined the majority of their singing as operatic (>50%), with the second most common style classical (>20%) (Bunch & Chap-man, 2000). All reported that they were in good health and able to perform the tasks.
Singer Protocol
The Musical TasksThree musical tasks were chosen in order to test
different demands of good singing, but were not musically difficult. They were designed to test the use of open throat, and contained musical features derived from a previous qualitative study on the use of the technique (Mitchell et al., 2003), where use or lack of the technique was deemed to be particularly valuable or noticeable. These features
were: high tessitura, sustained or legato singing, dynamic range control, and vocal agility.
The messa di voce, a crescendo-diminuendo on a single note is a tool for accomplishing and maintaining uniform timbre throughout the sing-ing range (Miller, 1996). The singers sang three messe di voce across their passagio beginning with pitches determined by their voice type, soprano or mezzo soprano (A4, G4, G-flat4).
The Mozart song Ridente la Calma, K 152, bars 1-27 (Figure 1a) was selected as it is a nominally simple song in the Italian language (Wapnick & Ekholm, 1997) with a mixture of common musi-cal statements involving repeated legato lines as well as the initial stylised leaps of a major 4, and short scale figures. All 6 singers sang this aria in the same key (F major).
The third verse of the Schubert lied, Du bist die Ruh D. 776 (Op. 59, No. 3) (Figure 1b), bars 54 to 80 was chosen for its demanding vocal control, sustained musical line and high climactic tessitu-ra. The three sopranos and 3 mezzo-sopranos sang this in an appropriate key depending on voice type (E-flat, D-flat and C major).
Experimental conditionsSingers sang each of the two song excerpts un-
der three conditions: optimal, sub-optimal and loud sub-optimal. ‘Optimal’ (O) provided the best possible sound quality the singer could produce using their best open throat technique. ‘Sub-opti-mal’ (SO) condition involved the use of a reduced (open throat) technique but still with an acceptable singing technique and without consciously alter-ing any other aspect of their technique. The loud
Figure 1a: First 7 bars used as perceptual stimuli of Mozart song Ridente la Calma, K 152.
&b8
3‰ ‰
R
œ .œ
œ j
œ
J
œ .œœ
J
œ
Ù œœœ œœ œ œ œ œ
œ œ
œ œ
J
Ϩ
œœœ
&b
6
œ œ œ œ œœ
r
œ
r
œ
‰ ‰ ∑ ‰ ‰ ≈
R
œ .œœ
j
œ ≈
R
œ .œ
œ
j
œ≈R
œ
≈
Ri - den - te la cal - ma, nell' al - ma si - desti, nell'
al - ma si desti.
&b
b
b 8
3
J
œ .
J
œ
R
œœb
J
œ œb
J
œ œ
J
œ œ
J
œ
.œ œœ j
œ œ œœ œn œ œ
&b
b
b
6.œ
.œ
Dies Au - gen zelt, Von dei - nem Glanz Al - lein er -
hellt.
Figure 1b: Bars 54 to 60 used as perceptual stimuli of the Schubert lied, Du bist die Ruh D. 776 (Op. 59, No. 3).
97
Perceptual identification of open throat 5Musicae Scientiae in press
sub-optimal condition provided additional acous-tic information discussed in earlier papers (Kenny & Mitchell, in press; Mitchell & Kenny, 2004a, 2004b).
Each task was performed twice in the O and SO conditions. On completion of each O and SO con-dition in each task, participants were asked to self-rate their singing on a 10 cm visual analogue scale (VAS) for ‘openness’ and ‘my best singing’. Each singer completed six of these descriptive self-rat-ing scales.
Reliability check
A pedagogue was present during the recording sessions to provide accompaniment for warm-ups and practice of the tasks where necessary. Prior to the commencement of the study, the students and the pedagogue had several practice sessions in which the singer was instructed to use either O or SO at random. The pedagogue indicated which version she thought the singer had used. Practice with each singer continued until both the singer and the pedagogue reached 100% agreement on the occasions that the technique was applied or reduced.
Recording
The voice was recorded using a high-quality mi-crophone (AKG C-477) positioned on a head boom a constant 7 cm distance from the singer’s lips. This ensured that direct energy of the performers’ voices was recorded rather than room reflections, enabling us to use a studio environment with low ambient noise rather than an anechoic studio (Ca-brera, Davis, Barnes, Jacobs, & Bell, 2002). The signal was then amplified using a Behringer Ul-tragain preamplier and digitally recorded to a CD recorder (Marantz CDR 630).
Calibration was carried out in each recording by playing pink noise samples immediately follow-ing each recording session at the same recording gain used for recording the singer’s voice. For cal-ibration of absolute sound pressure levels (SPL), a sound level meter (Rion NL-06 SPL) was placed adjacent to the AKG microphone 7 cm from a speaker (Bose Lifestyle) from which the pink noise was played. The SPL shown on the sound level meter was noted for the pink noise signal and used later for calibration.
Pink noise enables calibration of an audio play-back system across the frequency range for tasks such as perceptual testing and comparative analy-sis on a computer. Pink noise is at least as good as
any other steady-state known signal. A pure sine wave tone is more susceptible to interference in the environment, and hence would give less re-peatable results.
This data set was used as a basis for perceptual and acoustic evaluations.
Perceptual Test
StimuliThe audio recordings of the singers were digit-
ally extracted from CD using Audiograbber soft-ware (www.audiograbber.com) on a standard PC computer to wav audio format in stereo at 41000 Hz and 16 bit sample rate. This methodology en-sures as close as possible that the original recorded sound was played back to the judges (no filtering or normalization was applied to minimize the ef-fects of digital artifacts).
The recorded pink noise for each singer was used to equalize the peak levels of each sample to ensure that relative SPL for each singer was the same. The amplification tool in Cool Edit calcu-lated the SPL necessary to equalize each sample’s peak SPL, thus making each recording relative to the level (in dB) of the other samples produced on the day.
The files were then edited in Cool Edit Pro 1.2 (www.adobe.com) and final samples saved to CD.
ProcedureThe perceptual test was conducted in a quiet en-
vironment and samples were played on CD from a Sony CD Walkman (DEJ885W) via circum-aural closed-back stereo monitoring headphones (Sen-nheiser HD 270). This enabled the study to be con-ducted in participants’ singing studios. This meth-odology was favoured as optimal sound quality, rather than using a computer sound card (Erick-son & Perry, 2003) or sending tapes to participants (Ekholm et al., 1998; Wapnick & Ekholm, 1997). Prior to presentation of stimuli, participants were given information on the two singing conditions, O and SO, based on qualitative results in a previ-ous study (Mitchell et al., 2003). Pedagogues were instructed to assess each singing sample for use of open throat technique. In the first study, peda-gogues were presented with single notes in messa di voce and in the second study, extended singing samples in both O and SO conditions.
Single notes as one of the test stimuli were used to assess the timbral qualities of open throat technique without potential distractors such as
98
Perceptual identification of open throat 6Musicae Scientiae in press
intonation (Geringer & Madsen, 1998) or diction (Wapnick & Ekholm, 1997). The term ‘timbre’ has attracted different definitions over time. Tradition-ally, ‘timbre’ was defined as the perceptual differ-ence between two sounds which are equal in pitch and loudness (ANSI, 1973). Following this defini-tion, musical ‘timbre’ has been investigated as the difference between pairs of sounds rather than the ‘timbre’ of a single sound. More recently, ‘timbre’ has been defined as overall sound quality (Handel & Erickson, 2004). Since the aim of this study was to assess the perceived overall sound quality when a pedagogical technique is applied, we assessed the messa di voce responses individually.
Study 1: Messa di voce
For the messa di voce pairs, participants were asked to assess timbre in each sung note. Each pair contained a combination of O and/or SO by the same singer. There were four possible paired com-binations: O/O, O/SO, SO/O and SO/SO. Each listener had a trial session with four sample pairs, O/O, O/SO, SO/O, SO/SO before commencing the study. There were 720 samples of messa di voce in 360 pairs.
Study 2: Song samples
In the second study, participants were presented with individual samples of Mozart or Schubert. They were presented with the musical score of each musical task and asked to assess each sample for use of open throat technique. Each listener had a trial session with four samples, 2 O, 2 SO for each of the musical tasks (Mozart and Schubert) before starting the second perceptual test. Each of the 15 judges responded to 24 pairs of singing samples plus six repeats.
Study design
Assessment of inter-rater agreement between judges was calculated using Cohen’s Kappa (Siegel & Castellan, 1988). Kappa varies between -1 and 1, where 1 is perfect agreement and 0 is agreement not better than expected by chance. Although its use in medical settings has been criticised because it does not adjust for bias (the difference in rated prevalence of a disorder between judges), this is not a problem in a controlled experimental design where the prevalence of the two conditions being rated is identical.
As a test statistic, Kappa can verify that agree-ment exceeds chance levels. It is an omnibus in-dex of agreement and does not make distinctions among various types and sources of disagreement.
Further, Kappa is influenced by distributions and base-rates. As a result, Kappas are seldom com-parable across studies, procedures, or populations. Thus, attempts to categorize ranges of kappa as “good”, “fair”, or “poor” may be inappropriate. Krippendorf (1980) advocates agreement of at least .70.
Altman (1991) argued that any value of Ka-ppa below .50 would indicate “poor” agreement. Fleiss (1981) recommends the following Kappa benchmarks: < .40 = poor agreement; 0.40-0.75 = intermediate to good agreement; > .75 = excel-lent agreement. Gardner (1995) argues that Kappa should exceed .70 before one proceeds with ad-ditional data analyses. McGinn, Wyer, Newman, Leipzig & Guyatt, (2004) state that a safe rule of thumb is that .5 indicates moderate agreement, and anything lower is considered fair to poor agree-ment.
RESULTS
MESSA DI VOCE
In the first analysis, judges’ individual ratings of samples as either O or SO were assessed. Table 1 shows the results of these ratings.
A number of tests were conducted to determine the reliability of the judges in their ratings of the samples (Cicchetti & Feinstein, 1990). All confi-dence intervals were set at 95%. All indices dem-onstrated high levels of accuracy in the ratings of these judges. The measure of efficiency (ie correct overall classification rate) was 0.8597 (95% CI: 0.8322 - 0.8843); sensitivity (ie the proportion of true positives rated as positive) = 0.8111 (95% CI: 0.7668 - 0.8502) and specificity (ie the number of true negatives rated as negative) = 0.9083 (95% CI: 0.8737 - 0.9361). The chance levels of effi-ciency were 50%; of sensitivity, 45.14% and of specificity, 54.86%. The overall misclassification rate was 0.1403 (95% CI: 0.1157 - 0.2094); the false positive rate = 0.0917 (95% CI: 0.0639 - 0.1263); and the false negative rate = 0.1889 (95% CI: 0.1498 - 0.2332).
Cohen’s Kappa was calculated to test the hy-pothesis that Kappa = 0. The hypothesis was not confirmed (z=19.40, p =0.0000. Cohen’s Kappa = 0.7194 (95% CI: 0.6690 - 0.7699), indicating agreement substantially above chance among this sample of judges on this perceptual test.
The reliability of the best and worst judges were
99
Perceptual identification of open throat 7Musicae Scientiae in press
tested individually. For the best judge, the measure of efficiency = 0.9792 (95% CI: 0.8893 - 0.9995); sensitivity = 0.9583 (95% CI: 0.7888 - 0.9989) and specificity = 1.00 (85%); overall misclassifi-cation rate = 0.0208 (95% CI: 0.0005 - 0.1398); and false negative rate = 0.0417 (95% CI: 0.0011 - 0.2112). Cohen’s Kappa = 0.9583 (95% CI: 0.8776 - 1.0391) indicated almost perfect agreement in the best judge’s responses. (z=6.65, p =0.0000)
For the worst judge, the measure of efficiency = 0.6458 (95% CI: 0.4946 - 0.7784); sensitivity = 0.5833 (95% CI: 0.3664 - 0.7789) and specificity = 0.7083 (95% CI: 0.4891 - 0.8738). The overall misclassification rate = 0.3542 (95% CI: 0.2216 - 0.5389); the false positive rate = 0.2917 (95% CI: 0.1262 - 0.5109); and the false negative rate = 0.4167 (95% CI: 0.2211 - 0.6336). Cohen’s Kappa = 0.2917 (95% CI: 0.0232 - 0.5601) indicated poor agreement in the worst judge’s responses (z=2.04, p =0.0417). These data are presented in Table 2.
Song samples
In the individual song samples, listeners identi-fied the use of open throat in 84% of O samples, and 69% of SO (Table 3). They were more likely to make a correct identification in the Mozart task (85% correct) than in the Schubert task (68%). Listeners identified > 83% of Mozart O and SO, and Schubert O. They were least reliable in judg-ing Schubert SO (53%).
In 180 song samples, for O, judges rated 80 of 90 correctly in Schubert and 70 of 90 correctly in Mozart, that is, judges were less consistent in rat-ing the Mozart task (k=.3) than the Schubert task (k=.47) in both O and SO.
Reliability of judges
The 15 judges identified the experimental con-dition in 372 of 450 samples (82.7%). There is a 95% chance that this sample comes from a popu-lation where the true proportion comes from either side of 0.827. Kappa analysis was performed to assess listener consistency. Twelve of fifteen judg-es were moderately consistent in their judgments (k≥0.600); judges 1, 5 and 7 were inconsistent in their judgements. These data are presented in Ta-ble 4.
Individual singers
Individual singers’ O samples, in both Mozart and Schubert were more readily identified by lis-teners than SO. Figure 2a-d presents these data graphically, showing the correct identifications and misclassifications of O and SO for each sing-er in each musical task. Four of six singers (1-4) were consistently identified correctly as singing with open throat in O and less open throat in SO. Singers 5 and 6 were outliers. Judges incorrectly judged Singer 6’s intentions in Mozart O and SO
All judgments
Sample O SO
O 81.11% 18.89%
(89.85%) (17.22%) 50.00%
SO 9.17% 90.83%
(10.15%) (82.78%) 50.00%
(45.14%) (54.86%) N = 720
Table 1: Messa di voce ratings. Percentages of all correctly and incorrectly identified samples by Overall percentages of correctly and incorrectly judged samples. Top row represents row percent-ages and bottom row in parentheses represents col-umn ratings.
Best Judge (3) Worst Judge (1)
O SO O SO
O 23 1 14 10
(95.8%) (4.2%) (58.3%) (41.7%)
SO 0 24 7 17
(0.0%) (100.0%) (29.2%) (70.8%)
Table 2: Messa di voce responses of the best and worst judges. The top row represents the total number of judgments from 24 responses to O and 24 to SO and the bottom row represents the corre-sponding percentages of correctly and incorrectly identified samples.. Task Hit % Miss %
All O 84% 16%
All SO 69% 31%
Mozart O 86% 14%
Mozart SO 84% 16%
Schubert O 83% 17%
Schubert SO 53% 47%
Table 3: Song responses. Percentages of correct (hit) and incorrect (miss) responses to the song task, by condition (optimal and sub-optimal) and by condition and task (Mozart and Schubert).
100
Perceptual identification of open throat 8Musicae Scientiae in press
and identified her O as SO and SO as O in over half of the 30 judgements. Singer 5 was incorrect-ly identified in Schubert SO only, that is, judges viewed her singing in SO as O.
Singers’ self-perceptions of their sound
Overall means of singers’ VAS results indicated that singers associated their performance of the O condition with a high degree of openness and strongly representative of their best singing (Fig-ure 3). Subject 6 was dissatisfied with all of her singing on the test day and cast “informal votes” when asked to rate her singing. Her responses were omitted from the final result.
These responses are in accordance with the sing-ers’ intentions in their performance in each con-dition. Pearson r varied between the responses to best singing and openness according to condition and task. The O condition responses were moder-ately correlated in the Mozart task (r=0.45), but highly correlated in the Schubert (r=0.84) but nei-ther was statistically significant (p>0.05). In the
SO condition, responses were both very highly correlated (r=0.98, r=0.94), and significantly dif-ferent from zero (p<0.01, p<0.05). At least 90% of the singers’ perception of good singing can be attributed to their perception of openness.
DISCUSSION
This study confirmed that the use of open throat technique is a perceptual reality to singers and that it produces a specific vocal quality in classi-cal singing that can be reliably identified by expert listeners. While previous acoustic studies proved inconsistent in differentiating between experimen-tal conditions O and SO, the human ear produced the most reliable assessment of vocal quality.
Messa di voce
Collectively, judges were accurate in their iden-tification of O and SO and in each messa di voce sample, the majority of listeners correctly recog-nised the use of open throat when it occurred. Re-liability of listeners’ judgements ranged from fair to highly consistent. The high proportion of cor-rect judgements of O and SO suggests that these singers produced a specific sound quality or tim-bre that expert pedagogues could reliably differen-tiate, thus strengthening the claim that open throat technique is associated with a particular desirable vocal quality.
In a previous study, expert pedagogues confirmed their collective understanding of the technique of open throat and its use as an essential pedagogical instruction in the singing studio (Mitchell et al., 2003). This study confirmed their assertions in the previous study that open throat technique produc-es a specific audible quality, readily identifiable by vocal pedagogues.
Perceptually, we considered it an important methodological advance to assess open throat tech-nique in single notes (Erickson, 2003; Erickson & Perry, 2003) before progressing to extended sing-ing samples (Ekholm et al., 1998; Howard et al., 2002; Robison et al., 1994; Wapnick & Ekholm, 1997). While extended singing samples are more representative of listeners’ and pedagogues’ expe-riences in the singing studio and in singing assess-ment, testing timbre of single notes removes any influence of intonation, diction and musical stim-uli on judgements and focuses listeners directly to core sound quality. Proficiency in the messa di voce is a key indicator of vocal mastery (Miller, 1996) and these samples provided listeners with
Listener Kappa Asymp. Std. Error(a)
Approx. T(b)
Approx. Sig.
1 0.400 0.167 2.196 0.028
2 0.867 0.091 4.747 0.000
3 0.733 0.123 4.053 0.000
4 0.800 0.109 4.392 0.000
5 0.467 0.161 2.556 0.011
6 0.867 0.090 4.790 0.000
7 0.200 0.179 1.095 0.273
8 0.600 0.145 3.316 0.001
9 0.733 0.124 4.017 0.000
10 0.667 0.133 3.727 0.000
11 0.733 0.120 4.168 0.000
12 0.667 0.133 3.727 0.000
13 0.667 0.136 3.660 0.000
14 0.600 0.134 3.586 0.000
15 0.800 0.107 4.472 0.000
Table 4: Reliability of judges’ responses. Cohen’s Kappa, Asymptomatic standard error (not assum-ing the null hypothesis), asymptotic standard error (assuming the null hypothesis) and approximate significance for listeners 1-15.
101
Perceptual identification of open throat 9Musicae Scientiae in press
sufficient vocal information to respond to timbre. As the definition of ‘timbre’ is generally accept-
ed as overall or intrinsic sound quality (Handel & Erickson, 2004), it was less important to the aims of this part of our study to report the difference between one sound and another (ANSI, 1973).
Song samples
Twelve of fifteen expert listeners consistently identified the use of open throat technique in O and a reduction of the technique in SO in the song samples. Listeners demonstrated reliability in their judgements through the duration of the song task, repeating their judgement of O or SO in the
0
5
10
15
1 2 3 4 5 6
0
5
10
15
1 2 3 4 5 6
0
5
10
15
1 2 3 4 5 6
0
5
10
15
1 2 3 4 5 6
Mozart OptimalO hitO miss
Schubert Sub-OptimalSO hitSO miss
Schubert OptimalO hitO miss
Mozart Sub-OptimalSO hitSO miss
Tota
lres
po
nse
s
Singer
Singer
Singer
Singer
Tota
lres
po
nse
sTo
talr
esp
on
ses
Tota
lres
po
nse
s
b
c
d
a
Figure 2 Correct identifications and misclassifications for each singer (1-6) in (a) Mozart optimal (O); (b) Mozart sub-optimal (SO); (c) Schubert optimal and (d) Schubert sub-optimal.
102
Perceptual identification of open throat 10Musicae Scientiae in press
six repeated samples in the perceptual test. These results support previous research that indicates high intra-judge agreement (Davidson & Da Costa Coimbra, 2001; Ekholm et al., 1998) in assessing overall vocal quality.
However, when Mitchell and Kenny assessed the use of open throat in these singers through acoustic characteristics of vibrato (Mitchell & Kenny, 2004b) and energy distribution in LTAS (Mitchell & Kenny, 2004b), their findings did not accord with experts’ perceptual evaluation (Ken-ny, Davis, & Oates, in press). While the vibrato analysis revealed significant reductions in vibrato extent and increases in onset time in SO compared to O, conventional measures of singing power ra-tio (SPR) (the difference between the peak height) and energy ratio (ER) (the difference between peak area) on LTAS of O and SO were not statistically significantly different (Mitchell & Kenny, 2004b). Changes in vibrato rate and extent, inappropriate to genre are considered indicative of poor sing-ing so it was hypothesized that testing the energy distribution in these singers’ voices in each condi-tion would identify the timbral changes associated with open throat. LTAS results were not consistent with the vibrato findings for O/SO, and SPR and ER were not sensitive to the timbral differences between O/SO identified perceptually by expert
listeners in this study.We examined our hypotheses using both extend-
ed singing samples as well as single notes (as in the messa di voce) because eachmusical task tests a different demand of good singing and the former more closely resembles singing as heard and as-sessed in the studio and in formal assessment situations (Davidson & Da Costa Coimbra, 2001; Ekholm et al., 1998; Howard et al., 2002; Robison et al., 1994; Wapnick & Ekholm, 1997).
An interesting and somewhat unexpected finding in our study was that listeners were more likely to identify the correct experimental condition in the Mozart song sample rather than Schubert. Previ-ous perceptual studies have recommended using the same musical stimuli to reduce inter-subject variables that may influence responses (Howard et al., 2002; Saunders & Holahan, 1997); yet, these findings indicated an effect of musical task. In-deed, as open throat is of critical importance in high sustained singing (Mitchell et al., 2003), the scale motif and high tessitura of the Schubert task may have proved too difficult for singers to truly reduce open throat in SO. Conversely, the Mozart task was less vocally demanding, opening up the possibility that singers require a less robust tech-nique to make the desired sound.
When Mitchell and Kenny (2004a) assessed these singers’ vocal qualities in O and SO through the energy distribution in LTAS, the effect of mu-sical task approached significance, that is, both SPR and ER were slightly greater in Schubert than in Mozart. According to current singing literature, lower ER and SPR should indicate improved sing-ing quality, by boosted energy above 2 kHz (Omori et al., 1996; Thorpe et al., 2001). A future study by these authors will examine the significance of SPR and ER rankings with perceptual judgment of overall quality (Kenny & Mitchell, in press).
Examination of correct judgements by singer in-dicated singers 1-4 were correctly identified more often than singers 5 and 6. In testing the sound quality of open throat by singer, singers 1-4 were consistently identified correctly in O and SO. Singer 5 and 6 were outliers in certain conditions and tasks. Singer 6, in the Mozart task, produced the most inconsistent responses amongst the judg-es. Her O and SO were incorrectly identified in over 50% of occurrences, suggesting there was little difference between her O and SO, and that judges collectively were unclear whether she was attempting to use maximum open throat. On the other hand, listeners had little trouble identifying
0
10
20
30
40
50
60
70
80
90
100
Mozart Schubert
OOpenness
OBestS
inging
SOOpenness
SOBestS
inging
OOpenness
OBestS
inging
SOOpenness
SOBestS
inging
VASof100m
m
Figure 3: Overall means of singers’ VAS scores from 1-100, for singers 1-5 (singer 6 omitted) to report the degree of openness and representation of her best singing for each musical task (Mozart and Schubert) in each condition (optimal and sub-optimal).
103
Perceptual identification of open throat 11Musicae Scientiae in press
singer 5’s Schubert O, but were misled by her SO, judging it as O in 8 of 15 cases. The overall high quality of singer 5’s voice may have misled judges to rate her SO as O.
For an accomplished singer, it may be possible to sing using reduced open throat technique but still produce an aesthetically pleasing sound, or equally, use an adequate open throat technique but produce an aesthetically displeasing sound as in the case of singer 6. Clearly, there are many fac-tors that converge in producing an aesthetically pleasing vocal quality, of which open throat is but one. Listeners have clearly attended to multiple dimensions of the vocal production in arriving at their perceptual judgments of the singers. In fact, singers 5 and 6 were the highest and lowest scor-ing singers respectively in a study of overall vo-cal quality (Kenny & Mitchell, in press) but these rankings showed no relationship with correspond-ing rankings of SPR and ER, which are intended as objective measures of voice or voice quality.
Singers in this study were acutely aware of the impact of open throat technique on their overall singing sound. Singers 1-5 equated good singing with use of maximal open throat technique and felt their overall singing sound decreased when they reduced the technique. Matching a perform-er’s intention to a pedagogical concept is valuable to singing pedagogy. Singer 6 was dissatisfied with her singing on the day of recording, which accorded with listeners’ judgements of her vocal quality and the similarities between her O and SO. It may be, as one listener pointed out, that it is ‘hard to tell if it is a good singer singing badly, or a bad singer singing well’.
Sophisticated acoustic measurement technolo-gies are now being routinely applied to voice sci-ence and vocal pedagogy the goal of which is to provide objective measurement of voice. To date, studies that attempt to define a good voice per-ceptually and acoustically (Ekholm et al., 1998; Wapnick & Ekholm, 1997) or assess vocal mas-tery throughout musical education (Mendes et al., 2003; Vurma & Ross, 2000) by linking perceptual ratings with acoustic measurements have not as-sessed the relationship between vocal mastery and the development of specific technical proficiencies. If singers’ and expert listeners’ reports of sing-ing accord, focussing research on specific sing-ing techniques could enhance singing pedagogy. Matching optimal singing quality with performer technique or intention enhances communication between singers and teachers. These findings will
be strengthened by replication in male voices and professional singers.
Current work in singing has not sufficiently in-corporated perceptual ratings and descriptions of sound quality. Vocal quality results from a com-plex combination of acoustical parameters that in-teract. To date, no single objective evaluation cap-tures or characterizes vocal quality in a systematic way (Omori et al., 1996; Thorpe et al., 2001). The future of experimental studies in the teaching of singing depends on their incorporation into a the-ory of the singing voice and on their relation to current practice.
Most singers and teachers of singing understand vocal functioning and control through audiation, sensory feedback and mental imagery. Teachers have a mental image of the particular sound indi-vidual students are capable of producing and use verbal explanation, vocal modeling and feedback to teach this. Despite the fact that teachers and as-sessors of singing process sound using sensory and auditory methods as their primary feedback (Kitch & Oates, 1994), few perceptual studies use these professional skills in assessing vocal qual-ity. If expert listeners can reliably identify a tech-nique, such as open throat, in the overall singing sound, future perceptual and acoustic studies of voice may provide more useful information for the dissemination of current vocal practices, their as-sessment and documentation in singing pedagogy. It is now feasible to track the methods by which good singers are trained. In linking the use of spe-cific pedagogical techniques with perceptual judg-ment and acoustic measurement, future studies in voice can provide a basis for a more systematic pedagogy of the singing voice.
ACKNOWLEDGEMENTS
We thank Ms Maree Ryan for her pedagogical advice and support, Mr Peter Thomas and Dr Den-sil Cabrera for their advice on acoustical matters, the six singers for their willing participation in the project, and the reviewers for helpful comments on an earlier draft of the manuscript.
REFERENCES
Altman DG (1991). Practical Statistics for Medical Re-search. Chapman and Hall.
ANSI. (1973). Psychoacoustical terminology: S3.20. New York: American National Standards Institute.
Brown, W. S., Jr, Rothman, H. B., Morris, R. J., & Sa-
104
Perceptual identification of open throat 12Musicae Scientiae in press
pienza, C. M. (2001). Acoustic analyses of trained singers perceptually identified from speaking sam-ples. Journal of Voice, 15(1), p. 25-35.
Brown, W. S., Jr., Rothman, H. B., & Sapienza, C. M. (2000). Perceptual and acoustic study of profession-ally trained versus untrained voices. Journal of Voice, 14(3), 301-309.
Bunch, M., & Chapman, J. (2000). Taxonomy of sing-ers used as subjects in scientific research. Journal of Voice, 14(3), 363-369.
Burgin, J. C. (1973). Teaching singing. Metuchen, N.J.: Scarecrow Press.
Cabrera, D., Davis, P., Barnes, J., Jacobs, M., & Bell, D. (2002). Recording the operatic voice for acoustic analysis. Acoustics Australia, 30, 103-108.
Callaghan, J. (2000). Singing and voice science. San Diego, California: Singular Pub. Group.
Cicchetti, D. V., & Feinstein, A. R. (1990). High agree-ment but low kappa: II. Resolving the paradoxes. J. Clin. Epidemiol, 43, 551-558.
Davidson, J. W., & Da Costa Coimbra, D. (2001). In-vestigating performance evaluation by assessors of singers in a music college setting. Musicae Scientiae, 5(1), 33-53.
Ekholm, E., Papagiannis, G. C., & Chagnon, F. P. (1998). Relating objective measurements to expert evaluation of voice quality in Western classical sing-ing: critical perceptual parameters. Journal of Voice, 12(2), 182-196.
Erickson, M. L. (2003). Dissimilarity and the Clas-sification of Female Singing Voices: A Preliminary Study. Journal of Voice, 17(2), 195-206.
Erickson, M. L., & Perry, S. R. (2003). Can listeners hear who is singing? A comparison of three-note and six-note discrimination tasks. Journal of Voice, 17(3), 353-369.
Fields, V. A. (1947). Training the singing voice: an analysis of the working concepts contained in recent contributions to vocal pedagogy. New York: King’s Crown Press.
Fleiss, JL. (1995) Statistical Methods for Rates and Proportions (2nd edition). New York: Wiley 1981.
Foulds-Elliott, S., Thorpe, C., Cala, S., & Davis, P. (2000). Respiratory function in operatic singing: ef-fects of emotional connection. Logopedics Phoniat-rics Vocology, 25(4), 151-168.
Gardner, W. (1995). On the reliability of sequential data: measurement, meaning, and correction. In John M. Gottman (Ed.), The analysis of change. Mahwah, N.J.: Erlbaum.
Geringer, J. M., & Madsen, C. K. (1998). Musicians’ ratings of good versus bad vocal and string perform-ances. Journal of Research in Music Education, 46(4), 522-534.
Handel, S., & Erickson, M. L. (2004). Sound Source Identification: The Possible Role of Timbre Transfor-
mations. Music Perception, 21(4), 587-610.Hemsley, T. (1998). Singing and imagination: a human
approach to a great musical tradition. Oxford; New York: Oxford University Press.
Herbert-Caesari, E. F. (1951). The voice of the mind. London: Robert Hale.
Howard, D. M., Szymanski, J., & Welch, G. F. (2002). Listeners’ perception of English girl and boy choris-ters. Music Perception, 20(1), 35-49.
Joiner, J. R. (1998). Charles Amable Battaille: pioneer in vocal science and the teaching of singing. Lanham, Md.: Scarecrow Press.
Kenny, D.T., Davis, P. & Oates, J. (2004). Music per-formance anxiety and occupational stress amongst opera chorus artists and their relationship with state and trait anxiety and perfectionism. Journal of Anxi-ety Disorders, 18, 6, 757-777.
.Kenny, D. T., & Mitchell, H. F. (in press). Acoustic and perceptual appraisal of vocal gestures in the female classical voice. Journal of Voice.
Kitch, J. A., & Oates, J. (1994). The perceptual features of vocal fatigue as self-reported by a group of actors and singers. Journal of Voice, 8(3), 207-214.
Kreiman, J., Gerratt, B. R., Kempster, G. B., Erman, A., & Berke, G. S. (1993). Perceptual evaluation of voice quality: review, tutorial, and a framework for future research. Journal of Speech & Hearing Research., 36(1), 21-40.
Krippendorf, K. (1980). Content Analysis: An Intro-duction to its Methodology. Sage Publications, Thou-sand Oaks, CA.
Landis, J., & Koch, G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159-174.
Manèn, L. (1987). Bel canto: the teaching of the clas-sical Italian song-schools its decline and restoration. Oxford: Oxford University Press.
Marafioti, P. M. (1981). Caruso’s Method of Voice Pro-duction. New York: Dover Publications.
McKinney, J. C. (1982). The Diagnosis and Correction of Vocal Faults. Nashville, Tennessee: Broadman Press.
McGinn, P. T., Wyer, C., Newman, T. B., Leipzig, R., & Guyatt, G. (2004). Tips for learners of evidence-based medicine: Measures of observer variability (kappa statistic). Canadian Medical Association Jour-nal, 171(11), 1369-1373.
Mendes, A. P., Rothman, H. B., Sapienza, C. M., & Brown, W. S., Jr. (2003). Effects of vocal training on the acoustic parameters of the singing voice. Journal of Voice, 17(4), 529-543.
Miller, R. (1996). The structure of singing: system and art in vocal technique. New York: London: Schirmer Books.
Miller, R. (1997a). The Open Throat (La gola aperta).
105
Perceptual identification of open throat 13Musicae Scientiae in press
In Miller, R. (Ed.), On the Art of Singing (pp. 17-19). New York: Oxford University Press.
Miller, R. (1998). The singing teacher in the age of voice science. In Sataloff, R. T. (Ed.), Vocal Health and Pedagogy (pp. 297-300). San Diego, CA: Sin-gular.
Miller, R., 1926-. (1997b). National schools of singing : English, French, German, and Italian techniques of singing revisited / Richard Miller. ([Rev. ed.]. ed.). Lanham, Md.: Scarecrow Press.
Mitchell, H. F., & Kenny, D. T. (2004a). The effects of open throat technique on long term average spectra (LTAS) of female classical voices. Logopedics Pho-niatrics Vocology, 29(3), 99-118.
Mitchell, H. F., & Kenny, D. T. (2004b). The impact of “open throat” technique on vibrato rate, extent and onset in classical singing. Logopedics Phoniatrics Vocology, in press.
Mitchell, H. F., Kenny, D. T., Ryan, M., & Davis, P. J. (2003). Defining open throat through content analy-sis of experts’ pedagogical practices. Logopedics Phoniatrics Vocology, 28(4), 167-180.
Monahan, B. J. (1978). The art of singing: a compen-dium of thoughts on singing published between 1777 and 1927. Metuchen, N.J.: Scarecrow Press.
Nair, G. (1999). Voice tradition and technology: a state-of-the-art studio. San Diego: Singular Publishing Group.
Omori, K., Kacker, A., Carroll, L. M., Riley, W. D., & Blaugrund, S. M. (1996). Singing power ratio: quan-titative evaluation of singing voice quality. Journal of Voice, 10(3), 228-235.
Puritz, E. (1956). The teaching of Elisabeth Schumann. London: Methuen.
Reid, C. L. (1975). Voice: psyche and soma. New York: Joseph Patelson Music House.
Reid, C. L. (1983). A dictionary of vocal terminology. New York: Joseph Patelson Music House.
Robison, C. W., Bounous, B., & Bailey, R. (1994). Vo-cal Beauty: A study proposing its acoustical defini-tion and relevant causes in classical baritones and fe-
male belt singers. Journal of Singing, 51, 19-30.Saunders, T. C., & Holahan, J. M. (1997). Criteria-spe-
cific rating scales in the evaluation of High School instrumental performance. Journal of Research in Music Education, 45(2), 259-272.
Siegel, S., & Castellan, J. N. J. (1988). Nonparametric Statistics for the Behavioural Sciences (Second ed.): McGraw Hill International.
Stanley, M., Brooker, R., & Gilbert, R. (2002). Exam-iner perceptions of using criteria in music perform-ance assessment. Research Studies in Music Educa-tion, 18, 43-52.
Stark, J. A. (1999). Bel canto: a history of vocal peda-gogy. Toronto, Buffalo: University of Toronto Press.
Sundberg, J. (1977). The acoustics of the singing voice. Scientific American(March), 82-91.
Sundberg, J., Gramming, P., & Lovetri, J. (1993). Com-parisons of pharynx, source, formant, and pressure characteristics in operatic and musical theatre sing-ing. Journal of Voice, 7(4), 301-310.
Thorpe, C., Cala, S., Chapman, J., & Davis, P. (2001). Patterns of breath support in projection of the singing voice. Journal of Voice, 15(1), 86-104.
Thurman, L., & Welch, G. F. (Eds.). (2000). Bodymind & voice: foundations of voice education (Rev. ed. ed.). Collegevile, Minn: VoiceCare Network.
Vennard, W. (1968). Singing: the mechanism and the technic (5th ed.). New York: Fischer.
Vurma, A., & Ross, Y. (2000). Priorities in voice train-ing: carrying power or tone quality. Musicae Scien-tiae, 4(1), 75-93.
Wapnick, J., & Ekholm, E. (1997). Expert consensus in solo voice performance evaluation. Journal of Voice, 11(4), 429-436.
Wapnick, J., Flowers, P., Alegant, M., & Jasinskas, L. (1993). Consistency in piano performance evalua-tion. Journal of Research in Music Education, 41(4), 282-292.
Ware, C. (1998). Basics of vocal pedagogy: the founda-tions and process of singing. New York: McGraw-Hill.
106
5.1 Conclusion
Expert pedagogues recognised the voice qualities associated with use of open throat
technique and reduction of the technique. Measurements performed on the LTAS in
paper 3 were unable to differentiate between experimental conditions, whereas the
human ear produced the most perfect and integrated assessment of vocal quality.
The fifth paper addressed these problematic acoustic measurements applied to LTAS
and compared them to listeners’ ratings of the voices by ranking perceptual and acoustic
measurements for overall goodness as defined in the singing literature (Barnes et al.,
2004; Omori et al., 1996; Thorpe et al., 2001; Vurma & Ross, 2000).
107
6 Vocal quality
To date, few scientific studies link acoustic measures with perceptual judgments. There
are two such studies that provide some basis for our proposed project. One, by Wapnick
and Ekholm (Wapnick & Ekholm, 1997) established 12 generally accepted perceptual
criteria applied to the assessment of voice quality in classical singing, and the other, by
Ekholm, Papagiannis and Chagnon (Ekholm et al., 1998) applied four of these criteria
(appropriate vibrato, resonance/ring, color/warmth, and clarity/focus) and related them
to objective measurements taken from acoustic analysis of the voice signal. However,
most current research relies on acoustic findings generated by averaging data from
groups of voices in order to characterize a good voice. The dilemma and paradox in this
approach is that the averaged findings may not adequately represent any of the
individual voices used to generate the desirable acoustic parameters that become the
benchmark of vocal quality.
Studies linking perceptual judgment and acoustic evaluation of performance quality in
singing encounter numerous problems. These include descriptors of voice quality,
selection of samples to evaluate and the absence of a gold standard against which to
assess even expert judgments, since inter-rater reliability is generally very poor, and
there are not well established relationships between acoustic profiles and perceptual
judgements.
Establishing criteria specific rating schemes does guide a listener’s thought process and
focuses them on the task. However, Ekholm, Papagiannis et al. (1998) found high
108
correlations between perceptual criteria an overall judgment of quality. Using a number
of criteria, rather than a single overall judgement may provide an illusion of objectivity.
Despite the widespread use of LTAS in music and voice research to date, there has been
little consideration given to the methodologies used to analyse and compare LTAS. As
early as 1975, Jansson and Sundberg (1975) recognised the benefits of using LTAS
analysis to show the individuality of each measured timbre. Their preliminary studies
identified the impact of musical task, length of time, performer or instrument and
performance instruction on the shape and reproducibility of LTAS (Jansson &
Sundberg, 1975). To date, there has been no similar study specifically for the singing
voice, and in particular, to test the value of current analysis methods performed on
LTAS in singing.
More recently, LTAS have been used to quantify perceptual cues guiding or misleading
listeners to judge singing samples as boys or girls (Howard et al., 2002). Exemplars of
LTAS were presented to explain the acoustic cues which identified girls as girls, boys as
boys or girls as boys. This indirectly assessed the vocal quality through perceptual cues
and acoustic output. While there has been major progress in the use of LTAS in singing
research, no study cross referencing perceptual response to LTAS exemplar exists for
operatic or classical voices, or as an explicit investigation of vocal quality and acoustic
cues. In light of the extensive vocal literature, this lack of perceptual evaluation
represents a considerable defect of current acoustic evaluations.
109
Further, conventional measures of SPR and ER on LTAS have not been sufficiently
sensitive to evaluate vocal quality to a given instruction within the same singer or across
singers (Mitchell & Kenny, 2004). Having established that ER and SPR may not be
representative of vocal quality, this study aims to assess the sensitivity of SPR and ER
to evaluate and report LTAS similarities and differences in singing. Central tendencies
in LTAS have been used to represent voice types (soprano to baritone) (Sundberg,
2001) or mean trends of spectral changes between pairs of LTAS (Barnes et al., 2004;
Mitchell & Kenny, 2004; Thorpe et al., 2001). These averages of LTAS may, in fact,
not represent any one voice and may be manufactured by clustered measurement
(Mitchell & Kenny, 2004).
In our fifth paper LTAS measurements were examined with respect to the perceptual
ratings of singers. Acoustic measurements were ranked for vocal quality, specifically,
carrying power after the work of (Barnes et al., 2004; Omori et al., 1996; Thorpe et al.,
2001). As expected, there was a strong relationship between the two acoustic measures,
but there was no relationship between perceptual rankings of vocal beauty and acoustic
rankings of vocal quality. It was argued that the human ear makes the most effective
acoustic evaluation and current analyses used to evaluate singers’ LTAS are not
sufficiently sensitive to discriminate vocal quality and the conclusions of this paper
suggest that LTAS should not be viewed as templates for overall vocal quality.
110
PAPER 5
Kenny, D. T., & Mitchell, H. F. (in press). Acoustic and perceptual appraisal of vocal gestures in the female classical voice Journal of
Voice.
111
ARTICLE IN PRESS
Acoustic and Perceptual Appraisal of VocalGestures in the Female Classical Voice
Dianna T. Kenny and Helen F. Mitchell
New South Wales, Australia
Summary: Long-term average spectra (LTAS) have identified features in thesounds of singers and have compared different vocal qualities based on energychanges that occur during different vocal tasks. In this study, we comparedthe perceptual ratings of vocal quality of expert pedagogues with acousticmeasures performed on LTAS. Fifteen expert judges rated 24 samples withsix repeats of six advanced singing students under two conditions: “optimal”(O), which represented the application of the maximal open throat technique;and “suboptimal” (SO), which represented the application of the reduced openthroat technique. LTAS were performed on each singing sample, and twoconventional assessments of peak energy height [singing power ratio (SPR)]and peak area [energy ratio (ER)] were calculated on each LTAS. Perceptualscores, SPR, and ER were rank ordered. We then compared perceptual rankingswith rankings of acoustic measures (SPR and ER) to assess whether theseacoustic measurements matched the perceptual judgments of vocal quality.Although we found the expected significant relationship between SPR andER, there was no relationship between perceptual ratings of vocal samples orsingers based on SPR or ER. These findings suggest that because LTASmeasures are not consistent with perceptual ratings of vocal quality, suchmeasurements cannot define a voice of quality. Future research with LTAS toassess vocal quality should consider alternative measures that are more sensi-tive to subtle differences in vocal parameters.
Key Words: Long-term average spectra—Perceptual rankings—Voicequality.
Accepted for publication December 9, 2004.From the Australian Centre for Applied Research in Music
Performance (ACARMP), Sydney Conservatorium of Music,The University of Sydney, New South Wales, Australia.
Address correspondence and reprint requests to DiannaKenny, Australian Centre for Applied Research in Music Perfor-mance (ACARMP), The Conservatorium of Music, The Uni-versity of Sydney, New South Wales, Australia 2006. E-mail:[email protected]
Journal of Voice, Vol. ■, No. ■, pp. ■
0892-1997/$30.00� 2005 The Voice Foundationdoi:10.1016/j.jvoice.2004.12.002
1
INTRODUCTION
What qualities of the singing voice define goodvocal quality and what constitute its specific vocalfeatures? This question has come into sharp focussince the advent of acoustic analysis. Few empiricalstudies classify good sound quality in singing,although perceptual studies have attempted to de-scribe the features of good singing1–3 and link these toacoustic features of voice.
Expert listeners are generally reliable in theirjudgments of overall quality. Ratings of good andpoor performance are correlated most strongly with
112
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL2
tone quality and intonation.1,2,4 Strong correlationshave also been observed among different descrip-tors that assess quality, such as color/warmth, reso-nance/ring, clarity/focus, and appropriate vibrato,which indicate that these factors converge on thesame underlying construct of overall vocal quality.Similar findings were reported in Merritt et al,5
where listeners matched their perception of perfor-mance anxiety of speakers with a variety of vocaland physical features. The ratings on the separatevocal and performance features (eg, physical ease,presence, gesture, eye contact, pace, clarity, andeye contact) all intercorrelated highly. Rating theseparate features added nothing to the overall ratingof the total performance. It seems that judges applypersonal constructs to assist with their judgmentstrategies as they often cannot articulate the compo-nents of sound quality on which their judgmentswere based.6
Vocal pedagogues in the singing studio mustassess each voice and devise a technical and aes-thetic program to improve its basic sound. “Openthroat” is one technical component taught in manymodern singing studios, and pedagogues agree thatit produces an identifiable quality in the sound,which is described as “even and consistent,” “bal-anced and coordinated,” “round,” and “warm.”Mitchell et al,4 in subsequent studies, comparedthe open throat technique (O) with a reduction ofthe technique described as suboptimal (SO) in thesame advanced singing students and found that ex-perts could identify the technique in 83% ofsamples.7,8
Applied research in voice must address the goalsof singing pedagogy in assessing and refining avoice of quality and generate perceptually viablemeasures to systematically define voice. Ideally,we need exemplars of vocal quality and to find mea-surements that enable us to rank voices in accor-dance with acoustic measures. One key questionaddressed in this article is whether acoustic featuresare associated with perceptual preferences of overallquality. To date, research into the singing voice hasdescribed acoustic features of voice and its visualrepresentation with few or no links to either percep-tual preferences or pedagogy.
Recent research has struggled to identify acousticcues that attract the highest perceptual rankings from
Journal of Voice, Vol. ■, No. ■, 2005
expert listeners.2,3,9 The voice can be representedacoustically in many ways. In current singing re-search, long-term average spectra (LTAS) are widelyapplied to represent timbre and vocal features, bothin speech and in singing. An LTAS gives an overallimpression of an entire excerpt by identifying certainconsistent features contained in the sound overtime.10 In singing samples, it averages out short-termvariations in phonetic structure11 and is a measureof the decibel level of the time-average of the powerof the acoustic signal at each frequency. A conven-tional way of assessing LTAS is to reduce the infor-mation it contains to a single meaningful number, bycomputing the ratio of energies in a low- and a high-frequency band.12,13 In singing, measures of spectracompare energy peak height [singing power ratio(SPR)] and peak area [energy ratio (ER)] between0–2 and 2-4 kHz. The difference between the heightof the major peaks between 0–2 and 2–4 kHz12
quantifies the relative energy between 2 and 4 kHz,although it does not account for the shape of theseenergy peaks. Assessing the areas under the LTAScurve at 0–2 and 2–4 kHz, rather than at the highestpeaks of energy, also identifies the areas in whichthe energy is reinforced or reduced.13 Both SPR andER enable effective comparison between individualsingers and between groups of singers.
LTAS exemplars of voice types14 and singinggenres15,16 are available. LTAS curves have also dif-ferentiated male and female voices,17 singers andspeakers,18 solo voice and choral voice,19 and popor country from opera singers15,16 based on thechanges in energy distribution necessary for thesedifferent vocal tasks.
In singing literature, particular LTAS models orexemplars are associated with particular vocal quali-ties; high-range energy within the LTAS identifiesvoices that produce a “ringing quality”2,3 or carryingpower or amplification over an orchestra,9,13,14
which is considered essential to operatic voices. InBarnes et al,20 the LTAS provides information aboutthe formant structure. However, it is premature toassert a connection between projection as demon-strated in LTAS and vocal quality, as Thorpe et al13
and Barnes et al20 have been the only studies onvocal projection in which the precise standard ofthe singing subjects has been specified and a homo-geneous group of singers of national and interna-tional reputation have been used.21 However, with
113
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 3
conservatory students, Vurma and Ross9 reportedno correlation between measurable carrying powerand the perception of tone quality by experts. Theyalso found no association between these measuresand length of training.
To date, such LTAS patterns have not been consis-tently associated with “good” or “beautiful” singing.Bartholomew22 associated “good voice quality” withhigh energy but did not resolve whether qualitywas associated with energy. Since Bartholomew22
suggested that male singers produced a specificenergy around 2800 Hz, and Sundberg14 identifiedthe “singer’s formant” as a critical component ofoperatic singing, the spectral distribution of the clas-sical and operatic voice has been measured againstthe standards established by this pioneering research.From this research, the energy produced in an oper-atic voice to project over an orchestra is consistentlyassociated with performers of the highest standardsin operatic singing.13,20 Ekholm et al2 matchedacoustic features to specific descriptors, but like ear-lier studies,22 they reported relationships betweenacoustic and descriptive features and presented ex-emplars of high-energy only in the most highlyrated voices.
LTAS have also been used to “validate” perceptualcues. For example, Howard et al23 and White24 con-ducted studies in which raters were asked to identifychild voices as male or female. Both studies foundtypically different LTAS for each gender and similar-ities between the spectra of girls identified as boysand boys identified as girls. LTAS is therefore reli-able in distinguishing male from female child andadult voice.23–25 However, this is a long way fromdetermining vocal quality among like voices, partic-ularly singing voices (eg, experienced adult femaleclassical singers).
A variation of the study type that applies percep-tual ratings to validate acoustic findings is withacoustic parameters to predict qualities, such as emo-tion in voice. These studies have found judges to beinconsistent and unreliable26 and/or to have awide range of responses to the proposed emotionalcontent of the music.27–30 Where listeners are consis-tent in identifying emotional content, stimuli havea tendency to be exaggerated, for example, in vibratoparameters,30 dynamics,31 or tone onset.32
Moreover, in a previous study, Mitchell andKenny33 tested the energy distribution of singingwith and without the use of open throat technique.Although visual inspection of LTAS illustrated smalldifferences, conventional measures performed onthese LTAS (SPR and ER) did not conclusively dif-ferentiate the high energy produced with and withoutopen throat across all singers. There is no doubtthat an operatic singer must produce energy in sucha way as to amplify his/her voice over an orchestra,but the visual presentation of operatic vocal qualitydoes not necessarily reflect its perception by expertlisteners. Similarly, Titze et al34 found the acousticoutput in “twangy” quality boosted energy around3 kHz. If vocal qualities other than operatic alsoproduce marked energy between 2 and 4 kHz, tonequality may not easily be demonstrated in the acous-tic measure of LTAS. The timbral effects of theopen throat technique have not been tested withrespect to the LTAS parameters and rankings oroverall perceptual rankings. Identifying trends in ERand SPR across all singers may result in an averagethat does not represent an actual voice and may notadequately represent the timbre of any individualvoices.
In this study, we tested whether the spectral mea-sures made on LTAS in female classical singerscould assess the difference in vocal quality betweenO and SO techniques and whether the ratings ofpedagogues of overall quality matched LTAS.
METHOD
ParticipantsListeners
Listeners were 15 experienced singing peda-gogues, 12 women and 3 men, aged between 37 and76 years, with a mean of 54 years. Six participantshad a postgraduate qualification in singing, five hada diploma of music or singing, and three a bachelorsdegree in music. One cited extensive internationalperforming experience as their qualification in sing-ing. All had taught singing for 4 to 40 years, with anaverage for 20 years. Thirteen of fifteen participantstaught singers in a Conservatorium of music. Over-all, the singing studios comprised an average of39.5% operatic students and 36.7% classical stu-dents. For 11 pedagogues, most of their studio com-prised these two genres. Six pedagogues also taught
Journal of Voice, Vol. ■, No. ■, 2005
114
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL4
�20% of musical theater students in their studio.Eleven taught a proportion of international and na-tional level singers, nine at a big city or regional/touring, and eight at the local community level.Participants were either known to the researchersvia affiliations with key music centers in Australiaor volunteered in response to an advertisement ina national singing organization newsletter. Partici-pants were sent information about the project andwere invited to take part in a perceptual study ofsinging technique. They were required to participatein a single listening session at a time and locationconvenient to them.
Before commencement, participants completed aquestionnaire detailing their musical and teachingexperience and current singing studio. Each partici-pant was asked for information related to age,number of years teaching, and highest qualificationsattained in music and/or singing. Participants wereasked to classify their singing studio according totheir students. Specifically, they gave the proportionof their private studio and studio at a musical conser-vatorium. Finally, participants were asked to classifytheir studio according to their primary singing genres(opera, contemporary music theater, musical the-ater, concert/oratorio/recital, choral) and the level atwhich their students performed (superstar, interna-tional, national/big city, regional/touring, local com-munity, full-time students of singing, amateur) andestimate the percentage each played in the per-forming career of their students.21
SingersSix female singers, three sopranos and three
mezzo-sopranos, volunteered to participate in thisproject. They were advanced students with excellenttechnique of an experienced singing pedagogue, whois a lecturer in vocal studies and opera at a stateConservatorium of Music in Australia. It is thepremier institution for musical education in thecountry and has produced singers of internationalrepute. Criteria for participant selection includedsingers who (1) had a good classical singing tech-nique for their level of training and experience and(2) understood and demonstrated skillful control of“open throat” or “retraction” techniques in theirsinging.
Journal of Voice, Vol. ■, No. ■, 2005
Before the voice recording, participants com-pleted a questionnaire seeking information on age,years of singing study, number of years of studywith each singing teacher and highest qualificationsattained or currently undertaken in music and/orsinging, and singer type (soprano or mezzo soprano).The participants were also asked to classify thegenres of singing they performed in public (opera,classical, choral, music theater, and contemporary)and to estimate the percentage each style played intheir total performing career.
Singer participants were aged between 23 and 30years, with a mean of 26 years. All had studiedsinging for at least 7 years (average 9.8 years)and had spent an average of 5 years studying withtheir present singing teacher. Each singer held aqualification in singing or music (four had Bachelorof Music degrees and two had diplomas, in musicand/or singing) and five of six were currently under-taking a second degree in singing (three postgraduateDiploma of Opera and two Bachelor of Musicdegrees). All defined most of their singing as oper-atic (�50%), with the second most common style asclassical (�20%), in accordance with the Bunchand Chapman21 taxonomy of singing voices. Allreported that they were in good health and couldperform the tasks.
Singers were sent information about the projectand were invited to take part in an acoustic andperceptual study of singing technique. They wererequired to attend a single recording session lastingno more than an hour and were told that the objectof the study was to investigate acoustical and percep-tual features of the open throat in singing and todiscover the sound qualities associated when a singeruses some form of open throat technique comparedwith when a singer does not use the open throattechnique.
Singer protocolA protocol was developed to assess the effect
of “open throat” technique on singing. Two musicaltasks were chosen to investigate the applicationof the technique in two song excerpts. Before singing,each singer selected the sequence of their tasksbefore commencing the experiment by selecting ablank card, the reverse side of which representedone task, to reduce the possible effects of task order.
115
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 5
The musical tasksMusical tasks were chosen to test different de-
mands of good singing, but they were not musicallydifficult. They were designed to test the applicationof open throat and contained musical features de-rived from a previous qualitative study on the appli-cation of the technique,4 where application or lackof application of the technique was deemed to beparticularly valuable or noticeable. These featureswere high tessitura, sustained or legato singing, dy-namic range control, and vocal agility.
The Mozart song Ridente la Calma, K 152, bars1–27 (Figure 1A), was selected as it is a nominallysimple song in the Italian language1 with a mixtureof common musical statements involving repeatedlegato lines as well as the initial stylized leaps of amajor 4 and short scale figures. All six singers sangthis aria in the same key (F major).
The third verse of the Schubert lied, Du bist dieRuh D. 776 (Op. 59, No. 3) (Figure 1B), bars 54 to80, was chosen for its demanding vocal control,sustained musical line, and high climactic tessitura.The three sopranos and the three mezzo-sopranossang this in an appropriate key depending on sopranoor mezzo-soprano voices (E-flat, D-flat, and Cmajor).
Experimental conditionsSingers sang each of the two song excerpts under
three conditions: O, SO, and loud SO (LSO). Oinvolved the maximal application of open throat.SO involved a reduced (open throat) techniquebut still with an acceptable singing technique andwithout consciously altering any other aspect of theirtechnique. It was hypothesized, from interviews withpedagogues,4 that the SO condition would result ina reduction of sound pressure level (SPL), so a thirdcondition, LSO, involved the same instruction asthe SO condition, but with the added instruction thatthe singer should try to achieve a louder dynamic thanin SO. This condition addressed SPL as an additionalvariable to the SO condition and was similar totechnical conditions established by Foulds-Elliottet al26 in her study of emotional connection.
Each task was performed twice in the O and SOconditions, whereas the LSO condition was per-formed only once because of the slight possibility
of adverse effects on vocal health. In total, eachsinger performed each musical task five times.
InstructionsA pedagogue was present during the recording
sessions to provide accompaniment for warm-upsand practice of the tasks where necessary and toinstruct singers to achieve the required vocal pos-tures for each experimental condition. For example,she instructed them to pay attention to producingthe most open sound in their throat in the O condi-tion, and to a lesser degree in the SO and LSOconditions. Some singers asked how to produce theLSO condition, and the pedagogue instructed themto “use more twang,” as taught in their lessons.35
RecordingParticipants were given time to warm up in the
singing studio and become familiar with the roombefore recording. Recording levels for each singerwere set during this time. The voice was recordedwith a high-quality microphone (AKG C-477, AKGAcoustics, Vienna, Austria) positioned on a headboom, a constant 7-cm distance from the lips of thesinger. This placement ensured that the direct energyof the voices was recorded rather than room reflec-tions, which enabled us to use a studio environmentwith low ambient noise rather than an anechoicstudio.36 The signal was then amplified with a Beh-ringer Ultragain preamplier and digitally recordedto a CD recorder (Marantz CDR 630, Marantz JapanInc., Kanagawa, Japan).
Calibration was carried out in each recording byplaying pink noise samples immediately after eachrecording session at the same recording gain appliedfor recording the voice. For calibration of absoluteSPLs, the a sound level meter (Rion NL-06 SPL,Rion Co. Ltd, Tokyo, Japan) was placed adjacentto the AKG microphone 7 cm from a speaker (BoseLifestyle, Bose Corporation, Framingham, MA),from which the pink noise was played. The SPLshown on the sound level meter was noted for thepink noise signal and applied later for calibration.
Pink noise enables calibration of an audio play-back system across the frequency range for tasks suchas perceptual testing and comparative analysis ona computer. Pink noise is at least as good as anyother steady-state known signal. A pure sine wave
Journal of Voice, Vol. ■, No. ■, 2005
116
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL6
FIGURE 1. A. We used bars 1 to 7 as perceptual stimuli of the Mozart song Ridente la Calma, K 152. B. We used bars 54 to 60as perceptual stimuli of the Schubert lied, Du bist die Ruh D. 776 (Op. 59, No. 3).
tone is more susceptible to interference in the envi-ronment and hence would give less repeatableresults.
This data set was applied as a basis for the percep-tual and acoustic evaluations.
Preparation of perceptual CDsThe audio recordings of the singers were digitally
extracted from the CDs on a standard PC to waveaudio format in stereo at a 16-bit, 41,000-Hzsample rate [with Audiograbber software (www.audiograbber.com)]. This methodology ensures asclose as possible that the original recorded soundwas played back to the judges (no filtering or normal-ization was applied to minimize the effects ofdigital artifacts).
The recorded pink noise for each singer wasapplied to equalize the peak levels of each sampleto ensure that relative SPL for each singer wasthe same. The amplification tool in Cool Edit Pro1.2 (www.adobe.com) calculated the SPL necessaryto equalize the peak SPL of each sample, thus
Journal of Voice, Vol. ■, No. ■, 2005
making each recording relative to the level (in deci-bels) of the other samples produced on the day.
The files were then edited in Cool Edit. We usedthe first 7 bars of Mozart and bars 54 to 60 of theSchubert musical tasks for the perceptual study. Eachconsisted of an entire musical phrase lasting around20 seconds. Only samples with correct intonationwere selected in the study. We used 24 CD tracksin the study: six singers singing Mozart O and SOand Schubert O and SO. CD tracks of the sampleswere presented in random order (generated atwww.randomizer.org), and six additional repeatstracks were generated, three O and three SO, ofwhich three were the Mozart task and three wereSchubert.
Pilot studiesTwo pilot studies were conducted to consider the
methodological issues discovered in this perceptualstudy. The first pilot study was conducted, witha single expert pedagogue, to determine whetherthe three experimental conditions performed by thesingers were detectable in a perceptual study. The
117
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 7
samples of three subjects were presented in eachsinging condition (O, SO, and LSO) in random order.LSO compared with SO did not produce a suffi-ciently discernible voice quality to be correctly dis-tinguished by an expert pedagogue. The pilot testlistener marked all LSO samples as SO and remarkedthat it was not possible to differentiate between thetwo conditions, SO and LSO. As LSO was createdas acoustic confirmation of SO, it is unreasonable toexpect the human ear to detect so small a differencein timbre. The LSO condition was generated toensure acoustic differences could be attributed to theopen throat technique and not solely to loudness.The LSO condition was therefore deleted from theperceptual study.
The second pilot study addressed normalizationof SPL of perceptual samples. Normalizationreduces the possible effects of SPL as a variable. Totest the perceptual difference in normalized samples,samples were normalized with the amplification toolin Cool Edit, saved, and then reduced back to theiroriginal SPL. Pairs of samples, each containing onenormalized and returned and one original, werepresented to two expert pedagogues. Pedagoguesdescribed a discernable difference in quality, saying,for example, “the overall recording quality is re-duced” or the overall timbre had changed, with dif-ferences in the “woofy” and “tweety” balance inthe recording.
Although other perceptual studies have normal-ized SPL across stimuli,23 singing pedagogue sub-jects in Wapnick and Ekholm1 commented on thedifficulty of evaluating voices not presented withrealistic loudness. On the basis of the pilot study,we decided not to normalize samples but insteadmake them relative to known SPL, based on SPLat the recording.
Perceptual testProcedure
The perceptual test was conducted in a quiet envi-ronment, and samples were played from a SonyCD Walkman (DEJ885W; Sony Corporation, Tokyo,Japan) via closed-back stereo monitoring head-phones (Sennheiser HD 270, Sennheiser, Tullamore,Ireland). This setup enabled the study to be con-ducted in the singing studios of the participants.With headphones, it was possible to eliminate room
effects from the listening environment. This method-ology was favored as optimal sound quality, ratherthan with a computer sound card37 or sending tapes toparticipants.1,2 Only one listener took the test on anyoccasion, and each listener took the test once. Levelswere checked before each test was conducted andset at a consistent and comfortable SPL volume foreach subject.
Before presentation of stimuli, participants weregiven information on the two singing conditions, Oand SO, and were presented with the musical score ofeach musical task. They were asked to rate the overallvocal quality of each sample on a 10-point scale ata standard relevant to advanced singing students at anational Conservatorium of Music. They were alsoasked to give qualitative comments and recommen-dations about the sound quality. Each listener hada trial session with four samples, two O and twoSO, in each musical task before starting the percep-tual test.
Acoustic analysisThe audio recordings were digitized at a 16-kHz
sample rate with Phog Version 2.0 (Hitech, Sweden)software and analyzed with Soundswell Version 4.0(Hitech, Sweden), with A-weighting for SPL pinknoise calibration and subject SPL measurement.
LTAS analyses (bandwidth 300 Hz) wereperformed on the 24 musical task files.25,38 LTAScontained only voiced data.11,25 A 300-Hz LTASbandwidth is less sensitive than a narrow-band anal-ysis to movement in the partials when the singer isat higher fundamental frequencies. A complicationwith LTAS is to account for loudness, which is prob-lematic as frequencies above 2 kHz increase fasterthan those below 2 kHz as SPL increases.39
To relate these LTAS plots to the known decibelof the calibration tone, an LTAS was performed onsteady 5-second portions of the pink noise samplefor each singer, with the same 300-Hz bandwidth.The Sect tool in Soundswell does not compute theoverall equivalent level (Leq) for an LTAS.Therefore, to calculate the absolute SPL of eachcalibration tone, it was necessary to find the meanSPL of the pink noise LTAS. Each point of the LTAScurve was linearized [y � 10^(χ/10)], then all datapoints were summed, and this total was convertedback to decibels [χ � 10·log(y)]. The mean SPL of
Journal of Voice, Vol. ■, No. ■, 2005
118
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL8
pink noise (in decibels down from full scale) wassubtracted from the known SPL of the calibrationnoise measured during recording to produce the cali-bration offset. This calibration offset (in decibels)was applied to each LTAS. In earlier research withLTAS methodology, White25 arrived at a “calibrationoffset” by calculating the mean SPL of a sung /i/vowel by adding the peak level (in decibels) of theLTAS with the known level. Applying the LTASarea to calculate the mean SPL of pink noise waspreferential to calculating the LTAS peak height andwas considered a valid method for evaluating theoverall level (in decibels) of the pink noise at 7 cm.
Acoustic analysis focused on two frequency areas:0–2 kHz and 2–4 kHz. From the LTAS plots, thehighest peaks in the 0–2-kHz and 2–4-kHz regionswere labeled P1 and P2, respectively. Peak levels (indecibels) and peak center frequencies (in Hertz)were calculated for each singer in each task andcondition. We used two measures to quantify theseLTAS data. The SPR, described by Omori et al12
compares the peak levels of P1 and P2. The SPRis the difference between the level of P1 and P2 indecibels (LP1 – LP2) and is a measure of spectralattenuation between 0–2 kHz and 2–4 kHz. Whenthe energy is focused in the 0–2-kHz region, it resultsin higherSPR results and there is typically less energyreinforcement in the 2–4-kHz region. A low SPRindicates a stronger energy peak �2 kHz. AlthoughSPR does not prove the presence of the formant ofa singer, it enables inter- and intrasinger comparisonof spectral energy in the voice.12,40
The ER, described by Thorpe et al,13 comparesoverall energy between the area 0–2 kHz (A1) andthe area 2–4 kHz (A2). It is calculated by taking thedifference between average energy values for the twofrequency areas (A1 – A2). A low ER represents agreater reinforcement in the 2–4-kHz region,whereas a high ER represents a smaller energy boost�2 kHz. Areas below the LTAS curve were calcu-lated in the same way as the calibration tone. Whenthere is less reinforcement in the 2–4-kHz region, ERresults follow SPR.
Study designThe reliability of the ratings of the judges was
tested with the scores of the six repeats against theoriginal scores. Intraclass correlation coefficients
Journal of Voice, Vol. ■, No. ■, 2005
TABLE 1. Ranked Mean Perceptual Scores for EachSinger With Standard Deviations in Each Musical Task(Mozart and Schubert) in Each Condition (0 and SO)
Rank Singer Task Condition Mean SD
1 5 Schubert O 7.60 1.882 5 Mozart O 7.07 1.493 1 Mozart O 6.93 1.444 3 Mozart O 6.93 2.195 1 Schubert O 6.93 1.496 4 Schubert O 6.80 1.937 4 Mozart O 6.73 1.838 2 Schubert O 6.67 1.729 3 Schubert O 6.27 1.8710 2 Mozart O 5.80 1.3711 6 Schubert O 5.40 1.6412 5 Schubert SO 5.40 2.0313 1 Mozart SO 5.20 1.2114 2 Schubert SO 5.20 1.5215 6 Mozart O 4.67 1.5416 5 Mozart SO 4.40 1.5917 3 Schubert SO 4.13 1.6418 6 Schubert SO 3.87 1.2519 1 Schubert SO 3.73 1.7520 3 Mozart SO 3.67 1.2921 4 Schubert SO 3.53 1.6422 2 Mozart SO 3.47 1.4123 4 Mozart SO 3.47 1.3024 6 Mozart SO 2.67 1.23
(ICCs) were calculated with a two-way randomeffects model (absolute agreement).
The study design was a repeated-measures [24ratings per listener] randomized complete block witha 2 [Task (Mozart vs. Schubert)] x 2 [Condition(O and SO)] factorial structure. The main effects fortask and condition and interaction effects (contrasts)were calculated for each dependent measure.
Because the design of this study included foursamples from each of the six singers, the 24 samplesare potentially nonindependent, which would consti-tute a violation of the assumptions of correlationand other parametric tests. Before analyses of thesedata, each dependent measure was tested for serialdependency with autocorrelation. The criterionvalue for serial dependency was P � 0.404. Resultsindicated that each of the three measures wereindependent: perceptual ratings, P � 0.111; ER,P � 0.238; SPR, P � 0.151.
119
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 9
FIGURE 2. Highest ranked perceptual samples. LTAS, exemplars of the qualitative positive and negative descriptions of the soundquality of pedagogues, with details of rank, singer, task, condition, and perceptual score.
RESULTS
DescriptivesMean rating scores and standard deviations calcu-
lated from averaging all measured notes sung byeach singer, in each condition, for O and SO condi-tions in each musical task (Mozart and Schubert) arepresented in Table 1. Data are ranked by mean score.
Intrajudge reliability and test–retest reliabilityIntraclass correlations with two-way random
effects model of absolute agreement (ICC(2,1))compared the exact scores of the repeated sampleswith the original ratings. For the O condition,ICC � 0.7843 [CI 95% 0.4822 � 0.9613] F � 9.52,P � 0.0001. For SO, ICC � 0.7579 [CI 95% 0.4338� 0.9558] F � 8.60, P � 0.0001; that is, listenersmatched their exact rating in the first rendition andthe repeat in nearly five of six cases. Therefore,judges were reliable in their absolute ratings.
Overall ratingsOverall means of score by condition indicated a
significant reduction in SO compared with O in both
musical tasks. For mean score, there was a maineffect for condition (F(1, 5) � 111.872, P � 0.000),but not for musical task (F(1, 5) � 1.497, P �0.276); that is, ratings were significantly higher forO than for SO in both the Mozart and the Schuberttasks. There was no interaction between musicaltask and condition (F(1, 5) � 0.355, P � 0.577).
Singer 5 was rated highest overall, in both Mozartand Schubert tasks (7.60, 7.07), and singers 1 and3 were ranked third with a score of 6.93. For the Ocondition, the mean scores were 6.34 for Mozartand 6.61 for Schubert. The O condition scoresranged from 5.40 to 7.60. For SO, singer 5 was ratedhighest for Schubert with a score of 5.40. For SO,the mean scores were 3.81 for Mozart and 4.31 forSchubert. SO scores ranged from 2.67 to 5.40. Oscores consistently scored a higher rating than didSO. Subject 6 in Mozart O was an outlier.
LTAS analysisFigures 2–5 present LTAS performed on selected
samples in the perceptual study. LTAS may provideacoustical interpretation of the perceptual ratings, orconversely, acoustic cues that influenced ratings may
Journal of Voice, Vol. ■, No. ■, 2005
120
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL10
FIGURE 3. Third-ranked perceptual samples. LTAS, exemplars of the qualitative positive and negative descriptions of the soundquality of pedagogues, with details of rank, singer, task, condition, and perceptual score.
be evident in the LTAS. In qualitative interpretationof each sample, listeners categorized their responsesinto positive and negative comments about the vocaland technical quality and suggestions for improve-ment. In addition to LTAS, each figure providesthe overall perceptual ranking, task, condition, andscore of each sample. Finally, each figure presentsexemplars of the qualitative responses given bylisteners for these samples. These analyses wereconducted to explore the perceptual responses tothe individual samples. They illustrate the personalresponse of the listeners to the most important quali-ties contained in the sample: of the voice, the tech-nique, and the overall musical performance.
The LTAS plots of singer 5 (Figure 2) are mark-edly different than those of the other singers; thatis, above 2 kHz, she produces a different distributionof energy from the other singers.
Singers 1 and 3 produce similar LTAS plots inboth tasks and are still rated highly (Figure 3). Lowerscoring O samples showed lower amplitudes below
Journal of Voice, Vol. ■, No. ■, 2005
3 kHz and the center frequency of energy � 2 kHz,P2, at a higher frequency.
Midscoring samples included one O and three SOsamples. Figure 4 presents these LTAS data.
For lowerscoring SO samples, allLTAS plots showless-defined or prominent peaks, particularly � 2kHz. The lowest scoring SO samples (Figure 5)show a series of spectral peaks and a spectral rolloff with rising frequency with no amplified peaksabove 2 kHz.
As spectral differences were unique to each singer,the SPR of each task and condition was performed tocompare the relationship of energy peaks between0–2 and 2–4 kHz. The SPR results of the perceptualsamples are presented in Table 2, ranked from thebest to the worst SPR values. Lower SPR resultsindicate more energy between 2 and 4 kHz. The SPRranks did not correspond to the perceptual rankings.
In this study, ER was calculated for each musicaltask in each condition for each subject. The ER dataof the samples of each singer are presented in Table
121
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 11
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000
Hz
SPL
(dB
at 7
cm
)Positive Comments
"good foundation tone, voice is still building" "good round sound" "still solid core" "a tone with roundness, depth"
Negative Comments
"sound is constricted and unstable" "lacks freedom" "lacks vibrancy and warmth" "little colour"
Rank Singer Task Condition Mean Score SD min max
11 6 Schubert O 5.40 1.64 3 8
12 5 Schubert SO 5.40 2.03 2 8
13 1 Mozart SO 5.20 1.21 3 7
14 2 Schubert SO 5.2 1.52 3 7
Singer 6 Schubert O
Singer 5 Schubert SO
Singer 1 Mozart SO
Singer 2 Schubert SO
FIGURE 4. Mid-ranked perceptual samples. LTAS, exemplars of the qualitative positive and negative descriptions of the soundquality of pedagogues, with details of rank, singer, task, condition, and perceptual score.
3, ranked from the best to the worst ER. The changesto ER results by singer corresponded to SPR resultsin the Mozart and Schubert tasks from SO comparedwith O; that is, ER decreased with condition acrosssingers. The rankings of ER did not correspond toperceptual rankings.
Relationships between the rankingsTable 4 presents the rankings of each singer and
the respective musical task and experimental condi-tion, ordered by perceptual rank from highest tolowest.
The Pearson correlation coefficients were calcu-lated for each of the three dependent measures (per-ceptual rating, SPR, and ER). Table 5 shows thatalthough there is a very highly significant relation-ship between ER and SPR (as expected), there isno relationship between either of these acoustic mea-sures and perceptual judgments.
The Spearman rho correlation coefficients werecalculated for the rankings of samples and singers
on each dependent measure by the judges. Table 6displays these results.
These data indicate that there was no relationshipbetween perceptual ratings of samples or singersbased on ranked ER and SPR.
A series of ICCs (ICC2,1) were calculated to deter-mine the degree to which judges were consistent intheir rankings of samples and singers based on theirrankings with respect to ER and SPR. In the firstICC, perceptual ratings were compared with ER forsamples. The ICC was 0.012 (F � 0.987,P � 0.511), which indicated no consistency betweenthe two rankings. When this test was completed forsingers, the ICC 0.029 (F � 0.972, P � 0.527) indi-cated no consistency between the two rankings. Inthe second ICC, perceptual ratings were comparedwith SPR for both samples and singers. The ICCfor samples was 0.47 (F � 1.89, P � 0.07), whichindicated moderate consistency between the tworankings. When this test was conducted for singers,the ICC was 0.029 (F � 0.972, P � 0.527), which
Journal of Voice, Vol. ■, No. ■, 2005
122
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL12
FIGURE 5. Lowest ranked perceptual samples. LTAS, exemplars of the qualitative positive and negative descriptions of the soundquality of pedagogues, with details of rank, singer, task, condition, and perceptual score.
indicated no consistency between the two rankings.The ICC between rankings of singers on ER andSPR was 0.6 (F � 2.5, P � 0.016), which indicated asignificant relationship between the rankings ofsingers.
DISCUSSION
In this study, we recorded six female singers whoperformed two songs with two different singing tech-niques: “open throat” (O) and “reduced open throat”(SO). As expected, the audio samples related to theO production were judged by expert listeners tobe sung with a better technique than the audio sam-ples related to the SO production. These perceptualrankings, however, did not match the spectral mea-surements of two spectral parameters (SPR and ER,as measured on LTAS). We conclude that these twospectral parameters (SPR and ER) are not useful orrelevant in assessing the vocal quality differencesbetween “open throat” and “reduced open throat”singing techniques. Because we found that the maxi-mum open throat produced a vocal quality in the
Journal of Voice, Vol. ■, No. ■, 2005
singing voice that was judged perceptually by expertlisteners to produce the highest overall vocal quality,we therefore tentatively argue that optimal openthroat is key to vocal quality and that it is this aspectof vocal quality that has not been detected in LTAS.
Reliability of judgesListeners were consistent in their ratings of overall
quality, producing exactly the same score, on aver-age, for five of the six repeated samples. There wereno statistically significant differences between rat-ings of the two musical tasks. Overall preferencesfor particular singers were also remarkably similar,and judges ranked their favorite singers (5, 1, and 3)consistently higher than the others. Of these, singer 5scored the highest overall ratings.
Role of perceptual studies in assessing voiceThis study is a methodological improvement over
previous twinned acoustic and perceptual stud-ies.2,3,23 We reduced the information in LTAS toa single number to make it possible to compare
123
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 13
TABLE 2. Ranked SPR for Each Singer, in EachMusical Task (Mozart and Schubert) in Each Condition
(O and SO)
Rank Singer Task Condition SPR
1 3 Mozart SO 11.062 6 Mozart SO 12.013 3 Mozart O 13.004 1 Mozart SO 16.105 5 Schubert SO 16.596 4 Mozart O 17.657 5 Mozart SO 17.838 2 Mozart O 18.139 1 Schubert O 18.5110 1 Mozart O 18.7011 1 Schubert SO 18.8312 6 Schubert O 18.9013 5 Mozart O 19.5114 4 Mozart SO 19.6515 6 Mozart O 19.8116 5 Schubert O 20.2817 2 Mozart SO 21.1818 6 Schubert SO 21.5619 3 Schubert O 22.3420 3 Schubert SO 23.7621 4 Schubert O 24.2722 4 Schubert SO 25.3323 2 Schubert O 25.6524 2 Schubert SO 33.20
intersinger differences across pairs of LTAS aswell as intrasinger differences across pairs ofLTAS.12,13,20,33 Neither ER nor SPR measurementswere corroborated by the perceptual ratings of over-all quality. LTAS may successfully differentiate be-tween different qualities (opera versus country, pop,or speech), but this does not identify the timbralsubtleties that may be contained in LTAS. Ideally,a meaningful acoustic measurement would be com-parable with a perceptual rating by expert listeners.Although there may be acoustic cues in these singingsamples that elicit listener preferences, providing alink between these and a visual representation ofsuperior quality in LTAS presents a complex chal-lenge for the field. Although more consistency seemsto exist between LTAS and energy, we still do notknow how this affects quality as available studies didnot include a perceptual component.10,13,14,20 Also,SPR and ER may be uncorrelated to the perceived“goodness” of vocal quality because there are at
TABLE 3. Ranked ER for Each Singer, in EachMusical Task (Mozart and Schubert) in Each
Condition (O and SO)
Rank Singer Task Condition ER
1 6 Mozart SO 9.412 3 Mozart O 11.103 3 Mozart SO 11.534 5 Mozart O 14.375 5 Schubert SO 14.606 2 Mozart O 14.937 5 Mozart SO 15.108 6 Mozart O 15.269 5 Schubert O 15.4510 1 Mozart O 15.7211 1 Mozart SO 15.8212 1 Schubert O 15.9513 4 Mozart O 16.1314 6 Schubert O 16.1515 2 Mozart SO 16.1816 1 Schubert SO 16.9017 4 Mozart SO 17.1018 6 Schubert SO 18.9619 3 Schubert O 19.5120 4 Schubert O 20.0121 3 Schubert SO 20.2722 4 Schubert SO 20.6823 2 Schubert O 22.3324 2 Schubert SO 29.32
least three ways in which a singer can increase theenergy in the higher harmonics, which thus affectboth SPR and ER in a similar direction: (1) produc-tion of the formant (good quality), (2) employingpressed phonation (bad vocal quality, at least froma classical perspective), and (3) deliberate additionof twang to boost carrying power.
Acoustic measures such as SPR and ER per-formed on LTAS were never intended to providevocal quality descriptors,22 but when the lowestranked singer on the perceptual ratings achieved thehighest rankings in SPR and ER as in this study,these findings indicate that such measures on LTAScannot define vocal quality as perceived by ear. TheLTAS may contain subtle information on classicalsinging quality, such as differences between openthroat technique and a SO open throat condition. TheLTAS of O and SO conditions produced in this studywere both different than LTAS produced in otherstudies examining speech quality,11,18 pop singing,16
Journal of Voice, Vol. ■, No. ■, 2005
124
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL14
TABLE 4. Singer, Task, Condition, and Rankings forPerceptual Score, SPR, and ER, Sorted by Perceptual
Score Ranking from Highest to Lowest
Perceptual SPR ERRank Rank Rank Singer Task Condition
1 16 9 5 Schubert Optimal2 13 3 5 Mozart Optimal3 10 10 1 Mozart Optimal4 5 2 3 Mozart Optimal5 9 11 1 Schubert Optimal6 21 20 4 Schubert Optimal7 6 13 4 Mozart Optimal8 23 23 2 Schubert Optimal9 19 19 3 Schubert Optimal10 8 6 2 Mozart Optimal11 11 14 6 Schubert Optimal12 4 4 5 Schubert Suboptimal13 3 12 1 Mozart Suboptimal14 24 24 2 Schubert Suboptimal15 15 8 6 Mozart Optimal16 7 7 5 Mozart Suboptimal17 20 21 3 Schubert Suboptimal18 18 18 6 Schubert Suboptimal19 12 16 1 Schubert Suboptimal20 1 5 3 Mozart Suboptimal21 22 22 4 Schubert Suboptimal22 17 15 2 Mozart Suboptimal23 14 17 4 Mozart Suboptimal24 2 1 6 Mozart Suboptimal
or country singing.15 We therefore argue that energydistribution above 2 kHz may indicate a fundamentalquality of classical and operatic singing but is notnecessarily an indicator of overall vocal quality.Measures of SPR and ER may concur more closelywith perceptual studies of vocal projection, carryingpower, or potential amplification over an orchestra.
Exemplars of vocal quality rated by listenersWe compared exemplars of LTAS sampled from
high-, mid-, and low-perceptual ranks and presentedthese with representative qualitative descriptors.LTAS of the highest-ranking singers showed anincrease in energy between 2 and 4 kHz, whereasLTAS of mid- and low-scoring singers lacked theunified peak of energy increase above 2 kHz. Previ-ous similar findings9,13,14,18,41 have interpreted thesevisual cues in LTAS as representative of classicalvocal quality. However, as the third ranking singers1 and 3 in both the Mozart and the Schubert O
Journal of Voice, Vol. ■, No. ■, 2005
TABLE 5. Pearson Correlations Among PerceptualRating, SPR, and ER of the 24 Samples
Rating SPR
0.101SPR P � 0.638
0.064 0.960ER P � 0.767 P � 0.000
produced this unified energy peak �2 kHz (the most“masculine” LTAS shape), they should have hadthe “best” voices perceptually. However, their LTASwere different than those of the highest-rankingsinger (singer 5), who had a wider distribution ofenergy above 2 kHz. The low SO plots (Figure 5)showed a spectral rolloff more consistent with plotsof speech than with singing,11,42 but with reducedenergy from 0 to 2 kHz. Therefore, measurementsof comparative energy between 0–2 and 2–4 kHz,such as SPR or ER, were inconclusive indicators ofvocal quality. High-scoring singers were praised fortheir overall vocal quality, or combinations of posi-tive qualities that resulted in a “balanced” sound.4
Judges recognized an overall good quality in themid-scoring voices, but they identified technicalflaws that impacted on vocal production. In gen-eral, judges were more likely to comment on faultyvocal production or technique, particularly in thoseranked lowest. Negative comments across all singerswere more descriptive and focused on technical sug-gestions for improvement in the sound. Conversely,those singers ranked highly attracted little comment.It seems that a beautiful voice is difficult to describe.
LTAS, and measures applied to LTAS, are notsensitive to quality differences that were identifiedby other analyses, such as vibrato. In previousstudies, these voices were examined for vibrato andspectral differences in O and SO,7,33 but although thedifferences were significant for vibrato parameters,these differences were not confirmed by spectralanalysis, which suggests that current measures thatassess voice, such as LTAS, are not sufficientlysophisticated to represent voice. Spectrographicwork in other fields43–47 has not conclusively identi-fied the same voice in different recordings despitesimultaneous perceptual and visual cues. If this isthe case, it may not be possible to convincingly
125
ARTICLE IN PRESS
VOCAL GESTURES IN FEMALE CLASSICAL VOICE 15
TABLE 6. The Spearman Rho for Rankings ofSamples and Singers for Perceptual Rating,
SPR, and ER
Samples (n � 24) Rating SPR
0.308SPR P � 0.143
�0.006 0.169ER P � 0.977 P � 0.431
Singers (n � 6)0.214
SPR P � 0.315�0.314 0.014
ER P � 0.135 P � 0.947
illustrate good singing by visual representation alone.Indeed, with any acoustic analysis, isolation maylead to incorrect interpretation of acoustic cues inthe singing voice. The danger of such a methodis the creation of an ideal, but nonexistent, voiceof quality.
In measuring vocal quality in singing, the inter-pretation of LTAS is as subjective as is the perceptualassessment of expert listeners. Our listeners wereconsistent in their perceptual judgments and reli-able in their assessment of vocal quality. Althoughspectral assessments, such as LTAS, allow for acomparable holistic visual assessment of the voice,future voice research must assign a prominentplace to perceptual rating and avoid confusing spec-tral energy distributions with vocal quality.
Ideally, LTAS should provide an acoustic resourceto assess the spectra of a performance,10 rather thanof a single note.12,48,49 An LTAS plot, like an ex-tended listening sample, provides an overall impres-sion of the most regular features of the sound ofa singer. Given the advantages of LTAS, in that itstabilizes to a regular pattern over time10,33 whileretaining the individuality generated by its musicalstimuli and the individual singer, one would expectthe final LTAS curves to show a positive relationshipwith perceptual ratings. This result was not con-firmed in this study.
Current acoustic information available to singingstudents and singing pedagogues has limited appli-cation in the singing studio. Vocal quality is basedon multiple factors, which are unlikely to be inde-pendent. The next generation of acoustic analyses
must complement the human ear in integrating thecomplex dimensions of the human voice. Futureefforts could well be directed toward establishingperceptual benchmarks of vocal quality and theirassociated acoustical parameters. To remain viable,acoustic research in voice must attempt to emu-late the human ear.
Acknowledgments: We thank Ms. Maree Ryan for herpedagogical advice and support. We also thank Mr.Peter Thomas and Dr. Densil Cabrera for their advice onacoustical matters and the six singers for their willingparticipation in the project. We are grateful for the helpfulcomments of the reviewers in the revision of this paper.
REFERENCES
1. Wapnick J, Ekholm E. Expert consensus in solo voice per-formance evaluation. J Voice. 1997;11:429–436.
2. Ekholm E, Papagiannis GC, Chagnon FP. Relating objectivemeasurements to expert evaluation of voice quality in West-ern classical singing: critical perceptual parameters. J Voice.1998;12:182–196.
3. Robison CW, Bounous B, Bailey R. Vocal beauty: a studyproposing its acoustical definition and relevant causes inclassical baritones and female belt singers. J Singing.1994;51:19–30.
4. Mitchell HF, Kenny DT, Ryan M, Davis PJ. Defining openthroat through content analysis of experts’ pedagogicalpractices. Logoped Phoniatr Vocol. 2003;28:167–180.
5. Merritt L, Richards A, Davis PJ. Performance anxiety: lossof the spoken edge. J Voice. 2001;15:257–269.
6. Thompson WF, Diamond PCT, Balkwill L-L. The adjudica-tion of six performances of a Chopin etude: a study ofexpert knowledge. Psychol Music. 1998;26:154–174.
7. Mitchell HF, Kenny DT. The impact of “open throat” tech-nique on vibrato rate, extent and onset in classical singing.Logoped Phoniatr Vocol. 2004;29:171–182.
8. Mitchell HF, Kenny DT. Can experts identify “open throat”-technique as a perceptual phenomenon? Musicae Sci-entiae. In press.
9. Vurma A, Ross Y. Priorities in voice training: carryingpower or tone quality. Musicae Scientiae. 2000;4:75–93.
10. Jansson EV, Sundberg J. Long-time average spectra appliedto analysis of music. Part I: method and general applica-tions. Acustica. 1975;34:15–19.
11. Lofqvist A, Mandersson B. Long-time average spectrumof speech and voice analysis. Folia Phoniatr Logoped.1987;39:221–229.
12. Omori K, Kacker A, Carroll LM, Riley WD, BlaugrundSM. Singing power ratio: quantitative evaluation of singingvoice quality. J Voice. 1996;10:228–235.
Journal of Voice, Vol. ■, No. ■, 2005
126
ARTICLE IN PRESS
DIANNA T. KENNY AND HELEN F. MITCHELL16
13. Thorpe C, Cala S, Chapman J, Davis P. Patterns of breathsupport in projection of the singing voice. J Voice.2001;15:86–104.
14. Sundberg J. Articulatory interpretation of the “singing for-mant.” J Acoust Soc Am. 1974;55:838–844.
15. Cleveland TF, Sundberg J, Stone RE. Long-term averagespectrum characteristics of country singers during speakingand singing. J Voice. 2001;15:54–60.
16. Borch DZ, Sundberg J. Spectral distribution of solo voiceand accompaniment in pop music. Logoped Phoniatr Vocol.2002;27:37–41.
17. Mendoza E, Valencia N, Munoz J, Trujillo H. Differencein voice quality between men and women: use of the long-term average spectrum (LTAS). J Voice. 1996;10:59–66.
18. Barrichelo VMO, Heuer RJ, Dean CM, Sataloff RT. Com-parison of singer’s formant, speaker’s ring and LTA spec-trum among classical singers and untrained normalspeakers. J Voice. 2001;15:344–350.
19. Rossing TD, Sundberg J, Ternstrom S. Acoustic comparisonof soprano solo and choir singing. J Acoust Soc Am. 1987;82:830–836.
20. Barnes JJ, Davis PJ, Oates J, Chapman J. The relationshipbetween professional operatic soprano voice and high rangespectral energy. J Acoust Soc Am. 2004;116:530–538.
21. Bunch M, Chapman J. Taxonomy of singers used as subjectsin scientific research. J Voice. 2000;14:363–369.
22. Bartholomew W. A physical definition of “good voice qual-ity” in the male voice. J Acoust Soc Am. 1934;6:25–33.
23. Howard DM, Szymanski J, Welch GF. Listeners’ perceptionof English girl and boy choristers. Music Percept. 2002;20:35–49.
24. White P. Long-term average spectrum (LTAS) analysisof sex and gender-related differences in children’s voices.Logoped Phoniatr Vocol. 2001;26:97–101.
25. White P. A study of the effects of vocal intensity variation onchildren’s voices using long-term average spectrum (LTAS)analysis. Logoped Phoniatr Vocol. 1998;23:111–120.
26. Foulds-Elliott S, Thorpe C, Cala S, Davis P. Respiratoryfunction in operatic singing: effects of emotional connec-tion. Logoped Phoniatr Vocol. 2000;25:151–168.
27. Behrens G, Green S. The ability to identify emotional con-tent of solo improvisations performed vocally and on threedifferent instruments. Psychol Music. 1993;21:20–30.
28. Howes P, Callaghan J, Davis PJ, Kenny DT, Thorpe C. Therelationship between measured vibrato characteristics andperception in Western operatic singing. J Voice. 2004;18:216–230.
29. Siegwart H, Scherer K. Acoustic concomitants of emotionalexpression in operatic singing: the case of Lucia in Ardigli incensi. J Voice. 1995;9:249–260.
30. Sundberg J, Iwarsson J, Hagegard H. A singer’s expressionof emotions in sung performance. In: Fujimura O, Hirano
Journal of Voice, Vol. ■, No. ■, 2005
M, editors. Vocal Fold Physiology: Voice Quality Con-trol. San Diego, CA: Singular Publishing Group; 1995:217–231.
31. Kamenetsky S, Hill D, Trehub S. Effect of tempo anddynamics on the perception of emotion in music. PsycholMusic. 1997;25:149–160.
32. Gabrielsson A, Juslin PN. Emotional expression in musicperformance: between the performer’s intention and thelistener’s experience. Psychol Music. 1996;24:68–91.
33. Mitchell HF, Kenny DT. The effects of open throat tech-nique on long term average spectra (LTAS) of female classi-cal voices. Logoped Phoniatr Vocol. 2004;29:99–118.
34. Titze I, Bergan CC, Hunter EJ, Story B. Source and filteradjustments affecting the perception of the vocal qualitiestwang and yawn. Logoped Phoniatr Vocol. 2003;28:147–155.
35. Estill J. Primer of Basic Figures. 2nd ed. Santa Rosa, CA:Estill Voice Training Systems; 1996.
36. Cabrera D, Davis P, Barnes J, Jacobs M, Bell D. Recordingthe operatic voice for acoustic analysis. Acoust Australia.2002;30:103–108.
37. Erickson ML, Perry SR. Can listeners hear who is singing?A comparison of three-note and six-note discriminationtasks. J Voice. 2003;17:353–369.
38. White P. Formant frequency analysis of children’s spokenand sung vowels using sweeping fundamental frequencyproduction. J Voice. 1999;13:570–582.
39. Nordenberg M, Sundberg J. Effect on LTAS of vocal loud-ness variation. TMH-QPSR. 2003;45:93–100.
40. Lundy DS, Roy S, Casiano RR, Xue JW, Evans J. Acousticanalysis of the singing and speaking voice in singing stu-dents. J Voice. 2000;14:490–493.
41. Sundberg J. Level and center frequency of the singer’sformant. J Voice. 2001;15:176–186.
42. Novak A, Vokral J. Acoustic parameters for the evaluationof voice of future voice professionals. Folia PhoniatrLogoped. 1995;47:279–285.
43. Koenig BE. Spectrographic voice identification: a forensicsurvey. J Acoust Soc Am. 1986;79:2088–2090.
44. Hakes J, Shipp T, Doherty ET. Acoustic properties ofstraight tone, vibrato, trill and trillo. J Voice. 1987;1:148–156.
45. Hoit JD, Jenks CL, Watson PJ, Cleveland TF. Respiratoryfunction during speaking and singing in professional coun-try singers. J Voice. 1996;10:39–49.
46. Hollien HF. The Acoustics of Crime: The New Science ofForensic Phonetics. New York: Plenum Press; 1990.
47. Rose P, editor. Forensic Speaker Identification. New York:Taylor & Francis; 2002.
48. Cleveland TF. Acoustic properties of voice timbre typesand their influence on voice classification. J Acoust SocAm. 1977;61:1622–1629.
49. Sundberg J. The Science of the Singing Voice. DeKalb, IL:Northern Illinois University Press; 1988.
127
ICMPC8 CONFERENCE PROCEEDINGS
Kenny, D.T., & Mitchell, H.F., Visual and auditory perception of vocal beauty: conflict or concurrence? In: Lipscomb S, Ashley R,
Gjerdingen R, Webster P, editors. 8th International Conference on Music Perception & Cognition (ICMPC8); August 2004; Evanston, IL
USA: Causal Productions; 2004. p. 171-174.
128
1
VISUAL AND AUDITORY PERCEPTION OF VOCAL BEAUTY: CONFLICT OR CONCURRENCE?
Dianna T. Kenny & Helen F. Mitchell
Australian Centre for Applied Research in Music Performance (ACARMP) Conservatorium of Music, University of Sydney
ABSTRACT
In this study, we compared perceptual ratings of vocal beauty with acoustic measures performed on LTAS. Fifteen expert judges rated 24 samples with six repeats of six advanced singing students under two conditions: ‘optimal’ (O), representing maximal open throat and ‘suboptimal’ (SO), representing reduced open throat technique in singing. LTAS were performed on each singing sample and conventional measures of energy in singing, using peak height and peak area were calculated on each LTAS. Perceptual scores and acoustic measures of ratios of peak height and peak area of LTAS were rank ordered. While there was a significant relationship between LTAS measures, there was no relationship between perceptual ratings of vocal samples or singers based on LTAS measures. These findings suggest that since LTAS measures are not consistent with perceptual ratings of vocal beauty, they cannot be used to define a beautiful voice.
1. INTRODUCTION What features of the singing voice define vocal beauty? Can these features be measured accurately and more importantly, specifically taught? Since the 1980s technology to measure vocal acoustics has become more sophisticated and has been increasingly used in experimental research, with the findings of such research now being incorporated into texts on singing [e.g. 1-3]. While there is a proliferation of acoustic studies of the singing voice it is difficult to draw firm conclusions about the implications for singing pedagogy of many of these scientific studies, which typically use small numbers of subjects, rarely link the vocal strategies used to perceptual results, often use student singers rather than professionals, and fail to identify the teaching/learning approach of the subjects.
There is now both a vast literature on vocal acoustic properties and many unanswered questions about what conclusions can be drawn from such analyses. There are many ways that the voice can be represented acoustically, but the most common is the measurement of energy production. Long-term average spectra (LTAS) are used to represent singers’ sound [4, 5] and its different vocal qualities based on energy changes that occur during different vocal tasks. A conventional way of reducing the information in the LTAS to a single meaningful number is to compute the ratio of energies in a low and a high frequency band. Often 2 kHz is chosen as the delimiting frequency [6, 7]. Do these ratios determine a beautiful voice? This question has come into sharp focus since the advent of acoustic analysis. The utility of experimental findings for the teaching of singing depends on their incorporation into a theory of the singing voice and on their relation to current practice.
However, most current research relies on acoustic findings generated by averaging data from groups of voices in order to characterize a good voice [6, 8]. The dilemma and paradox in this approach is that the averaged findings may not adequately represent any of the individual voices used to generate the desirable acoustic parameters that become the benchmark of vocal quality. Further, vocal quality derives from a combination of acoustical parameters that interact. “The subtle individual properties that set one voice apart from another should not be averaged out” [9] (p. 299).
Studies of voice quality have taken two main forms. In the first type of study, LTAS is used to “validate” perceptual cues [10]. In the second type of study, acoustic findings are “validated” by perceptual ratings [11, 12]. There are two such studies that provide some basis for this project. One, by Wapnick and Ekholm [13] established 12 generally accepted perceptual criteria applied to the assessment of voice quality in classical singing, and the other, by Ekholm, Papagiannis and Chagnon [11] applied four of these criteria (“appropriate vibrato”, “resonance/ring”, “color/warmth”, and “clarity/focus”) and related them to objective measurements taken from acoustic analysis of the voice signal.
In this study, we investigated pedagogues’ rating of female classical singers using maximum open throat [14] and reduced open throat in their singing of a classical song and a romantic lied [15, 16]. We matched perceptual ratings to acoustic measures of each perceptual sample to determine whether acoustic analysis matched perceptual judgments of overall timbre. The aim of the study was to assess the degree of agreement between acoustic analysis and perceptual ratings of vocal beauty.
2. METHOD
Participants Listeners were 15 experienced singing pedagogues, 12 females and 3 males aged between 37 and 76 with a mean of 54 years. Six participants had a postgraduate qualification in singing, five had a diploma of music or singing, and three a bachelor degree in music. One cited extensive international performing experience as their qualification in singing. All had taught singing for between 4 and 40 years, with an average of 20 years. Thirteen of fifteen participants taught singers in a Conservatorium of music. Participating pedagogues taught, on average, 39.5% operatic students and 36.7% classical students. For 11 pedagogues, the majority of their studio comprised these two genres. Six pedagogues also taught ≥20% of musical theatre students. Eleven taught a proportion of international and national level singers, nine at big city or regional/touring and eight at local community level.
129
2
Subjects and Stimuli Six advanced singing students, studying opera at a Conservatorium of Music sang two song excerpts, Mozart song Ridente la Calma and Schubert lied Du bit die Ruh under two conditions: optimal, and sub-optimal. ‘Optimal’ (O) condition was necessary to provide a sound quality with the best technique that they could. This involved the maximal use of open throat. ‘Sub-optimal’ (SO) condition involved the use of a reduced (open throat) technique but still with an acceptable singing technique and without consciously altering any other aspect of their technique.
The acoustic signal was recorded digitally (Behringer Ultragain preamplifier/Marantz CDR 630) via a high-quality microphone (AKG C-477) positioned on a head boom a constant 7 cm distance from the singer’s lips. Pink noise was played from a speaker (Bose Lifestyle) and the microphone of a sound level meter (Rion NL-06 SPL) was coupled to the AKG microphone for calibration of absolute sound pressure levels (SPL).
The audio recordings of the singers were digitally extracted using Audiograbber software (www.audiograbber.com) on a standard PC computer to wav audio format in stereo at 41000 Hz and 16 bit sample rate. The files were then reopened in Cool Edit Pro 1.2. The first 7 bars of Mozart and bars 54 to 60 of the Schubert musical tasks were used for the perceptual study. Each consisted of an entire musical phrase lasting around 20 seconds.
Procedure The perceptual test was conducted in a quiet environment and samples were played from a Sony CD Walkman (DEJ885W) via circum-aural closed-back stereo monitoring headphones (Sennheiser HD 270). Prior to presentation of stimuli, participants were given information on the two singing conditions, O and SO, and were presented with the musical score of each musical task. They were asked to judge whether or not the singer was using open throat technique in each sample. They were asked to rate each sample on a 10-point scale. Judges who assessed singers were not the singers’ pedagogues and did not know any of their identities.
Listeners assessed 30 tracks [6 singers x 2 conditions, (O and SO) x 2 musical tasks] in random order, including six repeats, also selected at random.
Acoustic Analysis Long term average spectra (LTAS) analyses (filter 300 Hz) of the 24 song samples demonstrated the differences in spectral energy for the singers, conditions and musical tasks. LTAS contained only voiced data [17].
From the LTAS plots, the highest peaks in the 0-2 kHz and 2-4 kHz regions were labeled P1 and P2 respectively. Two measures were used to quantify these LTAS data. The singing power ratio (SPR), described by Omori et al. [13] compares the maximum energy by peak levels of P1 and P2. The SPR is the difference between the level of P1 and P2 in dB and is a measure of spectral slope between 0-2 kHz and 2-4 kHz. When the energy is focused in the 0-2kHz region,
it results in higher SPR results and there is typically less energy reinforcement in the 2-4 kHz region. Although SPR does not prove the presence of a singer’s formant, it enables inter- and intra-singer comparison of spectral energy in the voice [18].
The energy ratio (ER), described by Thorpe et al. [12] compares overall energy between 0-2 kHz and 2-4 kHz. It is calculated by taking the difference between average energy values for the two frequency ranges. A low ER represents a greater reinforcement in the 2-4 kHz region and better balance between the spectral energy 0-2 and 2-4 kHz. Areas below the LTAS curve were calculated in the same way as the calibration tone. When there is less reinforcement in the 2-4 kHz region, ER results follow SPR.
Study Design Perceptual scores, SPR and ER were rank ordered. We then compared perceptual rankings with rankings of acoustic measures (SPR and ER) to assess whether the acoustic characteristics matched the perceptual judgments of overall timbre.
Reliability of the judges’ ratings was tested with the scores of the six repeats against the original scores. Intraclass Correlation Coefficients were calculated with a two-way random effects model (absolute agreement).
The study design was a repeated measures [24 ratings per listener] randomized complete block with a 2 [Task (Mozart vs Schubert)] x 2 [condition (Optimal and Sub-optimal)] factorial structure. Main effects for task and condition and interaction effects (contrasts) were calculated for each dependent measure.
3. RESULTS Intraclass correlations with two-way random effects model of absolute agreement [ICC(2,1)] compared the exact scores of the repeated samples with the original ratings. For the O condition, ICC = .7843 [CI 95% .4822 - .9613] F = 9.52, p < .0001. For SO, ICC = .7579 [CI 95% .4338 - .9558] F = 8.60, p < .0001, that is, listeners matched their exact rating in the first rendition and the repeat in nearly 5 of 6 cases. Therefore, judges were reliable in their absolute ratings.
Overall means of score by condition indicated a significant reduction in SO compared to O in both musical tasks. For mean score, there was a main effect for condition (F(1, 5) = 111.872, p = .000), but not for musical task (F(1, 5) = 1.497, p=.276); that is, ratings were significantly higher for O than for SO in both the Mozart and Schubert tasks. There was no interaction between musical task and condition (F(1, 5) = .355, p = .577).
Table 1 presents the rankings of each singer and the respective musical task and experimental condition, ordered by perceptual rank from highest to lowest.
130
3
Singer 5 was rated highest overall, in both Mozart and Schubert tasks (7.60, 7.07), and singers 1 and 3 were ranked third with a score of 6.93. The singers’ O condition scores ranged from 5.40 to 7.60. For SO, singer 5 was rated highest for Schubert. Singers’ SO scores ranged from 2.67 to 5.40. Singers’ O scores consistently scored higher rating than SO.
Pearson’s correlation coefficients were calculated for each of the three dependent measures (Perceptual rating, SPR and ER). While there was a highly significant relationship between ER and SPR, as expected (r=.960* p=.000) there was no relationship between either of these acoustic measures and perceptual judgments (ER, r= 0.064, p=.767; SPR, r=0.101p=.638).
Spearman’s rho correlation coefficients were calculated for judges’ rankings of samples and singers on each of the dependent measures. Results indicated that there was no relationship between perceptual ratings of samples or singers based on ranked ER (r=-0.006, p=.977) and SPR (r=0.308, p=.143).
Figure 1a and b present LTAS exemplars sampled from the highest ranked singer 5, and three singers ranked 3= singing in O. Figure 1c presents the middle ranked singers, singing in both O and SO and finally 1d illustrates the lowest perceptually ranked singers, all in SO.
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000
Hz
dB
S5M5
(a)
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000
Hz
dB
M1M3S1
(b)
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000
Hz
dB
S6S5M1S2
(c)
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000
Hz
dB
M2M4M6
(d)
Figure 1: LTAS of highest (a), 3= (b), middle (c) and lowest (d) perceptually ranked singers. Legends correspond to task, Mozart (M) or Schubert (S), and singer number.
Perceptual Rank
SPR Rank
ER Rank Singer Task Condition
1 16 9 5 S O 2 13 3 5 M O 3 10 10 1 M O 4 5 2 3 M O 5 9 11 1 S O 6 21 20 4 S O 7 6 13 4 M O 8 23 23 2 S O 9 19 19 3 S O
10 8 6 2 M O 11 11 14 6 S O 12 4 4 5 S SO 13 3 12 1 M SO 14 24 24 2 S SO 15 15 8 6 M O 16 7 7 5 M SO 17 20 21 3 S SO 18 18 18 6 S SO 19 12 16 1 S SO 20 1 5 3 M SO 21 22 22 4 S SO 22 17 15 2 M SO 23 14 17 4 M SO 24 2 1 6 M SO
Table 1: Singer, task, condition and rankings for perceptual score, singing power ratio (SPR) and energy ratio (ER), sorted by perceptual score ranking from highest to lowest.
131
4
4. DISCUSSION These findings suggest that since LTAS rankings were not consistent with perceptual ratings of vocal beauty, they cannot be used to define a beautiful voice. Perceptual ratings were significantly higher for O than for SO in both the Mozart and Schubert tasks. There was no interaction between musical task and condition. While there was a significant relationship between SPR and ER, there was no relationship between perceptual ratings of vocal samples or singers based on SPR or ER. Intraclass correlations with two-way random effects model of absolute agreement (ICC(2,1)) indicated that judges were reliable in their ratings.
Characteristic LTAS shapes, particularly of classical or operatic voices have been accepted as an accurate visual representation of good sound in the genre and operatic singing is positively associated with an increase in energy between 2-4 kHz [6, 7, 12]. While measures of SPR [7] and ER [6] are not intended to give information about singer’s formant, they should illustrate differences in energy distribution, or carrying power. However, Vurma & Ross [12] found that increased energy above 2 kHz did not necessarily represent vocal quality judged perceptually. In this study, measurements of comparative energy were inconclusive indicators of vocal quality.
We compared exemplars of LTAS sampled from high, mid and low perceptual ranks. LTAS of the highest-ranking singers showed an increase in energy between 2-4 kHz, whereas LTAS of mid and low scoring singers lacked the unified peak of energy increase above 2 kHz. Previous similar findings [5, 9, 12] have interpreted these visual cues in singers’ LTAS as a good sound. However, as third ranking singers 1 and 3 in O produced this unified energy peak > 2 kHz (the most ‘masculine’ LTAS shape) they should have had the ‘best’ voices perceptually. However, their LTAS were different to the highest-ranking singer (singer 5), who had a wider distribution of energy above 2 kHz. The low SO plots (Fig 1d) showed a spectral roll-off more consistent with plots of speech rather than singing.
Linking perceptual rankings with corresponding LTAS plots expands upon previous twinned perceptual and acoustic studies [7, 8]. We argue that previous templates such as shapes of LTAS curves are no longer acceptable as indicators of vocal quality. They have been produced from averages generated from groups of singers. These averages may, in fact, not represent any voice and may be an artefact of clustered measurement. A science of the singing voice cannot progress without addressing the problem inherent in accepting long-term average spectra as analogues for vocal quality without providing a link between perceptual cues and a visual representation of ‘goodness’ in singing. We have demonstrated that conventional acoustic measures of voice quality do not correspond to perceptual measures of vocal beauty. This represents a major challenge to current wisdom that acoustic parameters of voice must emulate the established acoustic norms in order to achieve overall vocal excellence.
5. REFERENCES 1. Miller, R., The structure of singing: system and art in vocal
technique. 1996, New York: London: Schirmer Books.
2. Sundberg, J., The Science of the Singing Voice. 1988, DesKalb, Illinois: Northern Illinois University Press.
3. Thurman, L. and Welch, G.F., eds. Bodymind & voice: foundations of voice education. 2000, VoiceCare Network: Collegevile, Minn.
4. Sundberg, J. (1974). Articulatory interpretation of the "singing formant". J Acoust Soc Am, 55(4), 838-844.
5. Borch, D. Z., & Sundberg, J. (2002). Spectral distribution of solo voice and accompaniment in pop music. Log Phon Vocol, 27, 37-41
6. Thorpe, C., Cala, S., Chapman, J., & Davis, P. Patterns of breath support in projection of the singing voice. J Voice, 2001. 15(1), p. 86-104.
7. Omori, K., Kacker, A., Carroll, L.M., Riley, W.D., and Blaugrund, S.M., Singing power ratio: quantitative evaluation of singing voice quality. J Voice, 1996. 10(3): p. 228-35.
8. Sundberg, J. Level and center frequency of the singer's formant. J Voice, 2001. 15(2), p. 176-186.
9. Miller, R., The Singing Teacher in the Age of Voice Science, in Vocal Health and Pedagogy, R.T. Sataloff, Editor. 1998, Singular: San Diego, CA. p. 297-300.
10. Howard, D.M., Szymanski, J., and Welch, G.F., Listeners' Perception of English Cathedral Girl and Boy Choristers. Music Perception, 2002. 20(1): p. 35-49.
11. Ekholm, E., Papagiannis, G.C., and Chagnon, F.P., Relating objective measurements to expert evaluation of voice quality in Western classical singing: critical perceptual parameters. J Voice, 1998. 12(2): p. 182-96.
12. Vurma, A., & Ross, Y. (2000). Priorities in voice training: carrying power or tone quality. Musicae Scientiae, 4(1), 75-93.
13. Wapnick, J. and Ekholm, E., Expert consensus in solo voice performance evaluation. J Voice, 1997. 11(4): p. 429-36.
14. Mitchell, H.F., Kenny, D.T., Ryan, M., and Davis, P.J., Defining open throat through content analysis of experts' pedagogical practices. Log Phon Vocol, 2003. 28: p. 167-180.
15. Mitchell, H.F. and Kenny, D.T., The impact of "open throat" technique on vibrato rate, extent and onset in classical singing. Log Phon Vocol, in press.
16. Mitchell, H.F. and Kenny, D.T., The effects of open throat technique on long term average spectra (LTAS) of female classical voice. Log Phon Vocol, in press.
17. White P. A study of the effects of vocal intensity variation on children's voices using long-term average spectrum (LTAS) analysis. Logoped Phoniatr Vocol 1998; 23:111-120.
18. Barnes JJ, Davis PJ, Oates J, Chapman J. The relationship between professional operatic soprano voice and high range spectral energy. J Acoust Soc Am 2004; in press.
132
6.1 Conclusion
We are a society of vocal critics. We recognise a bad voice, a good voice and a truly
great voice when we hear them. Yet scientific characterisations and descriptions of
beautiful voices have to date been inadequate. There is now a vast literature on vocal
acoustic properties and the visual representation of voice as well as many unanswered
questions about what conclusions can be drawn from such analyses. This systematic
body of research has assessed the degree to which acoustic measures can inform our
perception of vocal quality and the development of quality through vocal pedagogy.
In addition to the above-mentioned published papers, and on the basis of the issues
outlined above, it follows naturally that what is needed are new methods of assessing
vocal quality. Few studies employ singers of the highest performing standards or
include a perceptual component by expert listeners in combination with spectral
analysis. The papers above highlight the problems in current singing research and the
scientific literature and find that current methods used to analyse singing are insufficient
to capture and define voices of quality.
133
7 Overall conclusions and future directions
In this section, the main findings of each study are presented along with implications for
future research. Limitations and applications of this research to such future projects are
also discussed.
7.1 Review of the findings
This body of work has defined and examined a major pedagogical technique called
‘open throat’ and assessed its acoustic and aesthetic correlates. This research
demonstrated that open throat does have unique acoustic indicators and vibrato
differences were apparent between those singers using open throat compared to those
who were not. As inconsistent vibrato is considered indicative of poor singing, it was
hypothesized that testing the energy distribution in singers’ voices in each condition
would identify the timbral changes associated with open throat. Visual inspection of
LTAS confirmed differences between open throat and sub optimal open throat (that is,
reduced open throat), but conventional measurements of singing power ratio (SPR) (the
difference between the peak height) and energy ratio (ER) (the difference between peak
area), were not statistically significant. These results were not consistent with the
vibrato findings and suggested that conventional measures of SPR and ER are not
sufficiently sensitive to evaluate LTAS. In addition, there was no relationship of any
kind between perceptual rankings of vocal beauty and acoustic rankings of vocal
quality.
Current literature has used the shape of the long term average spectra (LTAS) as the
‘objective’ measure of quality but this research has found that, although acoustic
134
analysis LTAS and other indicators of vocal quality such as singing power ratio (SPR)
and energy ratio (ER) have contributed to our understanding of vocal production, it is
clear that these methods are somewhat limited in their capacity to define vocal quality,
since we have found no relationship between these standard acoustic measures and
perceptual ratings of vocal quality by expert listeners.
These studies have assessed the degree to which acoustics can inform our perception of
vocal quality and its development through vocal pedagogy. In singing the human ear is
the more perfect than any acoustic measure in evaluating vocal quality, it remains the
most effective method of assessing voice. Pedagogues process the whole sound,
interpret and assess technical features from the overall quality. Our understanding of
what produces changes in vocal quality may need to be extended to include not only the
demands of genre and emotional intention, but also its deliberate cultivation through
instruction in particular pedagogical techniques such as open throat. These studies pose
a number of unanswered questions which will be addressed in future research.
7.2 Originality of the research
This is the first body of work to document a technique used throughout historical and
current singing pedagogy and assess its value in classical singing training. In addition, it
tested conventional acoustic analyses in defining overall vocal quality, and represents a
major challenge to current wisdom that acoustic parameters of voice must emulate the
established acoustic norms in order to achieve overall vocal excellence.
Each project design challenged current research methodologies used in singing research
and reported in the singing literature. Rigorous and novel acoustic and perceptual
135
methodologies developed for these studies will form the basis for future research.
Methodologies used in the recording process have ensured consistency across data
collection, and close-miking removed room reflections, based on the most recent
research in recording operatic voices (Cabrera, Davis, Barnes, Jacobs, & Bell, 2002).
Individual papers discuss the innovations in calibration techniques (Mitchell & Kenny,
2004) and their future use of particular software used to analyse voice.
Previous scientific studies have used single subjects or averaged groups of subjects to
generate an overview of a perfect voice. Through this series of studies we have rejected
this methodological style, because it creates a mythical voice that is representative of no
actual voice.
7.3 Limitations of the research
It is believed that the homogenous group of female singers used in this study was a
methodological strength. However, we recommend replication of these results using
other singer participants to confirm the data presented above. Future studies must test
the use of the technique in male voices and also in professional singers to ascertain its
value in the singing studio. Previous studies on singing students have used cross-
sections of singers at various stages in their training (Ekholm et al., 1998; Novak &
Vokral, 1995; Vurma & Ross, 2000), but singers in this study were of similar age and
level of vocal training. The numbers in the singer subject pool were comparable to
many other LTAS and singing studies (Barnes et al., 2004; Cleveland et al., 2001;
Thorpe et al., 2001). Using two experimental musical tasks increased the information,
but statistical trends observed between tasks made it important to consider each task
individually.
136
Physiological verification of open throat was beyond the scope of this current project.
Pedagogical interpretations of the instructions to achieve the technique were considered
within the current literature (Mitchell et al., 2003) and we speculated on the possible
anatomical configurations that could account for the action and the sound quality.
Whether or not it is possible to isolate the action of open throat from other components
of good singing technique is still unclear and there may be more than one physiological
action involved to achieve the technique. Confirmation of the use of open throat was
made solely on expert perceptual identification.
Technology used in these studies has been superseded by music industry standard
software. Recording to CD recorder has been replaced by recording to computer hard
disk and studies should demand testing standard microphones that were not available or
affordable when these studies commenced. While Soundswell is innovative software
with multiple channel analysis tracks through its integration with a real-time
phonetogram program (Phog), the voice channel files were not compatible with other
software. Perceptual wave files were initially created from Swell at a bit rate of 16000
Hz, which was a third of CD quality. Therefore perceptual samples were created for
these projects using Cool Edit Pro (now Adobe Audition). There are numerous
commercial and freeware recording and analysis software packages which are more
widely used across different fields of study (for example Protools, PRAAT and Adobe
Audition) and these deserve consideration for use in future studies.
137
The recording techniques used in this singing research provided ideal conditions to
consistently test vocal energy. Using microphones mounted on head booms 7 cm from
the subjects’ lips removed any effect of room reflections. However, this produces
unnatural stimuli for a perceptual recording, unlike singing in a concert hall or theatre
space and akin to singing while standing next to a teacher in the singing studio. While
listeners in this study are familiar with such proximity of their singing students, it may
be that future recording process must absorb this effect, and make multiple recordings
conducive to perceptual evaluations, using for example, reverberation in Protools to
place the recorded sound in a church or a concert hall acoustic setting.
7.4 Future directions
Future research will be conducted with a team of music researchers, audio engineers,
singers and singing pedagogues representing the disciplines of singing pedagogy,
musicology, psychology, acoustics and audio engineering who have the capacity to
revolutionise the conceptual, theoretical and methodological bases for future studies
related to the pedagogy, development and description of singing voices of quality.
7.4.1 Establishing a gold standard
Research studies to date have not used voices judged perceptually as the most beautiful.
Future research must establish exemplars of vocal quality in professional singers by
obtaining vocal samples of professional classical singers, collected under identical
acoustic and musical conditions.
Vocal quality derives from a combination of acoustical parameters that interact. The
simple correlation studies performed in this thesis cannot deal with the non-linear
138
interactions of multiple parameters. Voice quality is likely to be based on multiple
factors, which are not necessarily independent. Models of vocal quality should be
generated from multiple acoustical parameters using non-linear statistics, such as neural
network and genetic algorithm techniques (Nannariello, Osman, & Fricke, 2002;
Zwicker & Fastl, 1999); thereby modelling the more subtle aspects of quality that are
lost in the residual of traditional correlation studies.
Expert pedagogues, speech pathologists and audio engineers will evaluate their singing
acoustically and perceptually.
7.4.2 Recording techniques
The singing research field must develop new precision acoustical measurement
techniques. New methods of measuring vocal sound must be attuned to subjectively
assessed aspects of vocal quality. Such measurement methods will yield a model of
vocal quality that can be used for objective vocal assessment, as well as improving
vocal quality through teaching.
Precision acoustical measurement techniques must include near-field measurements of
the singers (Cabrera, 2003) as well as binaural measurements suitable for subjective
evaluation. Binaural techniques (using a dummy head microphone) virtually place the
listening subject in the same acoustical environment as the original dummy head, with
only minor artefacts. Unlike conventional headphone techniques, binaural reproduction
yields an externalised auditory image with intact distance cues (Nannariello et al.,
2002). With the precise control of audio signal gain, the sound level of the reproduced
139
song can be maintained from the original level, which is of primary importance in
auditory preference studies.
In future studies, we will make use of music industry hardware and software for the
recording process. The acoustic signal will be recorded digitally to hard disc (24-bit
audio, to accommodate the very large dynamic range of singers) using a measurement-
grade microphone positioned on a head boom a constant 7 cm distance from the singer’s
lips (Cabrera et al., 2002). This close microphone tracks singer movements, as well as
rendering room acoustical effects negligible (Cabrera et al., 2002). A second
measurement microphone will be fixed at a distance of 1 m from the singer, and a
dummy head (Neumann KU100) at 2 m.
Prior to recording, the room will be acoustically optimised. The first-order reflections in
the recording room will be suppressed by the addition of sound absorptive material. In
addition, we will measure the transfer function from an artificial mouth at the singer
position to each microphone, to account for room acoustical effects and these will be
factored out in the analysis.
To assist the singer with timing and pitch accuracy and singing in an unnatural
environment, each singer will hear an identical backing track of piano accompaniment.
Closed headphones will be used for vocal and piano foldback so singers can hear both
the piano and themselves in the headphones. We will apply reverberation while
recording so singers have the sense of recording in a real acoustic space eg church or
140
big hall. This reverberation is recorded on discrete channels so that analysis of voice
and if necessary reverberation can be made separately.
7.4.3 Perceptual assessors
A significant methodological innovation in the perceptual studies would be to compare
singing pedagogues with other groups of expert judges, such as speech pathologists and
audio engineers. The latter two groups are highly attuned to minute variations in vocal
timbre, but have hitherto not been used in studies of vocal quality in singers. The use of
commercial audio engineers has significant implications for singers and singing
pedagogues as well as the music industry. The established scientific language used by
this group to describe vocal characteristics has not generalized to other vocal specialists
who have yet to develop a precise descriptive vocabulary.
Expert audio/mastering engineers are able to acoustically define their perceptions of
vocal quality by making voices as beautiful as possible, using a defined set of sound
enhancement computer software in industry standard, Protools. Comparing these
enhanced recordings, acoustically and perceptually, with the original voice sample will
demonstrate what sound qualities these voices are required to achieve for an optimal
sound quality and expertly capture the acoustic measures needed to equate with
perceptual judgments of a voice of quality.
Panels of singing pedagogues will be used to assess and discuss vocal samples in an
adjudication setting. This would enable discussion in the use of subjective language
among judges to reach consensus which would allow researchers to build a descriptive
taxonomy of voice.
141
7.4.4 Integrating science into a systematic pedagogy
The acoustic information currently available to singing students and singing pedagogues
has limited application in the singing studio. Yet, the traditional approach to pedagogy
can be a lengthy and often frustrating process for both teacher and student. Linking
perceptual judgment with acoustic measurement has the potential to provide a basis for
a more systematic pedagogical approach leading to more consistent and predictable
results in less time than current approaches.
Previous studies on vocal quality have used cross-sections of singers at various stages in
their training. Future research should document longitudinally the acoustic and
perceptual changes of vocal quality and skill acquisition in response to particular
pedagogical techniques. This approach has the potential to progress the art and science
of vocal pedagogy beyond sciolism.
It was beyond the scope of the current project to test the development and outcome of
the technique across time. Such an approach represents a major extension on current
work that describes vocal features at one point in time.
142
8 Conclusion
Musical pedagogy has evolved through an oral, non-scientific tradition. Empirical
research does not invent new ways to sing, but can objectively compare singing
traditions and pedagogies by systematically assessing and documenting the acquisition
of vocal mastery in order to identify which elements of technical and aesthetic training
produce the most beautiful voices, thereby contributing to the science of voice and a
more efficient singing pedagogy.
In future singing research, sophisticated technologies and research methods will
enhance acoustic and perceptual measurement and pedagogical practices in the musical
arts, thereby enhancing the musical development and career prospects of the next
generation of young performers and teachers. The methods that we have begun to
develop to assess vocal quality in this series of studies will be applicable to all vocal
styles including pop and music theatre and to instrumental performance. These studies
will hopefully benefit professional musicians, music students and the music industry.
143
9 References
Anderson, E. (Ed.). (1991). The letters of Mozart and his family. London: Macmillan
Press.
Appelman, D. R. (1967). The science of vocal pedagogy. Bloomington, Ind.: Indiana
University Press.
Bacilly, B. d. (1668/1968). Remarques curieuses sur l'art de bien chanter. English A
commentary upon the art of proper singing [by] Benigne de Bacilly. Translated
and edited by Austin B. Caswell. Brooklyn, N.Y.: Institute of Medival Music.
Baken, R. J., & Orlikoff, R. F. (2000). Clinical measurement of speech and voice. San
Diego: Singular Publishing Group.
Barnes, J. J., Davis, P. J., Oates, J., & Chapman, J. (2004). The relationship between
professional operatic soprano voice and high range spectral energy. J Acoust Soc
Am, 116(1), 530–538.
Barrichelo, V. M. O., Heuer, R. J., Dean, C. M., & Sataloff, R. T. (2001). Comparison
of singer's formant, speaker's ring and LTA spectrum among classical singers
and untrained normal speakers. Journal of Voice, 15(3), 344-350.
Bartholomew, W. (1934). A physical definition of "good voice quality" in the male
voice. J Acoust Soc Am, 6, 25-33.
Borch, D. Z., & Sundberg, J. (2002). Spectral distribution of solo voice and
accompaniment in pop music. Logopedics Phoniatrics Vocology, 27, 37-41.
Boyle, J. D. (1992). Evaluation of music ability. In Colwell, R. (Ed.), Handbook of
Research on Music Teaching and Learning (pp. 247-265). New York: Schirmer
Books.
144
Brown, W. S., Jr, Rothman, H. B., Morris, R. J., & Sapienza, C. M. (2001). Acoustic
analyses of trained singers perceptually identified from speaking samples.
Journal of Voice, 15(1), p. 25-35.
Brown, W. S., Jr., Rothman, H. B., & Sapienza, C. M. (2000). Perceptual and acoustic
study of professionally trained versus untrained voices. Journal of Voice, 14(3),
301-309.
Burgin, J. C. (1973). Teaching singing. Metuchen, N.J.: Scarecrow Press.
Cabrera, D. (2003). Pink noise and calibrations. In Mitchell, H. F. (Ed.). Sydney.
Cabrera, D., Davis, P., Barnes, J., Jacobs, M., & Bell, D. (2002). Recording the operatic
voice for acoustic analysis. Acoustics Australia, 30, 103-108.
Caccini, G. (1602/1930). Le nuove musiche (Riproduzione dell'edizione del 1601 / a
cura di Francesco Mantica. ed.). Roma: Monteverdi, Claudio.
Cairns, C. (1969). The memoirs of Hector Berlioz. London.
Callaghan, J. (2000). Singing and voice science. San Diego, California: Singular Pub.
Group.
Callaghan, J., & Wilson, P. (2004). How to Sing and See. Sydney: Cantare.
Cleveland, T. F., Sundberg, J., & Stone, R. E. (2001). Long-Term-Average-Spectrum
characteristics of country singers during speaking and singing. Journal of Voice,
15(1), 54-60.
Davidson, J. W., & Da Costa Coimbra, D. (2001). Investigating performance evaluation
by assessors of singers in a music college setting. Musicae Scientiae, 5(1), 33-
53.
145
Dmitriev, L., & Kiselev, A. (1979). Relationship between the formant structure of
different types of singing voices and the dimensions of supraglottic cavities.
Folia Phoniatrica, 31(4), 238-241.
Duey, P. A. (1980). Bel canto in its golden age: a study of its teaching concepts. New
York: Da Capo Press.
Easley, E. (1932). A comparison of the vibrato in concert and opera singing. In
Seashore, C. E. (Ed.), The vibrato (pp. 269-275). Iowa: University of Iowa
Press.
Ekholm, E., Papagiannis, G. C., & Chagnon, F. P. (1998). Relating objective
measurements to expert evaluation of voice quality in Western classical singing:
critical perceptual parameters. Journal of Voice, 12(2), 182-196.
Ellis, W. A. (Ed.). (1994). Richard Wagner's Prose Works (Vol. iv A music-school for
Munich). Lincoln: University of Nebraska.
Fenlon, I. (1984). Monteverdi's Mantuan Orfeo. Early Music, 12(2), 163-172.
Fields, V. A. (1947). Training the singing voice: an analysis of the working concepts
contained in recent contributions to vocal pedagogy. New York: King's Crown
Press.
Foreman, E. V. (1969). A comparison of selected Italian vocal tutors of the period circa
1550 to 1800. Unpublished diss., U. of Illinois.
Foulds-Elliott, S., Thorpe, C., Cala, S., & Davis, P. (2000). Respiratory function in
operatic singing: effects of emotional connection. Logopedics Phoniatrics
Vocology, 25(4), 151-168.
Garcia, M. (1911). Garcia's Treatise on the Art of Singing. London: Leonard & Co.
146
Geringer, J. M., & Madsen, C. K. (1998). Musicians' ratings of good versus bad vocal
and string performances. Journal of Research in Music Education, 46(4), 522-
534.
Hakes, J., Shipp, T., & Doherty, E. T. (1987). Acoustic properties of straight tone,
vibrato, trill and trillo. Journal of Voice, 1(2), 148-156.
Hemsley, T. (1998). Singing and imagination: a human approach to a great musical
tradition. Oxford; New York: Oxford University Press.
Hirano, M., Hibi, S., & Hagino, S. (1997). Physiological aspects of vibrato. In
Dejonckere, P. H., Hirano, M. & Sundberg, J. (Eds.), Vibrato (pp. 9-33). San
Diego: Singular Publishing Group, Inc.
Horii, Y. (1989). Frequency modulation characteristics of sustained /a/ sung in vocal
vibrato. Journal of Speech & Hearing Research, 32(4), 829-836.
Howard, D. M. (2003). Editorial. Logopedics Phoniatrics Vocology, 28(4).
Howard, D. M., Szymanski, J., & Welch, G. F. (2002). Listeners' perception of English
girl and boy choristers. Music Perception, 20(1), 35-49.
Howes, P., Callaghan, J., Davis, P. J., Kenny, D. T., & Thorpe, C. (2004). The
relationship between measured vibrato characteristics and perception in Western
operatic singing. Journal of Voice, 18(2), 216-230.
Jander, O., & Harris, E. T. (2004). Singing, Grovemusic online. Retrieved 12 January
2004, 2004
Jansson, E. V., & Sundberg, J. (1975). Long-Time-Average-Spectra applied to analysis
of music. Part I: method and general applications. Acustica, 34, 15-19.
Johnson, P. (1997). Performance as experience: the problem of assessment criteria.
British Journal of Music Education, 14(3), 271-282.
147
Keidar, A., Titze, I. R., & Timberlake, C. (1984). Vibrato characteristics of tenors
singing high Cs. Paper presented at the Proceedings of the Thirteenth
Symposium: Care of the Professional Voice, New York City.
Kenny, D. T., & Mitchell, H. F. (2004, August). Visual and auditory perception of vocal
beauty: conflict or concurrence? Paper presented at the 8th International
Conference on Music Perception & Cognition (ICMPC8), Evanston, IL, USA.
Kenny, D. T., & Mitchell, H. F. (in press). Acoustic and perceptual appraisal of vocal
gestures in the female classical voice. Journal of Voice.
Löfqvist, A., & Mandersson, B. (1987). Long-time average spectrum of speech and
voice analysis. Folia Phoniatrica et Logopedica, 39, 221-229.
Lundy, D. S., Roy, S., Casiano, R. R., Xue, J. W., & Evans, J. (2000). Acoustic analysis
of the singing and speaking voice in singing students. Journal of Voice, 14(4),
490-493.
Mendes, A. P., Rothman, H. B., Sapienza, C. M., & Brown, W. S., Jr. (2003). Effects of
vocal training on the acoustic parameters of the singing voice. Journal of Voice,
17(4), 529-543.
Mendoza, E., Valencia, N., Munoz, J., & Trujillo, H. (1996). Difference in voice quality
between men and women: use of the long-term average spectrum (LTAS).
Journal of Voice, 10(1), 59-66.
Mersenne, M. (1636/1979). Harmonie universelle. In MacClintock, C. (Ed.), Readings
in the history of music in performance. Bloomington: Indiana University Press.
Michel, J. F., & Myers, R. D. (1991). The effects of crescendo on vocal vibrato. Journal
of Voice, 5(4), 292-298.
148
Miller, R. (1996). The structure of singing: system and art in vocal technique. New
York: London: Schirmer Books.
Miller, R. (1997). National schools of singing: English, French, German, and Italian
techniques of singing revisited (Rev. ed.). Lanham, Md: Scarecrow Press.
Miller, R. (1998). The singing teacher in the age of voice science. In Sataloff, R. T.
(Ed.), Vocal Health and Pedagogy (pp. 297-300). San Diego, CA: Singular.
Mitchell, H. F., & Kenny, D. T. (2004a). The effects of open throat technique on long
term average spectra (LTAS) of female classical voices. Logopedics Phoniatrics
Vocology, 29(3), 99-118.
Mitchell, H. F., & Kenny, D. T. (2004b). The impact of "open throat" technique on
vibrato rate, extent and onset in classical singing. Logopedics Phoniatrics
Vocology, 29(4), 171-182.
Mitchell, H. F., Kenny, D. T., Ryan, M., & Davis, P. J. (2003). Defining open throat
through content analysis of experts' pedagogical practices. Logopedics
Phoniatrics Vocology, 28(4), 167-180.
Mitchell, H. F., & Kenny, D. T. (in press). Can experts identify “open throat” technique
as a perceptual phenomenon? Musicae Scientiae.
Monahan, B. J. (1978). The art of singing: a compendium of thoughts on singing
published between 1777 and 1927. Metuchen, N.J.: Scarecrow Press.
Nair, G. (1999). Voice tradition and technology: a state-of-the-art studio. San Diego:
Singular Publishing Group.
Nannariello, J., Osman, M. R., & Fricke, F. R. (2002). Recent developments in the
application of neural network analysis to architectural and building acoustics.
Acoustics Australia, 29, 103-110.
149
Novak, A., & Vokral, J. (1995). Acoustic parameters for the evaluation of voice of
future voice professionals. Folia Phoniatrica et Logopedica, 47, 279-285.
Omori, K., Kacker, A., Carroll, L. M., Riley, W. D., & Blaugrund, S. M. (1996).
Singing power ratio: quantitative evaluation of singing voice quality. Journal of
Voice, 10(3), 228-235.
Playford, J. (1694/1972). An introduction to the skill of musick. New York: Da Capo
Press.
Potter, J. (1998). Vocal authority: singing style and ideology. New York: Cambridge
University Press.
Praetorius, M. (1619/1979). Syntagma musicum. In MacClintock, C. (Ed.), Readings in
the history of music in performance. Bloomington: Indiana University Press.
Prame, E. (1994). Measurement of the vibrato rate of ten singers. J Acoust Soc Am,
96(4), 1979-1984.
Prame, E. (1997). Vibrato extent and intonation in professional Western lyric singing. J
Acoust Soc Am, 102(1), 616-621.
Reid, C. L. (1975). Voice: psyche and soma. New York: Joseph Patelson Music House.
Reid, C. L. (1983). A dictionary of vocal terminology. New York: Joseph Patelson
Music House.
Rosand, E. (1989). Monteverdi's mimetic art: L'incoronazione di Poppea. Cambridge
Opera Journal, 1, 113-137.
Rossing, T. D., Sundberg, J., & Ternstrom, S. (1986). Acoustic comparison of voice use
in solo and choir singing. J Acoust Soc Am, 79(6), 1975-1981.
Rossing, T. D., Sundberg, J., & Ternstrom, S. (1987). Acoustic comparison of soprano
solo and choir singing. J Acoust Soc Am, 82(3), 830-836.
150
Rothman, H. B., & Arroyo, A. A. (1987). Acoustic variability in vibrato and its
perceptual significance. Journal of Voice, 1(2), 123-141.
Rothman, H. B., Diaz, J. A., & Vincent, K. E. (2000). Comparing historical and
contemporary opera singers with historical and contemporary Jewish cantors.
Journal of Voice, 14(2), 205-214.
Saunders, T. C., & Holahan, J. M. (1997). Criteria-specific rating scales in the
evaluation of High School instrumental performance. Journal of Research in
Music Education, 45(2), 259-272.
Seashore, C. E. (1938). Psychology of music. New York and London: McGraw-Hill
Book Company.
Seashore, C. E. (Ed.). (1932). The vibrato. Iowa: University of Iowa Press.
Seidner, W., Nawka, T., & Cebulla, M. (1995). Dependence of the vibrato on pitch,
musical intensity, and vowel in different voice classes. In Dejonckere, P. H.,
Hirano, M. & Sundberg, J. (Eds.), Vibrato (pp. 63-82). San Diego: Singular
Publishing Group.
Shaw, G. B. (Ed.). (1960). How to become a music critic. London: Rupert Hart-Davis.
Shipp, T., Sundberg, J., & Hadlund, S. (1984). A model of frequency vibrato. Paper
presented at the Transcripts of the Thirteenth Symposium: Care of the
Professional Voice, New York.
Stanley, M., Brooker, R., & Gilbert, R. (2002). Examiner perceptions of using criteria in
music performance assessment. Research Studies in Music Education, 18, 43-52.
Stark, J. A. (1999). Bel canto: a history of vocal pedagogy. Toronto, Buffalo: University
of Toronto Press.
151
Sundberg, J. (1974). Articulatory interpretation of the "singing formant". J Acoust Soc
Am, 55(4), 838-844.
Sundberg, J. (1977). The acoustics of the singing voice. Scientific American(March),
82-91.
Sundberg, J. (1988). The science of the singing voice. DesKalb, Illinois: Northern
Illinois University Press.
Sundberg, J. (1995a). Acoustic and psychoacoustic aspects of vocal vibrato. In
Dejonckere, P. H., Hirano, M. & Sundberg, J. (Eds.), Vibrato (pp. 35-62). San
Diego: Singular Publishing Group.
Sundberg, J. (1995b). The singer's formant revisited. Voice, 4, 106-119.
Sundberg, J. (1997). Expressivity in singing. TMH-QPSR, 2(3), 13-19.
Sundberg, J. (2001). Level and center frequency of the singer's formant. Journal of
Voice, 15(2), 176-186.
Sundberg, J., Iwarsson, J., & Hagegard, H. (1995). A singer's expression of emotions in
sung performance. In Fujimura, O. & Hirano, M. (Eds.), Vocal Fold Physiology:
Voice Quality Control (pp. 217-231). San Diego, California: Singular Publishing
Group, Inc.
Thorpe, C., Cala, S., Chapman, J., & Davis, P. (2001). Patterns of breath support in
projection of the singing voice. Journal of Voice, 15(1), 86-104.
Thurman, L., & Welch, G. F. (Eds.). (2000). Bodymind & voice: foundations of voice
education (Rev. ed. ed.). Collegevile, Minn: VoiceCare Network.
Titze, I. (1998). Voice research: the wide pharynx. Journal of Singing, 55(1), 27-28.
Titze, I. (2001). Acoustic interpretation of resonant voice. Journal of Voice, 15(4), 519-
528.
152
Titze, I., Bergan, C. C., Hunter, E. J., & Story, B. (2003). Source and filter adjustments
affecting the perception of the vocal qualities twang and yawn. Logopedics
Phoniatrics Vocology, 28, 147-155.
Titze, I. R., Long, R., Shirley, G. I., Stathopoulos, E., Ramig, L. O., Carroll, L. M., et al.
(1999). Messa di voce: an investigation of the symmetry of crescendo and
decrescendo in a singing exercise. J Acoust Soc Am, 105(5), 2933-2940.
Titze, I. R., Solomon, N. P., Luschei, E. S., & Hirano, M. (1994). Interference between
normal vibrato and artificial stimulation of laryngeal muscles at near-vibrato
rates. Journal of Voice, 8(3), 215-223.
Tosi, P. F. (1743/1967). Opinioni de' cantori antichi e moderni. English Observations
on the florid song translated into English by Mr. Galliard. London: William
Reeves.
Treitler, L., & Strunk, O. (Eds.). (1998). Foreword to Il quinto libro de' madrigali with
the "Declaration" of Giulio Cesare Monteverdi (Vol. 4). New York and London:
W. W. Norton & Co.
Vennard, W. (1968). Singing: the mechanism and the technic (5th ed.). New York:
Fischer.
Vurma, A., & Ross, Y. (2000). Priorities in voice training: carrying power or tone
quality. Musicae Scientiae, 4(1), 75-93.
Wapnick, J., & Ekholm, E. (1997). Expert consensus in solo voice performance
evaluation. Journal of Voice, 11(4), 429-436.
Weiss, R., Brown, W. S., Jr., & Morris, J. (2001). Singer's formant in sopranos: fact or
fiction? Journal of Voice, 15(4), 457-468.
Worsthorne, S. (1968). Venetian opera in the seventeenth century. Oxford.
153
Appendix A: Project 1 Ethics Proposal
Ethics Application Project Outline
Reference 1/06/20 Supralaryngeal and Pharyngeal Widening During Singing
Many pedagogical texts advocate ‘opening the throat’ in singing in order to achieve a
certain vocal quality and timbre. The broad objectives of this research are to describe
current views and practices on pedagogical approaches that advocate voluntary control
of widening or ‘retraction’ of the ventricular folds (Estill, 1996) as well as other
approaches for pharyngeal or throat widening, for example those taught by Miller
(1996) and others (Reid, 1983). All of these approaches are collectively termed as
‘supralaryngeal and pharyngeal widening’ in this study. This is the first of a series of
novel research projects which will assess the attitudes of pedagogues and performers
towards the different pedagogical approaches as well as discovering the physiological
changes that occur, if any, when different pedagogical approaches related to
supralaryngeal and pharyngeal widening are employed. These studies will inform
pedagogy and contribute basic knowledge to singing science.
This study seeks to describe the knowledge of professional and semi-professional
performers and experienced teachers with regard to supralaryngeal and pharyngeal
widening through qualitative interviewing, in order to illustrate different terminology
and approaches used in singing pedagogy. Both classical operatic and contemporary
music theatre singing will be studied as these styles reflect the current demand on many
classical teachers’ studio time. It may also be that there are intrinsic differences across
these different styles. The objective is to study a set of different pedagogical
156
instructions used by singing teachers focused on the control of the laryngeal and
pharyngeal region.
The interview format is designed to gain information about teacher and performer
approaches to singing pedagogy. The interview will commence with a series of open-
ended and general questions with regard to all aspects of how they achieve a good
sound while singing, and factors that they believe enhance or facilitate good singing
practice. When and if they mention one of the terms associated with ‘supralaryngeal and
pharyngeal widening’, a more direct line of questioning will follow seeking more
specific information related to terminology, sound and awareness, pedagogy,
physiology and perception. If interviewees do not spontaneously mention the concepts
of interest, a question to the effect of 'Have you ever heard of throat widening?' will be
asked and then the interviewees will be invited to disclose their knowledge of the
technique.
The purpose of this project is to address the knowledge and awareness of trained singers
teachers about supralaryngeal and pharyngeal widening. It will also determine how and
when it is used in the singing studio. The interviewing will give a clear response of the
knowledge and preference of the singers for such pedagogical techniques from an
empirical point of view. The research will convey the knowledge, preference and
reasons for expressed preferences of singers and teachers for supralaryngeal and
pharyngeal widening in pedagogy and establish whether other terms are used, for
example via imagery as an access to these experts’ views on the likely effects of
supralaryngeal and pharyngeal widening on voice quality and physiology. The results of
157
this study will provide empirical information used in singing training, its underlying
similarities and differences as well as draw attention to areas needing further scientific
study.
Participants will be from a variety of backgrounds. In the interview, teachers will be
asked to indicate the main pedagogical influences on their style of teaching.
Approximately the same number of teachers who indicate Estill (1996) as a major
influence will be sought as those who indicate more classically mainstream pedagogues
such as Miller (1996). Some leading teachers will be approached directly and invited to
participate in the study. Others will be recruited through a general advertisement.
Respondents will be sent a Subject Information Sheet which will provide information
about the project. All teacher subjects will be encouraged to recommend to their
students, who perform regularly, to take part in the same study. These subjects will
contact the research team directly in order to participate.
With regard to demographic details, subjects will be asked a few questions about their
own singing background in order to ascertain the level of teaching or performing. The
interviews will be conducted in private, with a single subject, either teacher or singer in
a venue convenient for them. For leading teachers not resident in Sydney, the interview
will be conducted by telephone. Prior to the interview, subjects will be asked about their
teaching and performing background. They will be asked to classify themselves
according to a taxonomy (Bunch and Chapman, 2000)
158
The study will provide a basis in order to develop a framework for future, more specific
studies which will assess the association between any changes in the configuration or
movement of the supralaryngeal and pharyngeal structures and any reproducible voice
quality change within subjects and across subjects.
Questions on which the interview will be based are shown below. Please note that in the
open-ended interview, these questions will be developed in the conversation rather than
in the order shown.
The conversation will begin with very general questions on pedagogy: 'How do you
achieve a good sound?' and 'What techniques would you use to correct a problematic
sound?'. At any mention of a term associated with supralaryngeal and pharyngeal
widening, the questions below will be worked into the converstion.
1. Based on your experience, what do the terms Open Throat, Throat Widening,
Retraction and Space at the Back (OT/TW/R/SB) mean to you?
2. How do you feel these terms are related?
How would you describe the practice, if not one of the terms above?
3. What sound quality do you associate with these words?
What differences can you hear when someone is using OT/TW/R/SB compared
to someone who is not?
4. How does the sound quality let you know that someone is using it?
5. In your own singing, when do you/would you use it?
6. Why do you/would you use OT/TW/R/SB?
7. Suppose I was a new student – how would you tell me about some of these terms?
159
Is it something you teach?
How do you teach it?
8. Do you follow any particular ‘school of thought’ or pedagogy or scientific approach?
How does that pedagogy/writer influence your view on OT/TW/R/SB?
9. To what extent do you think that the use of OT/TW/R/SB improves or enhances the
final singing sound?
10. For you, what part of the singing process is OT/TW/R/SB involved in?
All … beginning … preparation?
11. In terms of physiology, which muscles would you expect to be working, that is to
say, what actually happens when you use OT/TW/R/SB?
12. Where do you feel it?
13. Do you think it is possible to have voluntary control of muscles responsible for
OT/TW/R/SB? Either through sensation or imagery?
Do you believe it is possible to isolate and control the muscles used in
OT/TW/R/SB?
160
Appendix B: Project 1 Subject Information Sheet
National Voice Centre
The University of Sydney
SUBJECT INFORMATION SHEET A RESEARCH STUDIES
Project Title: Current Vocal Pedagogy Investigators: Associate Professor Pamela Davis Associate Professor Dianna Kenny Ms Maree Ryan Ms Helen Mitchell Project Contact person: Associate Professor Pamela Davis, phone 9351 5352 or 0411 449 274. This project has been approved by the University of Sydney Institutional Ethics Committee. Any persons with concerns or complaints about the conduct of the research study can contact the Manager of the Ethics and Biosafety Administration on 02-9351-4811. 1. You are invited to participate in the project described below. You will, at all times, be free to decline to continue with a particular task, or to cease participating in the project. Approximately two weeks after sending this Information Sheet, Ms Mitchell will contact you by telephone to discuss whether you wish to participate and if so, to make an appointment. 2. Background to project This project will describe aspects of current vocal pedagogy, its use and its application in the singing studio. 3. Description of project - methods and demands If you choose to participate in this study you will be required to: • Fill in a short questionnaire with generalized details of your background and current teaching in
singing. You will also be asked to self-classify yourself and/or your students as to what genre of music you perform and at what level, in accordance with a set taxonomy.
• Participate in a thirty-minute semi-structured interview conducted by Helen Mitchell, discussing current vocal pedagogy and its application in your singing studio. The interview will be taped, with your permission, for later transcription. Taping of the interview will not begin until we are ready to start the interview itself. When the interview has been transcribed, you will be sent a copy and have the opportunity to confirm your answers.
• Recommend to your students who perform regularly to participate as Performer Subjects and to provide with Information Sheet B which will be supplied to you. The students would be interviewed with similar questions but with a focus on their learning. The data analyses and reports from this study will not enable any person to link specific student and teacher answers.
161
Confidentiality of all your records will be maintained by coding your results with a number and not your name. The results of the analyses may be published but your name or any groups will not be identified (i.e. the data will not enable readers of any of the output, including drafts of data analysis provided to the various investigators which may include papers for publication) to link specific data with voice type, age, or other features which may enable identification of individuals. There will be a debriefing session after all subjects have been interviewed. 4. Possible risks There are no risks involved in the participation of this project. We will be delighted to answer any further questions that you may have. If you are satisfied to continue with the study, please contact one of us to arrange an appointment. Yours sincerely, Associate Professor Pamela Davis Associate Professor Dianna Kenny Ms Maree Ryan Ms Helen Mitchell
162
Appendix C: Project 1 Subject Consent Sheet
National Voice Centre
The University of Sydney
CONSENT SHEET RESEARCH STUDIES
Project Title: Current Vocal Pedagogy Investigators: Associate Professor Pamela Davis Associate Professor Dianna Kenny Ms Maree Ryan Ms Helen Mitchell 1. I acknowledge that I have read the above statement which explains the nature, object and the possible risks of the investigation, and the statement has been explained to me to my satisfaction. Before signing this document I have been given the opportunity to ask questions relating to any possible physical harm I may suffer as a result of my participation and I have received satisfactory answers. I have also been informed that I may not receive any benefits from participating in this study. I have been offered a debriefing session after the research has been completed. 2. My decision whether or not to participate will not prejudice my future relations with The National Voice Centre or any of the investigators listed above. If I decide to participate, I am free to withdraw my consent and to discontinue participation at any time without prejudice. 3. I agree that research data gathered from the results of the study may be published provided my name is not used. __________ _________________________________ DATE Signature of Subject 4. I have fully explained to the subject _______________________________ the nature and purpose of the programme and the procedures to be employed as described above and such risks as are involved in their performance. __________ _________________________________ DATE Signature of Responsible Investigator 5. The foregoing is an accurate summary of the explanation which was made to the volunteer, and the undersigned witnessed the signatures. __________ _________________________________ DATE Signature of Auditor-Witness
163
Appendix D: Project 1 Subject Questionnaire
National Voice Centre
The University of Sydney
Questions for Participants
1. Age
2. Please outline your professional education
3. How many years have you been teaching?
4. With whom have you trained?
5. Please list masterclasses and workshops you have attended that have influenced your teaching.
6. What proportion of your professional life and income is generated through singing and the
teaching of singing?
7. Where do your students perform and at what level? Please use the Bunch/Chapman
Taxonomy given on the next pages*.
*Bunch, Meribeth and Chapman, Janice, ‘Taxonomy of Singers Used as Subjects in Scientific Research’, Journal of Voice, 14.3 (September 2000), 363-369
164
Categories of Singers * Singers have been divided into nine categories based on proven performance achievement.
1. Superstar • A singer of world-wide fame and recognition who commands the highest fees • Is supported by powerful marketing and a personal entourage • Is never out of the public eye • Commands media attention • Professional support may include teachers, coaches, repetiteurs and choreographers
2. International • A singer who performs in the international arena • Is known to the profession and amongst the public according to specific genre or style category • Commands good fees • Agency representation • Has high levels of training and support
3. National/Big City
• A singer recognisable at the national and big city level, who usually sings in the vernacular • Has good levels of pay • Agency representation • Possible marketing • Training and support according to what can be afforded
4. Regional/Touring (often seasonal)
• Normally a young singer gaining experience • May still be in training • May have an agent • Rarely has marketing
5. Local Community (often semiprofessional)
• Usually paid soloists and club singers who have sporadic engagements • Paid local prevailing rates for performance • Have variable training
6. Singing Teachers
• Who do not fit any of the other categories
7. Full-time Students of Singing (ages 18-25 in:) • Tertiary specialist training courses • Universities • Music and drama colleges
8. Amateur
• Sings for pleasure (unpaid but not unrewarded)
9. Child • Prepubertal voice
Types of Singing Opera … Contemporary Music Theatre … Musical Theatre … Concert/Oratorio/Recital … Recording Artist … Pop … Rock … Rap … Cabaret and Club … Jazz … Folk … Gospel and Soul … Country and Western … Pub and Karaoke … Church ad Cathedral … World Music … Vocal Groups (inc. Barbershop) … Session Singer … Busker/Street Singer.
* This categorisation is taken directly from the ‘Taxonomy of Singers’ of Meribeth Bunch and Janice Chapman. Bunch, Meribeth and Chapman, Janice, ‘Taxonomy of Singers Used as Subjects in Scientific Research’, Journal of Voice, 14.3 (September 2000), 363-369
165
1** SUPERSTAR % 3 NATIONAL/BIG CITY (con’d)
%
1.1 Opera 3.2 Contemporary Music Theatre 1.2 Contemporary Music Theatre 3.2a Major Principal 1.3 Musical Theatre 3.2b Minor Principal 1.4 Concert/Oratorio/Recital 3.2c Ensemble 1.5 Recording Artist Primarily 3.3 Musical Theatre 1.6 Pop 3.3a Major Principal 1.7 Rock 3.3b Minor Principal 1.8 Rap 3.3c Ensemble 1.9 Cabaret and Club 3.4 Concert/Oratorio/Recital 1.10 Jazz 3.5 Recording Artist Primarily 1.11 Folk 3.6 Pop 1.12 Gospel and Soul 3.7 Rock 1.13 Country and Western 3.8 Rap 3.9 Cabaret and Club 2 INTERNATIONAL % 3.10 Jazz 2.1 Opera Principal 3.11 Folk 2.2 Contemporary Music Theatre
Principal 3.12 Gospel and Soul
2.3 Musical Theatre 3.13 Country and Western 2.4 Concert/Oratorio/Recital 3.14 Pub and Karaoke 2.5 Recording Artist Primarily 3.15 Church/Cathedral 2.6 Pop 3.15a Soloist 2.7 Rock 3.15b Professional Chorister 2.8 Rap 3.15b1 Adult 2.9 Cabaret and Club 3.15b2 Child 2.10 Jazz 3.16 World Music 2.11 Folk 3.17 Vocal Groups (inc. Barbershop) 2.12 Gospel and Soul 3.18 Session Singer 2.13 Country and Western 3.19 Busker (specify type, Opera,
pop, etc)
2.14 World Music (specify type) 2.15 Vocal Groups (inc. Barbershop) 4 REGIONAL/TOURING % 4.1 Opera 3 NATIONAL/BIG CITY % 4.1a Major Principal 3.1 Opera 4.1b Minor Principal 3.1a Major Principal 4.1c Chorus/Cover 3.1b Minor Principal 4.2 Contemporary Music Theatre 3.1c Chorus 4.2a Major Principal 4.2b Minor Principal 4.3c Chorus/Cover
166
4** REGIONAL/TOURING (con’d)
% 6 SINGING TEACHER %
4.3 Musical Theatre 6.1 University/Schools/Colleges 4.3a Major Principal 6.2 Private 4.3b Minor Principal 4.3c Ensemble/Cover 7 FULL-TIME VOICE % 4.4 Seasonal Music Theatre/Touring
Co STUDENT
Summer Rep/Pantomime 7.1a Post graduate specialist 4.5 Concert/Oratorio/Recital training courses 4.6 Pop 7.1b University and College 4.7 Rock 7.1c Music and Drama School 4.8 Rap 4.9 Cabaret and Club 8 AMATEUR % 4.10 Jazz 8 (sings for pleasure) 4.11 Folk 4.12 Gospel and Soul 9 CHILD % 4.13 Country and Western 9.1 Opera Principal 4.14 World Music (specify type) 9.2 Contemporary Music Theatre
Principal
4.15 Vocal Groups (inc Barbershop) 9.3 Musical Theatre Principal 9.4 Oratorio Soloist 5 LOCAL COMMUNITY 9.5 Recording artist (Often semi-professional) 9.6 Pop
5.1 Opera (principal parts only) 9.7 Rock
5.2 Contemporary Music Theatre 9.8 Rap
5.3 Musical Theatre 9.9 Children’s Choir
5.4 Concert/Oratorio/Recital
5.5 Church Soloist
5.6 Pop
5.7 Rock
5.8 Rap
5.9 Cabaret and Club
5.10 Jazz
5.11 Folk
5.12 Gospel and Soul
5.13 Country and Western
5.14 Pub and Karaoke
5.15 World Music
5.16 Vocal Groups (inc. Barbershop)
5.17 Busker
**The table and wording of the table used in this questionnaire is taken from the Bunch and Chapman Taxonomy. The % column is my addition for ease of use in this study. Bunch, Meribeth and Chapman, Janice, ‘Taxonomy of Singers Used as Subjects in Scientific Research’, Journal of Voice, 14.3 (September 2000), 363-369
167
Appendix E: Project 2 Ethics Proposal
Ethics Application Project Outline
Reference 02/05/08 Musical terminology, particularly in the field of performance has been a source of
uncertainty and mystification for those who use and practice it. Many of the terms used
to describe ‘openness’ as an intrinsic element of the singing voice have not come under
sufficient scrutiny to determine a consistent terminology or associated sound quality
among expert singing pedagogues. Seashore (1938) recommended that ‘musicians scrap
a mass of the current synonyms for tone quality, because these words do not connote
any demonstrable differences in content. The diversity of words simply adds to the
confusion.’(p 111) There is, and has always been a recognition of an overall ‘good’
sound, but in this case of ‘openness’, there is need of clarification, especially in the area
of singing at a professional and/or expert level.
The use of specific words to describe a sound quality may or may not be effective in the
singing studio. A term used by a teacher to elicit a particular ‘sound quality’ is arguably
effective only if it achieves the desired effect, and not a subsequent incidental change.
The ability to divide and distinguish different sound elements with different singing
instructions from the greater whole of an ‘excellent sound’ attracts significant support in
current vocal pedagogy (Estill 1996, 1997, Miller 1996).
This study seeks to discover if a set of specific instructions to a group of singers can
change their sound quality and if this is demonstrable when a recording of their singing
is played to a panel of expert judges. The ability to distinguish precisely which
169
instruction evokes which sound is important because it maximises teaching
effectiveness and efficiency. However, it remains difficult to determine precisely the
quality of sound achieved by individual instructions. We rely not on words to describe
the sound, but the ears and experience of the pedagogues working with the most
successful singers at the current time to judge the sound.
These instructions relate to the area of ‘opening or widening’ any of the vocal tract
structures (eg. ventricular folds, and/or pharynx) from an already unconstricted starting
point in the sound. The aim of the project is to find out if a precise verbal instruction or
actual gesture can trigger a sound quality difference. That is to say, can a singer
demonstrate a difference in sound quality in a specific way, when they are given a
precise instruction believed to change the sound in terms of openness?
From the earlier study (Ethics Approval 1/06/20), we have learnt from interviews with
leading pedagogues and from some of their students a number of terms used to describe
sound qualities associated with the concept of open throat in singing. This project will
use these terms in an attempt to derive actual associations with sound quality in singing.
Methodology
Singers: Ten female singer subjects (sopranos and mezzo-soprano) will be recruited. Ms
Maree Ryan, a co-investigator from the Conservatorium of Music, is one of Australia’s
leading singing pedagogues and has many students achieve national and international
level singing. She will select singers on the basis of her expert judgement on their
expected ability to perform the proposed protocol. They will be asked to attend a
170
recording session at the Conservatorium of Music or the National Voice Centre at a
time that is convenient for them and the researchers.
Protocol for singer subjects
Singers will be asked to sing a portion of ‘Ridente la calma’ (a Mozart aria), and a
portion of ‘Du bist die Ruh’ (a Schubert lied). They will perform each under two
conditions: the first with a healthy but with a ‘not truly open’ sound; and secondly with
a ‘fully open’. Ms Ryan will select the terminology and instructions necessary for each
subject to achieve both these sounds at an optimal level. She will also be able to
accurately discern, from her wide experience, whether or not each singer is achieving
her best possible rendition for the two conditions. Singers may be asked to repeat a take
if she or the singer judge it to be less than the usual standard for that singer. Singers will
also be asked to sing a messa di voce, from the most quiet singing to the loudest and
then back to the most quiet, on a number of notes throughout their range.
Sound recording: The acoustic signal will be recorded digitally (Behringer Ultragain
preamplifier/ Marantz CDR 630) via a high-quality microphone (AKG C-477)
positioned on a head boom a constant 7 cm distance from the subject's lips so that direct
energy of the performer's voice will be recorded rather than room reflections, enabling
us to use a studio environment at the Conservatorium of Music with low ambient noise
rather than an anechoic chamber for all the recordings. A sound level meter (Rion) will
be used to calibrate the sound pressure level by reference to the level of a calibration
tone played during the recording session at the same recording gain. The singer subjects
will be asked to describe their own assessment of the performance in both conditions,
171
both of sensation and sound quality. We will adopt a qualitative approach for the
description of these experiences, as this feedback is likely to be subtle and variable. The
performers will also self-rate their performance immediately after recording each of the
conditions on a 100 mm visual analogue scale (VAS) for: warmth, freedom throughout
range, consistent sound quality, warmth, appropriate vibrato, openness, my best singing.
Perceptual analysis: The 60 excerpts (~45 secs in duration) from the original recordings
will be randomly ordered onto a CD, with 20% extra duplication to assess intra-judge
reliability, and be played to expert pedagogues for evaluation. The same excerpts used
for the acoustic analysis will be analysed perceptually. The recordings will be played by
Ms Mitchell to the pedagogue subjects, with short breaks between excerpts and asked to
rate the performances in a questionnaire on a 100 mm Visual Analogue Scale. They will
be asked about a number of criteria (warmth, freedom throughout range, consistency of
sound quality, openness, brilliance/ring, appropriate vibrato and overall performance
excellence), but the particular focus will be on associations with the concept of
‘openness’. Subjects will be able to ask to hear any particular example again, at the end
of the session. They will work in a quiet environment and the sound will be delivered by
quality speakers from a CD player. They will listen to each perceptual set within the
same day. Ms Mitchell will be present during the perceptual ratings. Intra-rater
reliability which will be set at >70% point by point agreement for judgments of
duplicate performances.
Acoustic: The overall characteristics of the recorded voice signals from the experiments
will be described using Soundswell signal analysis software (HiTech, Sweden). Long
172
term average spectra (LTAS) calculated from the voiced segments will be used to
describe of the singer's spectra in addition to an analysis of messa di voce spectra. In
order to quantify the acoustic characteristics of any changes arising from different tasks,
we will select the same climatic section of 45 seconds in duration from the song. We
will characterise the LTAS by means of its peak frequency, level of the fundamental
relative to level of F1, average spectral slope, and the presence, frequency, and
magnitude of any spectral peak in the 2-4 kHz frequency range. We will also compare
performances in the conditions using as an index the relative change in the range 0-2 vs
2-4 kHz range and will correlate the spectral energy in the 0-2 and 2-4 kHz regions with
sound pressure level. Spectrographic analysis will be performed and the presence and
depth of vibrato or any other feature relevant to the performers' voices recorded will be
measured.
173
Appendix F: Project 2 Singer Information Sheet
National Voice Centre
The University of Sydney
RESEARCH STUDY INTO THE USE OF SUPRALARYNGEAL AND PHARYNGEAL WIDENING (OPEN THROAT) IN SINGING
SINGER SUBJECT INFORMATION SHEET
You are invited to take part in a research study into the use of supralaryngeal and pharyngeal widening (open throat) in singing. The study is being conducted by Helen Mitchell, a postgraduate student at The University of Sydney, Associate Professor Pamela Davis (Director of the National Voice Centre), Associate Professor Dianna Kenny (Director of Postgraduate Studies, National Voice Centre) and Ms Maree Ryan (Lecturer in Singing, The Conservatorium of Music), and will form the basis of Helen Mitchell’s PhD at The University of Sydney.
The object is to investigate acoustical and perceptual analysis of the use of open throat in singing and to discover the sound qualities associated when a singer uses some form of open throat technique compared with when a singer does not use open throat.
If you agree to take part in this study, you will be asked to complete a short questionnaire that tells us a bit about you (your age, education, occupation and singing experience) and then to attend a recording session at the Conservatorium of Music or the National Voice Centre for a single recording at a time arranged with the researchers, lasting no more than an hour.
At this recording session, you would be asked to sing for a recording a portion of Ridente la calma (Mozart), a portion of Du bist die Ruh (Schubert), some scales and messe di voce across your vocal range and then give feedback on your view of each performance. We will ask you to sing all of these things in two ways: the first with the maximum degree of openness you would use in your technique, and the second with a smaller degree of openness, but still with an acceptable singing technique. Both these conditions will be around the same volume, so that the test is of the difference in sound quality. There are no risks involved in the participation of this project.
All aspects of the study, including results, will be strictly confidential and only investigator Helen Mitchell will have access to information on participants. Results will be coded with a number and not your name and the data will not enable readers of any of the output to link specific data with voice type, age, or other features which may enable identification of individuals. A report of the study may be submitted for publication, but individual participants will not be identifiable in such a report. There will be a debriefing session on completion of all data collection.
Participation in this study is entirely voluntary: you are not obliged to participate and - if you do participate - you can withdraw at any time. Whatever your decision, it will not affect your relationship with any of the researchers, or with the National Voice Centre.
The ethics committee of The University of Sydney has approved this study. If you have a concern or complaint about the conduct of a research study, you can contact the Manager for Ethics and Biosafety Administration, University of Sydney on (02) 9351 4811.
When you have read this information, Helen Mitchell will discuss it with you further and answer any questions you may have. If you would like to know more at any stage, please feel free to contact Helen Mitchell on 9517 5435 or [email protected].
174
Appendix G: Project 2 Singer Consent Sheet
National Voice Centre
The University of Sydney
CONSENT SHEET
RESEARCH STUDIES
Project Title: The Use of Supralaryngeal and Pharyngeal Widening (Open Throat) in Singing Investigators: Associate Professor Pamela Davis Associate Professor Dianna Kenny Ms Maree Ryan Ms Helen Mitchell 1. I acknowledge that I have read the above statement which explains the nature, object and the possible risks of the investigation, and the statement has been explained to me to my satisfaction. Before signing this document I have been given the opportunity to ask questions relating to any possible physical harm I may suffer as a result of my participation and I have received satisfactory answers. I have also been informed that I may not receive any benefits from participating in this study. I have been offered a debriefing session after the research has been completed. 2. My decision whether or not to participate will not prejudice my future relations with The National Voice Centre or any of the investigators listed above. If I decide to participate, I am free to withdraw my consent and to discontinue participation at any time without prejudice. 3. I agree that research data gathered from the results of the study may be published provided my name is not used. __________ _________________________________ DATE Signature of Subject 4. I have fully explained to the subject _______________________________ the nature and purpose of the programme and the procedures to be employed as described above and such risks as are involved in their performance. __________ _________________________________ DATE Signature of Responsible Investigator
175
Appendix H: Project 2 Singer Questionnaire The Use of Supralaryngeal and Pharyngeal Widening (Open Throat) in Singing
Questionnaire for Singer Subjects Page 1 of 3
National Voice Centre
The University of Sydney
The Use of Supralaryngeal and Pharyngeal Widening (Open Throat) in Singing
SINGER SUBJECT DEMOGRAPHIC QUESTIONNAIRE
We are interested in finding out about you. Please complete the following questionnaire by ticking the box that best describes you or writing in the space provided. When you have finished, please hand the questionnaire back to one of the researchers.
The questions begin on the next page.
176
The Use of Supralaryngeal and Pharyngeal Widening (Open Throat) in Singing
Questionnaire for Singer Subjects Page 2 of 3
1. How old are you?
_____________years 2. At what age did you begin individual singing lessons?
_____________years 3. For how many years (or months) in total have you had individual singing lessons?
_____________years (or _____________months)
4. How many different singing teachers have you had?
_____________teachers 5. Please specify the duration of your training with each singing teacher (beginning with
your first teacher):
Teacher 1: ____years or ____months Teacher 2: ____years or ____months
Teacher 3: ____years or ____months Teacher 4: ____years or ____months Teacher 5: ____years or ____months Teacher 6: ____years or ____months
177
The Use of Supralaryngeal and Pharyngeal Widening (Open Throat) in Singing
Questionnaire for Singer Subjects Page 3 of 3
6. What is the highest level of singing education you have completed?
Postgraduate degree at University or Conservatorium
Bachelor degree at University or Conservatorium
Diploma of Music
Higher School Certificate or equivalent
7. What is your current level of singing education (if this applies)?
Postgraduate degree/diploma at University of Conservatorium
Bachelor degree at University or Conservatorium
8. What voice category are you? (please tick)
Soprano.....................................
Mezzo soprano .........................
Contralto ..................................
9. What types of singing do you perform in public? Please circle the relevant singing type
and write in the space the percentage of your total performance time spent performing each type of singing.
Type of Singing % of performance time
Opera
Other classical (eg. Song and concert repertoire, church music)
Choral singing
Musical Theatre
Contemporary styles
Other (please specify)
Thank you for completing this questionnaire
178
Appendix I: Vibrato onset raw data
Schubert Vibrato Onset Subjects 1-2
word SO 1 SSO 1 SLSO 1 SO 2 SSO 2 SLSO 2 1 Dies 0.63 0.26 0.04 0.41 0.43 0.48 2 Au 0.14 0.91 0.05 0.23 0.04 0.46 3 zelt 0.15 0.81 0.59 0.26 0.22 0.44 4 Dei 0.08 1.00 0.44 0.28 0.07 0.25 5 glanz 0.08 0.06 0.64 0.31 0.23 0.24 6 lein 0.09 0.09 0.31 0.10 0.19 0.25 7 hellt1 0.08 0.06 0.07 0.10 0.34 0.28 8 hellt 2 nil nil nil nil nil nil 9 ganz 0.03 1.02 1.62 0.26 0.27 0.79
10 ganz 0.09 0.90 0.50 0.30 0.31 0.55 11 Dies 0.16 0.13 0.54 0.25 0.41 0.29 12 Au 0.13 0.52 0.69 0.26 0.09 0.42 13 zelt 0.02 0.81 0.74 0.19 0.26 0.62 14 Dei 0.15 0.08 0.94 0.25 0.24 0.56 15 glanz 0.07 0.40 0.58 0.24 0.23 0.47 16 lein 0.08 0.49 0.79 0.12 0.48 0.24 17 hellt1 0.06 0.06 0.07 0.08 0.33 0.27 18 hellt 2 nil nil nil nil nil nil 19 ganz 0.03 1.25 1.53 0.23 0.29 0.56 20 ganz 0.11 1.24 1.24 0.03 0.28 0.39
180
Schubert Vibrato Onset Subjects 3-4
word SO 3 SSO 3 SLSO 3 SO 4 SSO 4 SLSO 4 1 Dies 0.03 Absent 0.67 0.20 0.15 0.06 2 Au 0.18 0.28 0.61 0.05 0.24 0.15 3 zelt 0.02 0.41 0.63 0.02 0.05 0.03 4 Dei 0.26 0.05 0.60 0.25 0.06 0.04 5 glanz 0.06 0.40 0.46 0.05 0.17 0.23 6 lein 0.22 0.24 0.08 0.07 0.40 0.06 7 hellt1 0.17 0.19 0.09 0.06 0.24 0.06 8 hellt 2 nil nil nil nil nil nil 9 ganz 0.05 0.05 0.21 0.06 0.27 0.04
10 ganz 0.05 0.36 0.39 0.04 0.23 0.22 11 Dies 0.17 Absent 0.22 0.25 0.08 0.07 12 Au 0.03 0.58 0.58 0.03 0.05 0.04 13 zelt 0.03 0.19 0.22 0.03 0.65 0.04 14 Dei 0.24 0.19 0.22 0.22 0.06 0.03 15 glanz 0.08 0.25 0.19 0.20 0.32 0.17 16 lein 0.21 0.06 0.76 0.25 0.55 0.03 17 hellt1 0.05 0.07 0.07 0.18 0.18 0.07 18 hellt 2 nil nil nil nil nil nil 19 ganz 0.24 1.05 0.38 0.04 0.03 0.15 20 ganz 0.22 0.69 1.96 0.28 0.20 0.22
181
Schubert Vibrato Onset Subjects 5-6
word SO 5 SSO 5 SLSO 5 SO 6 SSO 6 SLSO 6 1 Dies 0.28 0.20 0.19 0.23 0.25 0.23 2 Au 0.23 0.18 0.44 0.04 0.22 0.04 3 zelt 0.16 0.18 0.04 0.20 0.26 0.38 4 Dei 0.21 0.05 0.44 0.20 0.22 0.23 5 glanz 0.23 0.22 0.22 0.30 0.32 0.27 6 lein 0.06 0.06 0.19 0.05 0.04 0.04 7 hellt1 0.06 0.07 0.21 0.25 0.10 0.21 8 hellt 2 nil nil nil nil nil nil 9 ganz 0.24 0.19 0.04 0.21 0.08 0.04
10 ganz 0.27 0.33 0.20 0.20 0.21 0.22 11 Dies 0.26 0.19 0.26 0.22 0.22 0.08 12 Au 0.04 0.20 0.18 0.19 0.20 0.05 13 zelt 0.19 0.17 0.21 0.18 0.20 0.25 14 Dei 0.23 0.05 0.06 0.39 0.04 0.21 15 glanz 0.23 0.22 0.18 0.04 0.41 0.05 16 lein 0.06 0.05 0.38 0.05 0.06 0.04 17 hellt1 0.08 0.06 0.07 0.08 0.07 0.11 18 hellt 2 nil nil nil nil nil nil 19 ganz 0.23 0.19 0.24 0.22 0.44 0.05 20 ganz 0.23 0.57 0.66 0.19 0.20 0.19
182
Mozart Vibrato Onset Subjects 1-2
word MO 1 MSO 1 MLSO 1 MO 2 MSO 2
MLSO 2
1 den 0.05 0.23 0.24 0.07 Absent Absent 2 te 0.06 0.07 0.07 0.08 0.07 0.34 3 la 0.07 0.06 0.06 0.21 0.27 Absent 4 cal 0.03 0.22 0.22 0.22 0.23 Absent 5 ma nil nil nil nil nil nil 6 re 0.04 0.41 0.26 0.17 0.04 0.24 7 sti 0.05 0.20 0.22 0.05 0.05 0.06 8 se 0.18 0.43 0.21 0.26 0.41 0.22 9 gno nil nil nil nil nil nil
10 gno e nil nil nil nil nil nil 11 mor nil nil nil nil nil nil 12 den 0.05 0.19 0.22 0.24 0.05 Absent 13 te 0.04 0.35 0.05 0.07 0.06 0.25 14 la 0.08 0.09 0.08 0.08 0.38 0.26 15 cal 0.04 0.04 0.05 0.25 0.07 Absent 16 ma nil nil nil nil nil nil 17 al 0.07 0.21 0.45 0.29 Absent 0.23 18 ma nil nil nil nil nil nil 19 si 0.05 0.19 0.23 0.22 0.24 0.04 20 des 0.07 0.07 Absent 0.24 0.26 0.22 21 sti 0.20 0.05 0.07 0.06 0.05 0.06 22 re 0.07 0.39 0.06 0.03 0.06 0.24 23 sti 0.05 0.03 0.59 0.05 0.05 0.07 24 se 0.04 0.55 0.54 0.04 0.25 0.40 25 gno nil nil nil nil nil nil 26 gno e nil nil nil nil nil nil 27 mor nil nil nil nil nil nil 28 mor 2 nil nil nil nil nil nil 29 res 0.06 0.06 0.23 0.06 Absent Absent 30 sti 0.05 0.05 0.06 0.05 0.24 0.29 31 se 0.25 0.56 0.21 0.20 0.22 0.23 32 gno nil nil nil nil nil nil 33 gno e nil nil nil nil nil nil 34 ti 0.21 Absent 0.05 0.25 0.05 0.22 35 mor nil nil nil nil nil nil 36 di 0.22 0.43 0.46 0.20 Absent 0.20 37 mor nil nil nil nil nil nil
183
Mozart Vibrato Onset Subjects 3-4
word MO 3 MSO 3 MLSO 3 MO 4 MSO 4
MLSO 4
1 den 0.25 0.42 Absent 0.22 Absent 0.22 2 te 0.20 0.19 Absent 0.05 0.36 0.37 3 la 0.06 0.22 0.05 0.21 0.22 0.05 4 cal 0.37 0.19 Absent 0.19 0.19 0.22 5 ma nil nil nil nil nil nil 6 re 0.19 Absent Absent 0.22 0.37 0.23 7 sti 0.20 0.20 0.21 0.06 0.38 0.41 8 se 0.21 0.20 0.25 0.04 Absent 0.23 9 gno nil nil nil nil nil nil
10 gno e nil nil nil nil nil nil 11 mor nil nil nil nil nil nil 12 den 0.06 0.04 0.25 0.06 Absent 0.21 13 te 0.03 0.20 0.23 0.06 0.39 0.21 14 la 0.19 0.18 0.11 0.05 0.21 0.22 15 cal 0.21 0.16 0.41 0.04 0.34 0.40 16 ma nil nil nil nil nil nil 17 al 0.10 0.22 0.22 0.07 0.21 0.21 18 ma nil nil nil nil nil nil 19 si 0.20 Absent 0.40 0.05 Absent 0.19 20 des 0.22 0.22 0.41 0.04 Absent Absent 21 sti 0.21 0.20 Absent 0.05 0.20 0.41 22 re 0.19 0.58 0.24 0.05 0.22 Absent 23 sti 0.21 0.22 0.38 0.21 0.05 0.38 24 se 0.20 0.38 0.20 0.20 Absent 0.44 25 gno nil nil nil nil nil nil 26 gno e nil nil nil nil nil nil 27 mor nil nil nil nil nil nil 28 mor 2 nil nil nil nil nil nil 29 res 0.23 0.25 0.26 0.18 Absent 0.25 30 sti 0.21 0.20 0.24 0.22 0.36 0.23 31 se 0.22 Absent 0.04 0.23 0.39 Absent 32 gno nil nil nil nil nil nil 33 gno e nil nil nil nil nil nil 34 ti 0.17 0.19 0.18 0.05 Absent 0.23 35 mor nil nil nil nil nil nil 36 di 0.23 Absent Absent 0.23 Absent Absent 37 mor nil nil nil nil nil nil
184
Mozart Vibrato Onset Subjects 5-6
word MO 5 MSO 5 MLSO 5 MO 6 MSO 6
MLSO 6
1 den 0.07 0.21 0.21 0.25 Absent 0.21 2 te 0.06 Absent 0.16 0.18 Absent Absent 3 la 0.05 0.50 0.19 0.20 Absent 0.23 4 cal 0.05 Absent 0.04 0.05 Absent 0.25 5 ma nil nil nil nil nil nil 6 re 0.06 0.23 Absent 0.04 Absent 0.23 7 sti 0.05 0.36 0.21 0.05 Absent 0.41 8 se 0.22 0.35 0.18 0.05 Absent 0.21 9 gno nil nil nil nil nil nil
10 gno e nil nil nil nil nil nil 11 mor nil nil nil nil nil nil 12 den 0.05 0.06 0.54 0.06 Absent 0.19 13 te 0.05 0.19 0.18 0.03 0.42 0.19 14 la 0.04 0.06 0.20 0.05 0.23 Absent 15 cal 0.03 0.17 0.19 0.15 Absent 0.19 16 ma nil nil nil nil nil nil 17 al 0.09 0.06 0.37 0.08 0.23 0.12 18 ma nil nil nil nil nil nil 19 si 0.06 0.19 0.35 0.04 Absent 0.24 20 des 0.23 0.20 0.21 0.04 Absent 0.25 21 sti 0.19 0.18 0.18 0.04 0.23 0.41 22 re 0.04 0.06 0.06 0.06 Absent 0.25 23 sti 0.20 0.39 0.39 0.04 0.32 0.42 24 se 0.07 0.36 0.19 0.21 Absent Absent 25 gno nil nil nil nil nil nil 26 gno e nil nil nil nil nil nil 27 mor nil nil nil nil nil nil 28 mor 2 nil nil nil nil nil nil 29 res 0.06 0.05 0.20 0.07 Absent Absent 30 sti 0.07 0.35 0.38 0.04 Absent Absent 31 se 0.05 0.43 0.34 0.21 Absent Absent 32 gno nil nil nil nil nil nil 33 gno e nil nil nil nil nil nil 34 ti 0.17 0.18 0.33 0.16 0.18 Absent 35 mor nil nil nil nil nil nil 36 di 0.15 0.13 0.16 0.24 Absent Absent 37 mor nil nil nil nil nil nil
185
Appendix J: Vibrato extent raw data
Schubert Vibrato Extent Subjects 1-2
word SO 1 SSO 1 SLSO 1 SO 2 SSO 2 SLSO 2 1 Dies 0.87 0.82 0.80 1.22 1.22 0.87 2 Au 0.97 0.46 0.77 1.33 1.00 0.92 3 zelt 1.20 0.70 0.88 1.20 1.05 1.12 4 Dei 1.20 0.48 0.63 1.26 1.19 1.17 5 glanz 1.29 0.85 0.76 1.41 1.29 1.52 6 lein 1.14 0.67 0.63 1.32 1.16 1.17 7 hellt1 0.94 0.44 0.39 1.36 0.92 0.78 8 hellt 2 0.59 0.44 0.49 1.07 0.63 0.71 9 ganz 1.11 0.47 0.45 1.57 1.33 0.92
10 ganz 1.10 0.53 0.65 1.26 1.36 0.78 11 Dies 0.76 0.65 0.42 1.27 1.40 0.71 12 Au 1.12 0.65 0.70 1.20 1.22 0.84 13 zelt 1.13 0.81 0.80 1.33 1.34 0.93 14 Dei 0.99 0.52 0.50 1.56 1.24 0.77 15 glanz 1.27 0.69 0.66 1.54 1.36 1.19 16 lein 1.07 0.57 0.53 1.29 1.30 0.62 17 hellt1 0.82 0.46 0.42 1.36 1.07 0.56 18 hellt 2 0.61 0.42 0.49 1.12 0.98 0.65 19 ganz 0.94 0.33 0.43 1.51 1.57 0.71 20 ganz 0.79 0.61 0.72 1.28 1.20 0.61
186
Schubert Vibrato Extent Subjects 3-4
word SO 3 SSO 3 SLSO 3 SO 4 SSO 4 SLSO 4 1 Dies 0.93 absent 0.33 1.62 0.38 0.78 2 Au 1.40 0.48 0.46 1.82 0.75 0.54 3 zelt 1.45 0.70 0.58 1.52 0.58 0.78 4 Dei 1.60 0.86 0.39 1.68 0.60 0.62 5 glanz 1.31 0.77 0.71 1.63 0.56 0.69 6 lein 1.25 0.83 0.68 1.33 0.66 0.69 7 hellt1 1.21 0.67 0.61 1.21 0.46 0.46 8 hellt 2 1.07 0.51 0.60 1.04 0.44 0.34 9 ganz 1.30 0.60 0.78 1.66 0.69 0.64
10 ganz 1.56 0.36 0.40 1.62 0.81 0.68 11 Dies 1.28 absent 0.30 1.14 0.41 0.43 12 Au 1.47 0.30 0.35 1.62 0.90 0.73 13 zelt 1.66 0.38 0.71 1.62 1.00 0.88 14 Dei 1.96 0.74 0.69 1.63 0.83 0.65 15 glanz 1.46 0.74 0.54 1.52 0.68 0.62 16 lein 1.17 0.75 0.61 1.35 0.56 0.53 17 hellt1 1.18 0.60 0.40 1.16 0.31 0.37 18 hellt 2 0.96 0.45 0.34 1.00 0.29 0.33 19 ganz 1.09 0.30 0.43 1.37 0.68 0.41 20 ganz 1.18 0.81 0.34 1.23 0.55 0.48
187
Schubert Vibrato Extent Subjects 5-6
word SO 5 SSO 5 SLSO 5 SO 6 SSO 6 SLSO 6 1 Dies 1.17 0.94 0.37 0.62 0.57 0.63 2 Au 1.36 1.13 0.64 0.68 0.72 0.83 3 zelt 1.23 1.11 0.58 0.89 0.67 0.65 4 Dei 1.11 0.98 0.75 1.13 0.86 0.74 5 glanz 1.19 0.92 0.67 0.79 0.72 0.77 6 lein 1.11 0.83 0.62 1.11 0.89 0.73 7 hellt1 0.83 0.54 0.62 1.17 1.01 0.76 8 hellt 2 0.90 0.74 0.67 1.38 0.81 1.08 9 ganz 0.89 0.64 0.80 0.91 0.65 0.65
10 ganz 0.77 0.84 0.78 0.98 0.77 0.94 11 Dies 1.27 1.03 0.96 0.74 0.71 0.42 12 Au 1.21 1.03 0.84 0.66 0.62 0.68 13 zelt 1.22 1.13 0.82 0.84 0.68 0.80 14 Dei 1.28 0.82 0.73 0.87 0.75 0.85 15 glanz 1.19 0.74 0.77 0.77 0.74 0.77 16 lein 1.12 0.78 0.72 1.07 1.02 0.78 17 hellt1 0.85 0.60 0.68 0.94 1.02 1.03 18 hellt 2 0.82 0.73 0.72 0.79 0.77 1.27 19 ganz 0.87 0.72 0.78 0.71 0.61 0.74 20 ganz 0.96 0.76 0.73 1.00 absent 1.07
188
Mozart Vibrato Extent Subjects 1-2
word MO 1 MSO 1 MLSO 1 MO 2 MSO 2
MLSO 2
1 den 1.22 0.72 0.80 0.93 absent absent 2 te 1.21 0.69 0.62 1.21 0.70 0.52 3 la 1.21 0.91 0.65 1.09 0.60 absent 4 cal 1.02 0.96 0.80 1.09 0.29 absent 5 ma 1.19 0.83 0.71 1.20 0.75 0.50 6 re 0.85 0.61 0.59 1.00 0.34 0.50 7 sti 1.23 0.84 0.70 1.01 0.68 0.76 8 se 1.25 0.72 0.50 1.42 0.68 0.70 9 gno 1.23 0.90 0.89 1.18 1.06 0.77
10 gno e 1.27 0.55 0.49 1.22 1.03 0.88 11 mor 1.29 0.60 0.65 1.18 0.94 0.87 12 den 1.02 0.39 0.57 1.13 0.23 absent 13 te 1.26 0.74 0.46 1.32 1.05 0.93 14 la 1.29 0.69 0.74 1.29 0.70 0.57 15 cal 1.23 0.65 0.71 1.28 0.46 absent 16 ma 1.48 0.74 0.80 1.28 0.78 0.89 17 al 0.90 0.40 0.36 1.16 absent 0.35 18 ma 1.20 0.69 0.88 1.58 0.86 0.86 19 si 1.23 0.67 0.77 1.37 0.81 0.80 20 des 1.29 0.57 absent 1.28 0.91 0.75 21 sti 1.38 0.67 0.69 1.45 0.89 0.93 22 re 1.19 0.63 0.45 1.15 0.91 0.53 23 sti 1.38 0.82 0.59 1.34 0.79 0.73 24 se 1.02 0.70 0.40 1.25 0.74 0.79 25 gno 1.25 0.87 0.78 1.34 0.90 0.86 26 gno e 0.99 0.66 0.69 1.35 0.93 0.86 27 mor 1.02 0.60 0.71 1.38 0.97 0.84 28 mor 2 1.20 0.71 0.68 1.31 0.81 0.95 29 res 1.05 0.59 0.39 1.10 absent absent 30 sti 1.17 0.69 0.60 1.37 0.78 0.56 31 se 1.10 0.47 0.35 1.20 0.52 0.77 32 gno 1.18 0.77 0.89 1.31 0.79 0.78 33 gno e 1.14 0.61 0.84 1.69 0.91 0.74 34 ti 0.83 absent 0.49 1.46 1.00 0.80 35 mor 0.95 0.72 0.54 1.13 0.59 0.77 36 di 1.21 0.48 0.33 1.28 absent 0.30 37 mor 1.02 0.51 0.63 1.26 0.90 0.69
189
Mozart Vibrato Extent Subjects 3-4
word MO 3 MSO 3 MLSO 3 MO 4 MSO 4
MLSO 4
1 den 1.90 1.15 absent 1.46 absent 0.59 2 te 1.74 0.34 absent 1.63 0.53 0.68 3 la 1.61 0.36 0.63 1.52 0.50 0.58 4 cal 1.53 0.69 absent 1.43 0.50 0.38 5 ma 1.36 0.92 0.76 1.48 0.58 0.58 6 re 1.93 absent absent 1.64 0.37 0.53 7 sti 1.55 0.80 0.98 1.47 0.74 0.59 8 se 1.68 0.80 0.83 1.53 absent 0.62 9 gno 1.81 1.23 1.00 1.76 0.78 1.02
10 gno e 1.59 0.78 1.11 1.38 0.72 0.62 11 mor 1.67 0.52 0.67 1.62 0.87 0.44 12 den 1.27 0.60 0.24 1.30 absent 0.51 13 te 1.65 1.09 0.76 1.79 0.68 0.67 14 la 2.04 0.55 0.71 1.29 0.52 0.41 15 cal 1.65 1.09 0.85 1.33 0.74 0.39 16 ma 1.68 0.63 0.61 1.50 0.61 0.83 17 al 1.35 0.35 0.31 1.15 0.34 0.44 18 ma 1.33 0.90 1.00 1.68 0.64 0.99 19 si 1.65 absent 0.75 1.32 absent 0.54 20 des 1.77 0.72 1.07 1.41 absent absent 21 sti 1.23 0.90 absent 1.58 0.60 0.71 22 re 1.62 0.84 1.02 1.24 0.53 absent 23 sti 1.45 1.08 1.13 1.46 0.69 0.54 24 se 1.81 0.76 0.43 1.47 absent 0.54 25 gno 1.60 0.69 absent 1.56 absent 0.67 26 gno e 1.85 1.01 1.36 1.51 0.72 0.70 27 mor 1.68 0.51 1.24 1.51 0.44 absent 28 mor 2 2.07 0.95 0.94 1.64 0.72 0.49 29 res 1.56 1.12 0.36 1.36 absent 0.55 30 sti 1.61 1.16 0.80 1.57 0.72 0.62 31 se 1.64 absent 0.81 1.39 0.49 absent 32 gno 1.22 0.89 0.93 1.56 absent 0.76 33 gno e 1.70 1.06 1.23 1.54 0.46 0.49 34 ti 1.48 0.86 0.87 0.79 0.56 35 mor 1.78 0.74 0.30 1.63 0.43 0.58 36 di 1.64 absent absent 1.38 absent absent 37 mor 1.74 0.90 0.58 1.80 0.73 0.54
190
Mozart Vibrato Extent Subjects 5-6
word MO 5 MSO 5 MLSO 5 MO 6 MSO 6
MLSO 6
1 den 1.04 0.72 0.60 0.94 absent 0.33 2 te 1.24 absent 0.62 0.96 absent absent 3 la 0.99 0.63 0.64 0.78 absent 0.31 4 cal 1.05 absent 0.44 0.93 absent 0.47 5 ma 1.10 0.67 0.76 1.10 absent 0.77 6 re 1.09 0.68 absent 0.78 absent 0.41 7 sti 1.03 0.67 0.64 0.69 absent 0.58 8 se 0.85 0.69 0.63 0.72 absent 0.58 9 gno 1.12 0.71 0.74 0.95 0.51 0.84
10 gno e 0.88 0.55 0.50 0.86 absent 0.57 11 mor 1.06 0.66 0.62 0.92 0.59 0.84 12 den 0.99 0.71 0.62 1.21 absent 1.00 13 te 1.04 0.79 0.67 0.67 0.42 0.57 14 la 0.93 0.70 0.58 0.61 0.50 absent 15 cal 0.85 0.72 0.74 0.72 absent 0.50 16 ma 1.13 0.85 0.87 0.94 0.77 0.72 17 al 0.96 0.65 0.69 1.36 0.32 1.22 18 ma 1.12 0.81 0.54 0.80 absent 0.54 19 si 1.08 0.74 0.64 0.79 absent 0.56 20 des 0.93 0.57 0.52 0.78 absent 0.65 21 sti 1.01 0.66 0.69 0.78 0.62 0.93 22 re 1.04 0.59 0.56 1.06 absent 0.58 23 sti 1.15 0.93 0.84 0.85 0.40 0.64 24 se 1.01 0.89 0.44 0.75 absent absent 25 gno 1.06 0.81 0.63 0.76 absent absent 26 gno e 0.80 0.84 0.64 0.86 0.47 0.63 27 mor 0.83 0.56 0.39 0.88 0.36 0.83 28 mor 2 1.12 0.66 0.65 0.83 0.56 0.96 29 res 0.99 0.70 0.52 0.97 absent absent 30 sti 0.96 0.85 0.66 0.78 absent absent 31 se 1.03 0.85 0.43 0.64 absent absent 32 gno 0.97 0.66 0.57 0.76 absent 0.54 33 gno e 0.76 0.91 0.40 0.69 0.57 0.72 34 ti 0.97 0.68 0.52 0.63 0.45 absent 35 mor 0.94 0.64 0.62 0.90 0.59 0.60 36 di 0.86 0.75 0.35 0.76 absent absent 37 mor 1.01 0.61 0.61 1.00 0.66 0.77
191
Appendix K: Vibrato rate raw data
Schubert Vibrato Rate Subjects 1-2
word SO 1 SSO 1 SLSO 1 SO 2 SSO 2 SLSO 2 1 Dies 6.78 6.02 5.65 5.82 5.30 5.13 2 Au 5.59 4.76 5.56 5.30 4.88 5.02 3 zelt 5.59 5.73 5.83 5.15 5.07 5.07 4 Dei 5.60 5.56 5.38 5.20 5.04 5.30 5 glanz 5.93 5.12 5.80 5.21 5.14 5.23 6 lein 5.80 5.42 5.61 5.32 5.15 5.39 7 hellt1 5.98 5.68 5.63 5.25 4.95 5.26 8 hellt 2 5.69 5.74 5.66 5.06 5.02 5.04 9 ganz 5.69 5.66 6.25 5.17 5.15 5.42
10 ganz 5.65 5.52 5.45 5.03 4.96 4.97 11 Dies 5.80 6.28 6.25 5.73 5.21 5.24 12 Au 5.33 5.73 5.92 5.22 5.03 4.88 13 zelt 5.64 5.91 5.43 5.21 5.04 5.17 14 Dei 6.17 5.31 5.38 5.32 4.98 5.19 15 glanz 5.84 5.89 5.69 5.42 5.13 5.28 16 lein 5.86 5.39 5.36 5.48 5.20 5.27 17 hellt1 5.95 5.40 5.68 5.26 5.25 5.23 18 hellt 2 5.68 5.60 5.20 5.14 5.06 5.24 19 ganz 5.76 5.32 5.56 5.15 4.96 5.24 20 ganz 5.39 5.43 5.41 4.19 4.73 5.13
192
Schubert Vibrato Rate Subjects 3-4
word SO 3 SSO 3 SLSO 3 SO 4 SSO 4 SLSO 4 1 Dies 5.77 absent 6.25 6.01 7.49 6.41 2 Au 5.46 5.78 5.77 5.78 5.61 5.89 3 zelt 5.81 5.74 5.57 5.92 6.05 5.80 4 Dei 5.44 5.46 5.17 5.67 5.88 5.61 5 glanz 5.71 5.76 6.40 5.52 5.30 5.52 6 lein 5.62 5.76 5.75 5.68 5.56 5.79 7 hellt1 5.80 5.77 5.85 5.59 5.27 5.57 8 hellt 2 6.01 6.24 5.95 5.57 5.39 5.66 9 ganz 5.41 5.68 5.49 5.76 5.58 5.59
10 ganz 5.33 5.58 5.44 5.43 5.25 5.20 11 Dies 6.14 absent 5.49 6.17 6.18 6.71 12 Au 5.69 4.94 5.34 5.66 5.41 5.40 13 zelt 5.75 5.37 5.28 5.91 5.83 5.58 14 Dei 5.46 5.41 5.33 5.64 5.45 5.51 15 glanz 5.78 5.77 5.45 5.51 5.59 5.63 16 lein 5.77 5.77 5.75 5.69 5.82 5.69 17 hellt1 5.77 5.84 5.84 5.42 5.69 5.34 18 hellt 2 5.98 5.82 6.24 5.51 5.37 5.26 19 ganz 5.48 5.60 5.78 5.60 5.53 5.72 20 ganz 5.32 5.77 4.46 5.16 5.79 5.14
193
Schubert Vibrato Rate Subjects 5-6
word SO 5 SSO 5 SLSO 5 SO 6 SSO 6 SLSO 6 1 Dies 7.08 7.10 8.49 7.24 6.12 6.23 2 Au 7.02 7.44 6.78 6.76 5.76 6.08 3 zelt 6.61 6.71 7.30 6.12 5.75 6.66 4 Dei 6.86 6.80 6.62 5.38 5.18 6.54 5 glanz 6.42 6.65 6.71 7.06 6.41 7.44 6 lein 6.80 6.66 6.38 6.20 5.34 5.35 7 hellt1 6.33 6.98 6.10 5.55 5.16 5.15 8 hellt 2 6.33 6.37 6.43 5.57 5.69 5.11 9 ganz 6.63 6.45 6.55 5.20 4.88 5.24
10 ganz 6.59 6.61 6.65 5.06 4.90 5.21 11 Dies 6.50 7.14 7.27 6.69 5.57 6.58 12 Au 7.07 6.98 6.72 6.82 6.55 6.77 13 zelt 6.85 6.82 6.44 6.35 5.15 6.23 14 Dei 6.99 6.84 6.57 6.10 5.17 5.32 15 glanz 6.95 6.68 6.34 6.66 6.40 7.00 16 lein 6.87 6.70 6.55 5.93 5.23 5.02 17 hellt1 6.51 6.72 6.30 5.24 5.35 5.23 18 hellt 2 6.47 6.82 6.47 5.16 5.37 5.65 19 ganz 6.66 6.58 6.61 4.99 5.01 5.35 20 ganz 6.53 6.37 6.37 5.20 5.09 5.37
194
Mozart Vibrato Rate Subjects 1-2
word MO 1 MSO 1 MLSO 1 MO 2 MSO 2
MLSO 2
1 den 5.93 5.91 5.57 5.40 absent absent 2 te 5.88 5.86 5.89 5.50 5.23 5.56 3 la 5.92 5.85 5.77 5.75 5.43 absent 4 cal 5.70 5.88 5.99 5.94 5.88 absent 5 ma 5.86 5.77 5.79 5.51 5.64 5.13 6 re 5.96 6.49 5.77 5.72 5.57 5.56 7 sti 6.06 6.09 6.12 5.36 5.19 5.18 8 se 6.28 6.13 5.63 5.90 5.56 5.26 9 gno 5.95 5.82 5.81 5.27 5.19 5.11
10 gno e 5.95 5.68 5.77 5.50 4.96 5.18 11 mor 5.84 5.42 5.59 5.36 5.23 5.11 12 den 6.59 6.97 6.74 6.19 5.83 absent 13 te 5.63 5.73 5.59 5.49 5.34 5.46 14 la 5.91 5.85 5.75 5.89 5.41 5.66 15 cal 5.93 5.66 5.96 5.73 5.52 absent 16 ma 6.16 5.93 5.96 5.67 5.13 5.41 17 al 6.19 6.00 6.71 6.50 absent 5.56 18 ma 5.85 6.02 5.65 5.47 5.13 5.36 19 si 6.06 5.97 5.92 5.83 5.26 5.56 20 des 6.12 6.11 absent 5.46 5.13 5.46 21 sti 5.89 6.05 6.17 5.44 5.03 5.18 22 re 6.31 5.78 6.11 5.72 5.38 5.41 23 sti 5.95 6.07 6.45 5.27 5.19 5.69 24 se 5.91 6.13 6.17 5.72 5.28 5.41 25 gno 5.58 5.80 5.74 5.27 4.84 5.13 26 gno e 5.89 5.56 5.67 5.78 5.17 5.45 27 mor 5.70 5.76 5.70 5.20 4.92 5.10 28 mor 2 5.75 5.58 5.72 5.64 5.27 5.27 29 res 6.29 5.97 5.18 6.01 absent absent 30 sti 5.90 6.00 5.86 5.51 5.56 5.26 31 se 5.98 6.06 6.11 6.02 5.64 5.46 32 gno 5.74 5.69 5.89 5.38 5.60 5.00 33 gno e 5.68 5.92 5.65 5.66 5.69 5.22 34 ti 5.95 absent 6.32 5.64 5.46 5.09 35 mor 5.88 5.68 5.43 5.45 4.89 5.04 36 di 6.37 6.41 7.30 6.02 absent 6.67 37 mor 5.83 5.86 5.68 5.20 5.01 5.16
195
Mozart Vibrato Rate Subjects 3-4
word MO 3 MSO 3 MLSO 3 MO 4 MSO 4
MLSO 4
1 den 5.74 6.17 absent 6.11 absent 6.25 2 te 5.58 5.78 absent 5.96 5.88 6.27 3 la 5.92 6.94 5.57 5.63 5.57 5.57 4 cal 5.68 5.76 absent 6.05 5.83 5.00 5 ma 5.70 5.32 5.26 5.57 5.69 5.48 6 re 5.65 absent absent 6.02 5.26 5.28 7 sti 5.79 5.82 5.79 5.46 5.26 5.41 8 se 5.93 5.39 5.72 5.86 absent 5.56 9 gno 5.45 5.59 5.88 5.58 5.44 5.46
10 gno e 5.57 5.69 5.80 5.81 5.10 5.26 11 mor 5.30 5.55 5.26 5.33 5.16 5.09 12 den 5.97 5.91 6.70 6.28 absent 6.07 13 te 5.37 5.23 5.56 5.85 5.57 5.69 14 la 5.48 5.89 5.57 6.03 5.72 6.27 15 cal 5.63 5.53 5.28 5.60 5.41 6.07 16 ma 5.71 5.56 6.25 5.85 5.57 5.57 17 al 5.82 6.13 6.07 6.21 6.46 6.07 18 ma 5.16 5.26 5.57 5.74 5.41 5.26 19 si 5.33 absent 5.26 5.58 absent 5.36 20 des 5.42 5.24 5.26 5.51 absent absent 21 sti 5.46 5.64 absent 5.13 5.77 5.57 22 re 5.67 5.68 5.79 5.64 5.72 absent 23 sti 5.62 5.94 5.56 5.31 5.62 5.26 24 se 5.58 5.81 5.26 5.63 absent 5.26 25 gno 5.26 6.17 absent 5.41 absent 5.29 26 gno e 5.38 5.28 5.27 5.58 5.90 5.48 27 mor 5.62 5.51 5.27 5.31 5.90 28 mor 2 5.50 5.66 5.26 5.20 5.41 5.26 29 res 5.67 5.64 5.90 5.56 absent 5.00 30 sti 5.75 5.60 5.66 5.25 5.26 5.28 31 se 5.52 absent 5.26 5.63 5.88 absent 32 gno 5.16 5.42 5.26 5.27 absent 5.00 33 gno e 5.37 5.33 5.57 5.69 5.41 6.07 34 ti 5.64 5.56 5.37 absent 5.05 35 mor 5.45 5.43 5.90 5.24 5.23 5.05 36 di 6.31 absent absent 6.17 absent absent 37 mor 5.39 5.17 5.37 5.37 6.25 5.88
196
Mozart Vibrato Rate Subjects 5-6
word MO 5 MSO 5 MLSO 5 MO 6 MSO 6
MLSO 6
1 den 7.23 6.83 6.90 6.45 absent 5.88 2 te 7.04 absent 6.83 6.90 absent absent 3 la 7.04 7.42 6.83 7.08 absent 5.26 4 cal 6.52 absent 6.85 7.24 absent 6.07 5 ma 6.61 6.22 6.97 7.33 absent 6.28 6 re 6.83 6.70 absent 6.75 absent 7.11 7 sti 6.68 6.90 7.06 7.22 absent 7.69 8 se 6.77 6.90 6.46 7.39 absent 7.50 9 gno 6.85 6.83 6.83 8.22 6.70 7.35
10 gno e 6.63 6.63 6.86 7.12 absent 6.62 11 mor 6.55 6.58 6.68 7.38 6.87 7.01 12 den 6.79 6.46 7.69 7.13 absent 6.25 13 te 6.96 6.97 6.70 7.81 7.52 6.48 14 la 6.63 6.95 6.83 6.99 5.96 absent 15 cal 6.76 6.68 6.79 8.14 absent 5.61 16 ma 6.78 6.71 6.69 7.63 5.88 6.97 17 al 6.69 6.27 7.74 6.96 6.67 6.46 18 ma 6.69 7.04 6.67 7.81 absent 6.97 19 si 6.95 6.98 6.97 7.35 absent 6.19 20 des 7.06 6.69 6.67 7.18 absent 6.90 21 sti 6.85 6.79 6.53 7.17 6.67 6.79 22 re 6.46 6.68 6.59 7.12 6.62 23 sti 6.68 6.25 6.70 6.82 6.48 8.28 24 se 6.80 6.46 6.55 7.32 absent absent 25 gno 6.90 7.14 6.67 7.53 absent absent 26 gno e 6.89 6.77 6.80 7.38 6.70 6.43 27 mor 6.87 6.83 6.67 6.72 4.55 6.19 28 mor 2 6.88 6.87 6.69 8.05 6.62 6.81 29 res 6.68 6.27 6.27 7.22 absent absent 30 sti 6.89 6.90 6.67 6.87 absent absent 31 se 7.10 7.42 6.90 7.31 absent absent 32 gno 6.75 6.81 6.46 7.47 absent 7.49 33 gno e 6.67 7.14 7.54 7.59 6.79 7.35 34 ti 6.80 7.01 6.83 7.84 6.25 absent 35 mor 6.55 6.68 6.87 6.65 4.98 4.95 36 di 6.92 6.97 6.97 7.51 absent absent 37 mor 6.69 6.63 6.58 6.22 5.53 6.03
197
Appendix L: Long term average spectra raw data
All Schubert Optimal
Schubert Optimal Raw Data Hz SO1 SO2 SO3 SO4 SO5 SO6
0 125 -57.87 -61.8 -60.29 -61.5 -57.66 -60.62 250 -47.65 -54.59 -49.29 -52.61 -48.14 -50.75 375 -41.88 -44.04 -41.14 -42.89 -40.37 -42.46 500 -37.69 -33.25 -35.86 -36.4 -34.06 -34.63 625 -35.39 -27.18 -30.76 -31.86 -33.44 -31.13 750 -37.22 -26.71 -28.49 -29.68 -39.8 -33.55 875 -42.07 -31.44 -31.26 -31.23 -51.57 -43.41
1000 -40.76 -42.22 -39.88 -38.01 -51.79 -52.13 1125 -38.9 -52.82 -45.33 -43.17 -54.08 -51.08 1250 -38.62 -50.22 -44.48 -44.12 -58.46 -52.57 1375 -39.69 -50.38 -44.76 -48.57 -56.82 -55.1 1500 -44.19 -53.71 -44.08 -53.39 -49.52 -51.49 1625 -50.22 -55.45 -44.35 -53.81 -46.29 -47.16 1750 -52.33 -54.4 -47.32 -53.33 -48.31 -45.17 1875 -56.36 -56.13 -52.81 -55.32 -55.02 -46.5 2000 -61.01 -59.3 -59.58 -56.71 -61.97 -51.66 2125 -63.6 -61.15 -59.93 -57.61 -64.26 -59.1 2250 -63.49 -63.86 -57.75 -60.85 -61.74 -64.3 2375 -62.03 -64.17 -56.87 -62.62 -59.77 -64.96 2500 -59.93 -62.67 -56.85 -61.76 -59.86 -63.34 2625 -56.31 -62.21 -57.66 -60.36 -57.66 -60.93 2750 -53.94 -61.46 -54.16 -57.1 -55.22 -61.57 2875 -53.18 -59.03 -50.69 -54.97 -55.4 -62.66 3000 -53.19 -54.14 -49.61 -54.86 -56.25 -59.47 3125 -54.37 -51.5 -50.62 -54.54 -57.19 -55.05 3250 -54.5 -52.18 -53.21 -54.38 -55.8 -51.44 3375 -54.92 -54.62 -55.33 -54.66 -54.56 -51.25 3500 -57.46 -56.6 -56.89 -55 -55.41 -53.98 3625 -61.27 -57.46 -60.02 -57.15 -57.46 -55.99 3750 -65.62 -58.11 -65.62 -60.78 -60.96 -58.13 3875 -71.04 -60.05 -73 -64.87 -64.16 -58.91 4000 -76.56 -62.09 -79.03 -68.58 -65.64 -59.87 4125 -80.66 -65.12 -81.54 -70.29 -65.81 -62.64 4250 -84.75 -69.59 -83.06 -73.1 -67.92 -66.52 4375 -89.38 -73.35 -84.56 -78.21 -74.03 -71.05
199
4500 -92.27 -77.43 -83.12 -82.58 -83.85 -72.78 4625 -90.69 -81.09 -79.24 -83.72 -89.52 -72.73 4750 -89.29 -82.65 -78.49 -81.75 -90.26 -74.44 4875 -90.16 -82.94 -80.32 -78.93 -91.62 -77.97 5000 -93.06 -84.34 -82.36 -78.99 -92.83 -81.62 5125 -94.76 -92.17 -85.46 -82.4 -92.1 -83.84 5250 -93.79 -92.45 -87.25 -86.29 -89.04 -85.11 5375 -92.35 -92.19 -85.87 -88.34 -85.15 -83.22 5500 -92.32 -91.21 -85.15 -86.64 -82.72 -81.77 5625 -93.8 -90.73 -83.22 -84.14 -82.72 -82.32 5750 -94.66 -90.38 -79.97 -81.17 -82.62 -80.9 5875 -94.01 -89.99 -78.12 -78.68 -82.78 -79 6000 -92.16 -89.01 -78.71 -78.34 -85.02 -79.02 6125 -89.68 -87.07 -80.43 -80.65 -87.59 -80.74 6250 -87.72 -85.07 -81.66 -85.52 -88.42 -81.29 6375 -86.54 -83.76 -83.1 -88.64 -88.3 -80.28 6500 -86.14 -83.18 -84.15 -88.4 -87.34 -79.5 6625 -87.09 -82.92 -84.72 -88.72 -86.78 -79.73 6750 -88 -82.29 -85.53 -89.75 -87.52 -82.01 6875 -88.27 -81.68 -86.53 -88.53 -88.88 -84.26 7000 -88.7 -81.71 -86.97 -87.93 -88.93 -83.44
200
All Schubert Sub-optimal
Schubert Sub-optimal Hz SSO1 SSO2 SSO3 SSO4 SSO5 SSO6
0 125 -60.59 -62.52 -61.33 -62.08 -58.3 -61.43 250 -52.06 -60.88 -52.24 -54.37 -48.71 -52.1 375 -46.02 -50.76 -44.17 -45.55 -42.92 -44.26 500 -42.11 -44.3 -38.64 -39.84 -38.91 -37.21 625 -39.71 -39.4 -34.09 -33.74 -38.31 -33.4 750 -40.67 -38.84 -32.39 -30.04 -43.3 -34.89 875 -43.08 -43.82 -35.2 -31.34 -47.97 -42.66
1000 -41.62 -55.87 -43.11 -38.94 -48.25 -49.87 1125 -41.97 -64.74 -47.48 -45.75 -51.96 -49.33 1250 -44.48 -63.71 -47.41 -45.63 -55.74 -51.44 1375 -46.15 -63.3 -47.56 -48.29 -54.12 -53.01 1500 -49.5 -64.3 -45.29 -51.65 -50.34 -52.45 1625 -52.1 -64.42 -45.29 -51.35 -45.66 -50.25 1750 -53.26 -65.58 -49.52 -51.46 -44.03 -49.1 1875 -55.99 -69.27 -52.88 -53.25 -48.17 -50.32 2000 -57.6 -72.43 -56.93 -55.73 -58.15 -54.42 2125 -60.36 -74.78 -60.48 -55.21 -64.7 -59.67 2250 -64.42 -76.65 -61.81 -56.82 -65.83 -62.73 2375 -65.45 -77.54 -62.74 -57.34 -64.37 -63.01 2500 -63.59 -77.69 -65.43 -59.29 -64.56 -64.33 2625 -60.58 -78.44 -62.3 -63.23 -65.31 -63.15 2750 -59.68 -77.97 -60.71 -59.88 -60.63 -62.15 2875 -59.18 -77.33 -58.41 -58.71 -55.79 -62.81 3000 -59.78 -74.8 -55.57 -59.5 -54.2 -61.49 3125 -60.44 -71.56 -56.23 -56.9 -55.29 -57.73 3250 -61.26 -73.12 -58.04 -56.68 -58.08 -54.38 3375 -63.32 -77.12 -58.9 -59.06 -59.17 -52.44 3500 -65.06 -78.15 -60.78 -60.86 -58.18 -53.18 3625 -67.01 -78.7 -61.26 -62.11 -58.15 -55.35 3750 -71.47 -76.74 -64.04 -63.58 -59.7 -57.17 3875 -76.04 -77.34 -68.93 -62.79 -60.56 -58.27 4000 -79.17 -80.32 -71.31 -63.51 -60.28 -58.4 4125 -82.69 -82.62 -74.83 -67.27 -61.59 -59.53 4250 -85.48 -85.96 -79.34 -70.13 -65.67 -63.3 4375 -88.42 -87.33 -83.08 -74.97 -70.7 -68.46 4500 -86.53 -87.83 -85.69 -80.49 -73.31 -71.13 4625 -86.79 -87.62 -85.24 -81.48 -75.08 -71.36 4750 -88.43 -87.89 -84.92 -79.11 -78.24 -72.75
201
4875 -88.2 -91.78 -85.89 -78.47 -82.31 -75.83 5000 -89.84 -98.32 -88.29 -80.24 -83.14 -78.16 5125 -89.94 -101.46 -89.76 -84.39 -82.09 -81.53 5250 -89.14 -102.69 -90.05 -90.1 -82.55 -85.41 5375 -89.75 -102.64 -91.11 -89.81 -83.88 -87.65 5500 -88.78 -102.5 -92.52 -85.69 -83.43 -89.4 5625 -88.43 -103.29 -92.89 -84.9 -82.42 -91.08 5750 -89.63 -103.28 -92.49 -86.6 -83.52 -90.79 5875 -91.21 -102.64 -91.97 -89.18 -85.39 -89.16 6000 -94.39 -102.19 -90.89 -89.99 -86.61 -87.53 6125 -97.13 -101.2 -89.4 -88.56 -87.8 -86.63 6250 -96.25 -100.29 -88.01 -87.05 -89.32 -86.51 6375 -94.22 -98.92 -86.74 -87.78 -92.41 -85.37 6500 -94.21 -98.37 -86.49 -90.32 -94.52 -83.18 6625 -95.7 -97.91 -87.03 -90.9 -94.1 -81.6 6750 -97.04 -97.14 -88.26 -90.71 -93.54 -80.57 6875 -97.47 -96.06 -90.47 -90.94 -93.3 -80.52 7000 -97.83 -96 -92.25 -90.24 -93.12 -82.25
202
All Schubert Loud sub-optimal
Schubert Loud Sub-optimal Hz SLSO1 SLSO2 SLSO3 SLSO4 SLSO5 SLSO6
0 125 -59.42 -62.3 -61.1 -62.02 -58.14 -59.96 250 -49.93 -58.17 -51.83 -53.51 -48.69 -50.08 375 -44.11 -48.42 -43.11 -43.72 -43.53 -42.94 500 -40.5 -40.12 -37.49 -38.23 -40.08 -36.69 625 -38.13 -33.96 -32.92 -33.36 -39.04 -34.05 750 -38.82 -32.8 -30.6 -29.68 -41.36 -36.84 875 -40.7 -36.99 -32.92 -29.72 -42.9 -43.38
1000 -38.57 -48.33 -39.99 -35.53 -44.12 -43.37 1125 -38.58 -60.23 -43.15 -45.72 -47.96 -44.31 1250 -42.06 -56.55 -44.71 -46.18 -51.72 -47.51 1375 -44.63 -53.28 -46.84 -47.12 -51.01 -46.34 1500 -46.43 -51.2 -45.31 -49.55 -48.5 -44.62 1625 -46.53 -51.85 -44.69 -50.17 -43.92 -45.57 1750 -47.57 -53.76 -48.28 -49.61 -41.93 -42.82 1875 -49.95 -55.24 -52.34 -49.9 -45.69 -42.52 2000 -50.73 -59.38 -53.66 -50.32 -55.5 -47.59 2125 -53.09 -62.94 -55 -50.15 -61.95 -53.9 2250 -57.61 -64.31 -56.33 -52.77 -63.23 -56.66 2375 -59.37 -65.47 -56.57 -53.77 -63.78 -57.7 2500 -58.36 -66.09 -59.75 -54.19 -64.4 -58.69 2625 -54.55 -64.76 -59.04 -56.73 -63.73 -58.36 2750 -52.95 -64.53 -55.73 -59.17 -58.54 -58.06 2875 -52.45 -63.54 -52.53 -59.09 -54.24 -56.03 3000 -53.26 -61.6 -49.37 -58.81 -52.71 -53.17 3125 -53.86 -59.95 -48.71 -55.88 -53.15 -50.01 3250 -54 -59.5 -50.44 -54.49 -54.27 -47.67 3375 -55.64 -62.11 -53.13 -54.28 -54.75 -47.4 3500 -57.82 -62.82 -55.07 -54.96 -55.39 -47.76 3625 -60.74 -61.1 -56.39 -57.05 -56.7 -48.29 3750 -65.13 -58.39 -59.09 -57.93 -58.4 -49.82 3875 -69.33 -56.94 -65.26 -56.87 -58.2 -51.75 4000 -72.92 -58.87 -73.28 -56.95 -57.64 -53.58 4125 -76.92 -62 -76.5 -58.83 -58.99 -55.35 4250 -80.63 -65.26 -79.15 -60.63 -62.88 -58.3 4375 -86.44 -69.77 -82.72 -63.25 -67.55 -63.18 4500 -90.37 -71.51 -85.43 -67.88 -69.47 -67.45 4625 -87.79 -73.08 -83.78 -69.61 -70.97 -68.42 4750 -85.79 -76.53 -80.97 -68.72 -74.75 -70.63
203
4875 -85.1 -79.39 -80.24 -67.86 -80.34 -75 5000 -85.76 -83.53 -83.19 -68.27 -80.74 -78.8 5125 -84.39 -87.61 -88.88 -71.45 -78.52 -82.07 5250 -82.24 -87.38 -91.35 -77.77 -78.12 -84.64 5375 -83.12 -88.53 -90.1 -82.66 -78.86 -85.71 5500 -83.87 -89.45 -89.13 -80.48 -78.28 -87.08 5625 -84.26 -90.71 -89.12 -78.57 -77.26 -86.98 5750 -86.08 -91.79 -89.61 -80.11 -78.93 -83.88 5875 -86.82 -92.12 -88.28 -83.16 -83.74 -81.25 6000 -89.07 -92.12 -86.05 -84.51 -88.27 -80.33 6125 -92.79 -90.63 -84.14 -83.56 -89.9 -79.96 6250 -93.21 -89.64 -82.68 -82.2 -91.11 -79.26 6375 -91.13 -88.55 -81.78 -82.32 -92.94 -78.18 6500 -90.33 -87.52 -81.98 -84.45 -93.44 -77.9 6625 -90.9 -85.05 -81.96 -88.07 -92.8 -78.07 6750 -91.38 -82.55 -82.84 -89.25 -92.03 -78.36 6875 -90.67 -82.33 -85.72 -88 -91.99 -78.71 7000 -89.95 -83.88 -88.79 -86.49 -92.3 -78.76
204
All Mozart Optimal
Mozart Optimal Raw Data Hz MO1 MO2 MO3 MO4 MO5 MO6
0 125 -58.74 -59.97 -58.89 -58.95 -58.66 -60.4 250 -46.99 -48.98 -46.47 -47.97 -46.2 -49.58 375 -38.61 -40.74 -38.47 -39.6 -37.71 -41.41 500 -34.59 -36.86 -34.82 -35.39 -33.8 -37.58 625 -33.25 -33.2 -32.73 -33.73 -31.93 -35.89 750 -31.62 -31.67 -30.81 -32.54 -31.15 -35.26 875 -32.49 -35.56 -32.94 -34.18 -35.17 -38.98
1000 -37.86 -44.28 -39.06 -39.9 -43.23 -47.66 1125 -42.05 -49.43 -42.54 -45.76 -46.22 -52.85 1250 -44.83 -56.08 -46.71 -51.97 -52.2 -53.32 1375 -44.7 -54.63 -45.31 -55 -47.96 -49.34 1500 -41.65 -52.28 -42.05 -52.53 -43.68 -47.28 1625 -40.86 -51.56 -43.03 -51.42 -43.19 -48.4 1750 -44.64 -51.19 -47.38 -53.03 -45.34 -50.08 1875 -51.26 -51.75 -50.59 -56.83 -50.03 -51.92 2000 -55.15 -53.3 -53.51 -61.34 -54.98 -54.25 2125 -57.66 -55.49 -54.09 -63.76 -56.04 -56.14 2250 -59.15 -57.35 -54.32 -62.67 -56.77 -58.15 2375 -56.44 -60.03 -54.5 -61.02 -59.42 -62.94 2500 -54.17 -61.48 -54.75 -61.93 -60.11 -64.48 2625 -53.03 -61.89 -51.96 -62.29 -57.09 -64.5 2750 -51.96 -61.6 -48.1 -59.89 -54.59 -65.66 2875 -52.1 -59.68 -45.7 -56.97 -52.35 -62.38 3000 -52.14 -57.6 -44.89 -55.06 -51.39 -58.67 3125 -51.81 -56.11 -45.41 -54.53 -52.16 -57.56 3250 -51.94 -57.01 -47.09 -55.12 -54.9 -58.03 3375 -53.26 -58.09 -49.56 -55.84 -55.87 -57.64 3500 -56.25 -58.54 -51.76 -56.57 -54.8 -57.5 3625 -58.99 -60.09 -54.06 -58 -55.34 -58.1 3750 -62.18 -62.32 -58.44 -60.84 -57.29 -58.72 3875 -67.61 -63.78 -64.46 -64.09 -60.01 -59.56 4000 -72.79 -65.07 -69.47 -65.5 -63.46 -61.25 4125 -75.4 -67.35 -74.83 -66.18 -67.76 -63.26 4250 -78.74 -70.61 -79.15 -68.08 -71 -67.03 4375 -84.42 -73.83 -78.55 -72.55 -73.67 -70.67 4500 -88.46 -76.5 -77.29 -77.07 -74.27 -70.52 4625 -82.98 -80.42 -76.59 -79.02 -73.67 -71.33 4750 -79.29 -86.09 -75.68 -79.72 -75.68 -74.17
205
4875 -81.61 -90.87 -75.58 -79.23 -81.87 -76.91 5000 -87.97 -92.56 -77.84 -80.14 -86.99 -78.03 5125 -91.19 -92.25 -81.63 -83.43 -86.11 -79.53 5250 -91.58 -92.54 -84.18 -88.14 -84.72 -83.33 5375 -86.52 -92.2 -83.39 -88.87 -83.43 -88.17 5500 -81.52 -91.24 -81.72 -87.31 -83.01 -90.03 5625 -81.67 -90.72 -81.27 -85.14 -82.3 -89.7 5750 -86.66 -90.25 -81.26 -83.08 -81.22 -89.06 5875 -90.85 -89.91 -78.08 -81.72 -80.94 -87.9 6000 -90.9 -88.99 -73.96 -82.12 -80.5 -86.68 6125 -88.68 -87.08 -72.95 -83.69 -78.86 -85.71 6250 -84.24 -85.1 -74.29 -85.78 -78.21 -85.04 6375 -82.5 -83.75 -76.32 -86.94 -80.82 -84.9 6500 -84.38 -83.21 -78.89 -87.53 -83.62 -85.55 6625 -85.82 -82.95 -79.21 -88.79 -83.39 -86.04 6750 -85.59 -82.3 -77.33 -90.15 -82.99 -85.53 6875 -85.53 -81.72 -78.6 -90.17 -82.62 -85.16 7000 -85.26 -81.76 -82.11 -89.05 -83.05 -85.31
206
All Mozart Sub-optimal
Mozart Sub-optimal Hz MSO1 MSO 2 MSO3 MSO4 MSO5 MSO6
0 125 -60.12 -60.72 -60.08 -60.64 -60.47 -61.82 250 -49.82 -51.49 -49.47 -49.89 -49.86 -52.68 375 -42.63 -43.71 -41.72 -42.54 -42.43 -43.52 500 -40.01 -40.42 -38.54 -39.74 -39.83 -39.19 625 -39.25 -39.68 -37.56 -37.68 -39.38 -37.61 750 -38.13 -39.25 -37.12 -35.54 -39.09 -36.03 875 -39.09 -42.78 -39.79 -37.29 -42.07 -37.97
1000 -40.93 -51.19 -44.97 -40.79 -46.57 -44.84 1125 -44.83 -56.82 -49.08 -44.19 -50.79 -50.22 1250 -48.32 -60.96 -54.59 -50.87 -56.32 -54.84 1375 -47.13 -54.52 -53.5 -54.84 -52.8 -53.56 1500 -45.48 -51.81 -48.31 -53.22 -49.11 -48.83 1625 -47.35 -54.32 -47.89 -53.36 -48.83 -47.94 1750 -51.07 -56.43 -51.47 -54.08 -49.53 -51.36 1875 -54.63 -56.71 -55.22 -53.74 -52.29 -54.28 2000 -54.29 -57.13 -58.68 -52.53 -54.67 -55.79 2125 -54.99 -58.81 -59.56 -53.59 -56.57 -53.21 2250 -56.42 -61.25 -61.04 -55.81 -57.72 -52.69 2375 -59.91 -63.88 -62.75 -57.86 -61.15 -55.36 2500 -61.4 -63.92 -64.49 -61.43 -65.25 -60.16 2625 -60.14 -66.34 -62.34 -62.78 -65.51 -59.89 2750 -58.08 -67.9 -60.55 -64.08 -63.73 -60.49 2875 -54.63 -66.58 -57.92 -62.13 -60 -62.18 3000 -53.03 -64.72 -55.01 -58.37 -57.33 -56.57 3125 -55.55 -64.75 -54.47 -57.55 -58.01 -51.84 3250 -59.64 -66.95 -57.75 -60.68 -61.38 -52.2 3375 -60.24 -67.16 -61.54 -62.75 -60.75 -56.44 3500 -60.41 -66.44 -61.79 -61.76 -59.66 -57.27 3625 -61.63 -67.53 -62.98 -62.85 -59.92 -56.45 3750 -63.68 -68.5 -65.86 -64.11 -59.98 -58.12 3875 -67.42 -67.42 -69.23 -62.86 -59.82 -59.63 4000 -72.86 -67.44 -72.44 -62.46 -60.37 -59.59 4125 -78.96 -69.3 -77.42 -63.48 -62.65 -60.8 4250 -86.17 -72.23 -82.38 -66.45 -66.14 -63.32 4375 -87 -74.83 -84.89 -68.39 -67.78 -69.04 4500 -86.02 -76.56 -85.29 -70.47 -68.4 -73.22 4625 -88.69 -80.14 -83.63 -71.4 -70.31 -70.98 4750 -94.18 -87.34 -82.97 -73.21 -74.83 -70.3
207
4875 -94.6 -94.38 -83.98 -74.97 -80.1 -73.98 5000 -88.07 -96.83 -86.79 -74.14 -82.17 -79.03 5125 -83.11 -96.92 -87.93 -76.07 -81.99 -81.43 5250 -82.44 -96.44 -87.5 -79.26 -82.42 -82.61 5375 -85.43 -97.44 -89.05 -81.1 -83.46 -85.01 5500 -90.54 -98.08 -90.78 -82.53 -84.29 -87.3 5625 -93.1 -97.85 -91.06 -86.67 -84.73 -90.03 5750 -92.56 -96.77 -90.5 -89.9 -85.25 -89.99 5875 -90.19 -94.92 -90.18 -90.08 -85.44 -88.48 6000 -89.99 -93.54 -88.85 -86.97 -84.02 -88.02 6125 -92.37 -92.36 -86.73 -84.82 -83.58 -86.85 6250 -94 -91.26 -85.49 -85 -85.74 -84.49 6375 -93.84 -90.68 -84.6 -85.91 -89.01 -82.31 6500 -93.63 -89.31 -84.78 -87.68 -90.77 -80.99 6625 -92.83 -87.42 -86.54 -89.42 -91.21 -81.83 6750 -92.42 -86.8 -86.41 -89.9 -91.4 -83.68 6875 -92.63 -87.47 -83.74 -89.36 -91.12 -83.39 7000 -93.77 -88.69 -84.53 -88.14 -90.98 -81.49
208
All Mozart Loud sub-optimal
Mozart Loud Sub-optimal Hz MLSO1 MLSO2 MLSO3 MLSO4 MLSO5 MLSO6
0 125 -59.45 -60.04 -59.88 -60.41 -59.86 -61.12 250 -48.77 -48.99 -48.42 -49.19 -48.8 -50.06 375 -41.57 -40.89 -40.54 -41.87 -41.26 -40.64 500 -38.99 -37.67 -37.38 -39.63 -38.49 -35.82 625 -38.53 -34.65 -35.74 -38.54 -37.16 -33.42 750 -37.48 -32.26 -33.99 -36.86 -35.75 -31.56 875 -37.71 -35.76 -36.07 -38.04 -37.64 -33.37
1000 -38.73 -44.13 -40.88 -39.92 -40.62 -39.91 1125 -42.71 -48.9 -43.89 -43.19 -44.79 -45.63 1250 -47.48 -56.06 -48.92 -50.26 -51.27 -47.64 1375 -45.56 -50.86 -49.99 -52.95 -47.22 -44.17 1500 -43.39 -47.08 -48.11 -50.86 -43.46 -41.53 1625 -45.53 -49.04 -47.99 -51.22 -43.74 -42.16 1750 -49.29 -51.8 -51.49 -52.82 -45.23 -44.31 1875 -50.7 -49.34 -54.38 -52.78 -47.99 -45.43 2000 -50.16 -48.42 -56.36 -51.12 -50.09 -47.49 2125 -51.3 -50.52 -56.68 -52.18 -51.67 -47.02 2250 -52.08 -52.09 -55.82 -53.04 -52.56 -46.21 2375 -55.47 -55.06 -56.21 -53.56 -55.55 -48.88 2500 -59.11 -55.54 -58.48 -57.3 -57.5 -53.04 2625 -57.92 -56.57 -56.91 -58.67 -57.02 -53.95 2750 -55.85 -59.08 -55.23 -60.59 -57.47 -54.29 2875 -52.15 -55.37 -53.04 -60.73 -53.68 -53.38 3000 -50.07 -52.53 -50.55 -57.27 -50.21 -49.7 3125 -52.3 -54.42 -49.07 -55.81 -50.85 -46.72 3250 -57.12 -57.28 -51.78 -57.61 -54.09 -46.44 3375 -57.86 -56.64 -56.56 -58.41 -53.84 -48.33 3500 -58.08 -57.01 -56.61 -57.2 -53.01 -48.7 3625 -59.34 -58.44 -57.78 -58.18 -53.63 -49.61 3750 -61.67 -59.01 -62.26 -60 -53.92 -51.73 3875 -64.9 -56.88 -69.03 -59.47 -54.01 -52.58 4000 -69.3 -55.98 -73.06 -59.59 -54.22 -52.86 4125 -75.06 -58.48 -76.14 -60.6 -55.48 -54.44 4250 -82.65 -61.87 -78.5 -63.52 -58.86 -57.73 4375 -87.36 -61.38 -81.17 -66.81 -61.25 -61.22 4500 -86.83 -61.99 -82.38 -66.1 -60.92 -62.83 4625 -88.34 -66.17 -80.41 -63.99 -62.57 -62.41 4750 -91.85 -74.4 -79.13 -65.62 -68.13 -63.37
209
4875 -89.47 -82.87 -80.16 -71.03 -74.98 -66.74 5000 -85.5 -86.88 -82.41 -74.14 -77.59 -72.18 5125 -82.25 -84.2 -85.47 -74 -77.52 -76.48 5250 -79.51 -80.56 -87.05 -72.4 -77.98 -78.33 5375 -79.71 -83.34 -88.67 -70.56 -76.9 -80.05 5500 -83.76 -90.45 -88.3 -71.65 -74.15 -79.91 5625 -87.58 -91.65 -87.53 -76.61 -73.62 -79.9 5750 -86.51 -91.29 -86.66 -83.95 -75.69 -80.21 5875 -85.42 -85.28 -85.96 -85.28 -78.31 -80.84 6000 -85.2 -81.52 -84.61 -81.09 -78.68 -80.02 6125 -85.78 -83.04 -81.62 -79.11 -79.23 -79.29 6250 -88.81 -85.65 -79.61 -80.01 -81.54 -79.38 6375 -90.58 -84.48 -79.67 -82.55 -84.89 -78.75 6500 -89.73 -81.79 -81.06 -85.43 -87.65 -78.84 6625 -88.37 -78.94 -82.67 -87.59 -89 -79.67 6750 -87.26 -77.57 -83.92 -87.48 -88.88 -79.57 6875 -86.84 -78.55 -84.8 -86.03 -88.78 -78.07 7000 -87.71 -80.31 -84.82 -85.98 -88.8 -77.09
210
Appendix M: Music and Gesture Abstract
Evaluation of ‘open throat’ technique on sound production in classical singing
Helen F. Mitchell, Dianna T. Kenny, Pamela J. Davis and Maree Ryan
‘Open throat’ is thought to involve the maximizing of the pharyngeal space or abduction of the false
vocal folds (Miller 1996; Estill 1996). Mitchell, Kenny, Ryan & Davis (in press) recently reported
consensus among expert pedagogues that the technique of ‘open throat’ was fundamental to good
vocal pedagogy and the production of a beautiful sound, especially in classical singing. The sound
quality produced by this technique was variously described by these pedagogues as free, open,
balanced, coordinated and warm. In addition, open throat was thought to allow the singer to control
dynamic range more effectively than when the technique was not applied. This paper assesses the
acoustic qualities of open throat on the classical singing voice.
Six experienced female singers, deemed by an expert pedagogue to have mastered the technique of
‘open throat’, participated in the study. Singers were asked to perform the same arias under three
different conditions: (1) optimal technique, including maximum ‘open throat;’ (2) good technique
but with less open throat; (3) same as (2) but with the additional instruction to sing as loudly as in
condition 1. Measurements taken from spectrograms indicated that the singers’ vibrato parameters,
notably extent and onset, were affected by the experimental conditions and the decreased use of this
technique. Long-term average spectra plots confirmed timbral differences in centre frequency and
energy level of spectral peaks above 2 kHz.
The absence of open throat produced marked variability in sound quality on the acoustic measures.
This study therefore concluded that ‘open throat’ assisted in the production of a reliable and
consistent sound in classical singing. Further research will confirm its importance for male
classical singers and for singers in other genres.
211
References
Estill, J. (1996). Primer of Basic Figures. Santa Rosa, Estill Voice Training Systems.
Miller, R. (1996). The structure of singing: system and art in vocal technique. New York: London,
Schirmer Books.
212
Appendix N: Project 2 Pedagogue Information Sheet
Expert Perceptual Analysis of Singing Information Sheet for Pedagogue Subjects Page 1 of 1
The University of Sydney
SYDNEY CONSERVATORIUM OF MUSIC
AUSTRALIAN CENTRE FOR APPLIED RESEARCH IN MUSIC
PERFORMANCE
RESEARCH STUDY INTO EXPERT PERCEPTUAL ANALYSIS OF SINGING
PEDAGOGUE SUBJECT INFORMATION SHEET
You are invited to take part in a research study into the expert perceptual analysis of singing. The study is being conducted by Helen Mitchell, a postgraduate student at The University of Sydney, Associate Professor Pamela Davis, Associate Professor Dianna Kenny and Ms Maree Ryan (Lecturer in Singing, The Conservatorium of Music) and will form the basis of Helen Mitchell’s PhD at The University of Sydney.
The object is to discover associations between different terminology and sound qualities that you can hear in singing. We would use the information you give to present more useful information on how pedagogues interpret sound qualities into verbal descriptions. If you agree to take part in this study, you will be asked to complete a short questionnaire that tells us a bit about you (your age, sex, education, occupation and singing experience) and listen and give feedback on a series of singing examples, rating them for sound quality and excellence. To do this, you would be asked to attend a single session lasting less than an hour, at a time and location convenient to you to do this. At this perceptual analysis session, you will hear a number of examples of singing, and will be asked to rate them for a range of different types of sound qualities.
All aspects of the study, including results, will be strictly confidential and only investigator Helen Mitchell will have access to information on participants. Results will be coded with a number and not your name and the data will not enable readers of any of the output to link specific data with voice type, age, or other features which may enable identification of individuals. A report of the study may be submitted for publication, but individual participants will not be identifiable in such a report. There are no risks involved in the participation of this project. There will be a debriefing session on completion of all data collection. Participation in this study is entirely voluntary: you are not obliged to participate and - if you do participate - you can withdraw at any time. Whatever your decision, it will not affect your relationship with any of the researchers, or with the Australian Centre for Applied Research in Music Performance. The ethics committee of The University of Sydney has approved this study. If you have a concern or complaint about the conduct of a research study, you can contact the Manager for Ethics and Biosafety Administration, University of Sydney on (02) 9351 4811. When you have read this information, Helen Mitchell will discuss it with you further and answer any questions you may have. If you would like to know more at any stage, please feel free to contact Helen Mitchell on 0421 632 713 or [email protected].
214
Appendix O: Project 2 Pedagogue Consent Sheet Expert Perceptual Analysis of Singing Consent Sheet for Pedagogue Subjects Page 1 of 1
The University of Sydney
SYDNEY CONSERVATORIUM OF MUSIC
AUSTRALIAN CENTRE FOR APPLIED RESEARCH IN MUSIC
PERFORMANCE
CONSENT SHEET
RESEARCH STUDIES
Project Title: Expert Perceptual Analysis of Singing Investigators: Associate Professor Dianna Kenny Ms Helen Mitchell 1. I acknowledge that I have read the above statement which explains the nature, object and the possible risks of the investigation, and the statement has been explained to me to my satisfaction. Before signing this document I have been given the opportunity to ask questions relating to any possible physical harm I may suffer as a result of my participation and I have received satisfactory answers. I have also been informed that I may not receive any benefits from participating in this study. I have been offered a debriefing session after the research has been completed. 2. My decision whether or not to participate will not prejudice my future relations with the Australian Centre for Applied Research in Music Performance or any of the investigators listed above. If I decide to participate, I am free to withdraw my consent and to discontinue participation at any time without prejudice. 3. I agree that research data gathered from the results of the study may be published provided my name is not used. __________ _________________________________ DATE Signature of Subject 4. I have fully explained to the subject _______________________________ the nature and purpose of the programme and the procedures to be employed as described above and such risks as are involved in their performance. __________ _________________________________ DATE Signature of Responsible Investigator
215
Appendix P: Project 2 Pedagogue Questionnaire
Expert Perceptual Analysis of Singing Questionnaire for Pedagogue Subjects Page 1 of 3
Subject ID_____________
The University of Sydney
SYDNEY CONSERVATORIUM OF MUSIC
AUSTRALIAN CENTRE FOR APPLIED RESEARCH IN MUSIC
PERFORMANCE
Expert Perceptual Analysis of Singing
PEDAGOGUE SUBJECT DEMOGRAPHIC QUESTIONNAIRE
We are interested in finding out about you. Please complete the following questionnaire by ticking the box that best describes you or writing in the space provided. When you have finished, please hand the questionnaire back to one of the researchers.
The questions begin on the next page.
216
Expert Perceptual Analysis of Singing Questionnaire for Pedagogue Subjects Page 2 of 3
Subject ID_____________ 10. How old are you?
_____________years 11. What sex are you?
□ Female
□ Male 12. How many years have you been teaching singing?
_____________years 13. What is the highest level of singing education you have completed?
□ Postgraduate degree at University or Conservatorium
□ Bachelor degree at University or Conservatorium
□ Diploma in Music or Singing
□ Higher School Certificate or equivalent
14. What proportion of your singing studio is private and what proportion is at a University
or Conservatorium?
(a)_____________% private studio (b)_____________% university or conservatorium
Total of (a) and (b) = 100%
217
Expert Perceptual Analysis of Singing Questionnaire for Pedagogue Subjects Page 3 of 3
Subject ID_____________ 15. What types of singing do your students perform in public? Please circle the relevant
singing type and write in the space the percentage of your total studio.
Type of Singing % of performance time Opera
Other classical (eg. Song and concert repertoire, church music)
Choral singing
Musical Theatre
Contemporary styles
Other (please specify)
16. Of the major group of your students, what proportion of them perform at the following
levels? What types of singing do your students perform in public? Please circle the relevant singing type and write in the space the percentage of your total performance time spent performing each type of singing.
Level of Singing % of studio Superstar
International
National
Big City
Regional/Touring
Local Community
Full-Time Voice student
Amateur
Total 100%
Thank you for completing this questionnaire
218
Appendix Q: Project 2 Pedagogue Listener Response Sheet
The University of Sydney
SYDNEY CONSERVATORIUM OF MUSIC
Australian Centre for Applied Research in Music Performance
Expert Perceptual Analysis of Singing
Response Sheet
219
Subject ID_____________
Background to the Project
Expert pedagogues have defined “open throat” as a necessary component of classical singing technique. It was achieved by a widening of the pharynx or ‘retraction’ of the false vocal folds and elicited a sound quality described as, balanced and coordinated, free, open, even and consistent and warm (Mitchell, Kenny et al., in press).
In the study in which you have been invited to be a rater, sopranos and mezzo-sopranos, studying singing at a tertiary institution sang messe di voce on three notes in their range. They also sang excerpts from Ridente la calma by Mozart and Du bist die Ruh by Schubert. They were asked to sing in two different ways, which we have called Optimal and Technical.
These were defined as:
Optimal: the singer’s best singing sound, with maximum use of openness possible.
Technical: Use of an acceptable singing technique in all respects except that the singer has been instructed to use the minimum degree of openness possible.
We ask you to rate each track you hear to indicate which condition you think singers are using. Not all singers will perform each condition, and some conditions may be repeated. Excerpts are presented in a randomised order.
On the first CD, there are 24 messe di voce pairs
On the second CD, there are 30 song excerpts.
Instructions on rating are on the next page.
220
Subject ID_____________
Instructions PART 1: MESSA DI VOCE PAIRS
For each track, the same singer will sing twice. Please rate the sound quality for each of the two
samples as either optimal (O) or technical (T). There are four possible combinations for the messa
di voce pairs. If you think that singer sang with open throat on both occasions, you would complete
the boxes as in Example 1. If the singer sang technical on the first occasion and used maximum
open throat on the second, complete the boxes as given in Example 3 and so on.
Example 1 Example 3
Sample 1 Sample 2 Sample 1 Sample 2
O O T O
Example 2 Example 4
Sample 1 Sample 2 Sample 1 Sample 2
O T T T
We will first do a practice session using four singer pairs to familiarise you with the perceptual
discriminations required in this study. The investigator will play the four samples
PRACTICE MESSA DI VOCE PAIRS:
SINGER 1 SINGER 3
Sample 1 Sample 2 Sample 1 Sample 2
SINGER 2 SINGER 4
Sample 1 Sample 2 Sample 1 Sample 2
221
Messe di voce Pairs
Subject ID_____________ SINGER 1 SINGER 9 SINGER 17 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 2 SINGER 10 SINGER 18 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 3 SINGER 11 SINGER 19 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 4 SINGER 12 SINGER 20 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 5 SINGER 13 SINGER 21 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 6 SINGER 14 SINGER 22 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 7 SINGER 15 SINGER 23 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
SINGER 8 SINGER 16 SINGER 24 Sample 1 Sample 2 Sample 1 Sample 2 Sample 1 Sample 2
222
Singer Example Subject ID_____________
PART 2: SONG SAMPLES You will be required to make the same discriminations in this section as for the Messa di Voce task
except that you will be rating single excerpts from classical songs. You will hear 30 song tracks by
different singers, with a portion of a song, lasting about 20 seconds.
1. Please indicate which condition you think the singer in the excerpt is singing: Optimal or
Technical.
2. Make brief comments or recommendations in point form about the sound quality
3. Rate the singing in this sample out of 10.
Example Excerpt In the song track I heard, the singer: demonstrated ‘open throat’ demonstrated minimal ‘open throat’ OPTIMAL TECHNICAL Comments or Recommendations about the sound quality
•
•
•
Overall Rating: Circle a mark out of 10. 1 2 3 4 5 6 7 8 9 10
223
Singer X
Subject ID_____________ In the song track I heard, the singer:
demonstrated ‘open throat’ demonstrated minimal ‘open throat’ OPTIMAL TECHNICAL Comments or Recommendations
•
•
•
Overall Rating: Circle a mark out of 10. 1 2 3 4 5 6 7 8 9 10
224
Appendix R: Pedagogue Listener Debrief Document
Expert Perceptual Analysis of Singing Page 1 of 1
The University of Sydney
SYDNEY CONSERVATORIUM OF MUSIC
AUSTRALIAN CENTRE FOR APPLIED RESEARCH IN MUSIC
PERFORMANCE
EXPERT PEDAGOGUE DEBRIEF DOCUMENT RESEARCH STUDY INTO THE USE OF SUPRALARYNGEAL AND PHARYNGEAL WIDENING
(OPEN THROAT) IN SINGING Thank you for participating in this study at the Australian Centre for Applied Research in Music Performance – your input as a pedagogue is very important to the expansion and continuation of knowledge in this field.
In this study, the singers you have listened to were asked to sing in two different ways: the first with the maximum degree of openness they would use in their technique, and the second with a smaller degree of openness, but still with an acceptable singing technique. Both these conditions will be around the same volume, so that the test is of the difference in sound quality.
The aim of the study was to look at the use of open throat technique in singing, and this project was to discover if these instructions achieve a perceivable and intrinsic difference to the sound quality produced.
The words in the evaluation you have just completed were the result of an earlier study, where our sample of pedagogues gave a number of words that they associated with the use of open throat in singing. Words like freedom, consistency, warmth and openness were directly linked to the technique.
The more general terms were to ascertain if the overall sound was heard was better with or without use of the technique.
This study is not a test for pedagogues, and there are no correct or incorrect responses. We value your opinions and insights you have given into the examples played.
Thank you once again for taking part in this study.
On behalf of the research team,
Associate Professor Dianna Kenny
Ms Helen Mitchell
225
Appendix S: Perceptual judgments: messa di voce
Messa di voce perceptual responses Listeners 1-2
[Response code: 1=Optimal, 2=Sub-optimal]
Messa di voce LISTENER 1 LISTENER 2 sample sample
SAMPLE Subject singing 1st 2nd Answer1 Answer2 Answer1 Answer2
1 6 2 2 2 2 2 22 1 1 1 1 1 1 13 1 2 1 2 1 2 14 1 2 2 2 2 2 25 3 1 1 2 2 1 16 5 1 2 1 1 1 27 2 1 2 2 2 1 28 6 2 1 2 2 2 19 4 2 2 2 2 2 2
10 3 1 2 1 1 1 211 6 1 1 2 2 1 112 4 1 2 2 2 1 213 5 2 1 1 1 1 114 5 2 2 1 1 1 115 2 2 1 2 2 2 116 2 1 1 1 1 1 117 3 2 2 2 1 2 218 3 2 1 1 2 2 119 4 1 1 1 1 1 120 6 1 2 1 2 1 221 1 1 2 1 2 1 222 5 1 1 1 1 1 123 2 2 2 2 2 2 224 4 2 1 2 2 2 1
227
Messa di voce perceptual responses Listeners 3-6
LISTENER 3 LISTENER 4 LISTENER 5 LISTENER 6 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2
2 2 2 2 2 2 1 11 1 1 1 1 1 2 12 1 2 1 2 1 2 12 2 2 2 2 2 2 21 1 1 1 1 1 1 12 2 1 2 1 2 1 21 2 1 2 1 2 1 22 1 2 2 2 1 2 12 2 2 2 2 2 2 21 2 1 2 1 2 1 21 1 1 1 1 1 2 11 2 1 2 1 2 1 22 1 2 1 1 1 1 12 2 2 2 1 1 2 12 1 2 1 2 1 2 11 1 1 1 2 2 1 22 2 2 2 2 1 2 12 1 2 1 2 1 2 11 1 1 1 1 1 1 11 2 2 2 2 1 1 21 2 1 2 1 2 1 21 1 1 1 1 2 1 12 2 2 2 2 2 2 22 1 2 1 2 1 2 1
228
Messa di voce perceptual responses Listeners 7-10
LISTENER 7 LISTENER 8 LISTENER 9 LISTENER 10 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2
2 2 1 1 2 2 2 21 1 1 1 1 2 1 12 1 2 1 2 1 2 12 2 2 1 2 2 2 22 2 1 1 2 2 2 21 2 1 2 1 2 1 21 2 1 2 1 2 1 22 2 2 1 2 2 2 22 2 1 1 2 2 2 21 2 2 1 1 2 1 21 1 1 1 2 2 1 21 1 1 2 1 2 1 22 1 2 1 2 1 2 12 2 2 2 2 2 2 22 1 2 1 2 1 2 11 1 1 1 1 1 2 12 2 2 1 2 2 2 22 1 1 2 2 2 2 11 1 2 1 1 1 2 12 2 1 2 2 2 1 21 2 1 2 1 2 1 21 1 2 1 2 2 2 12 2 2 1 2 2 2 22 1 1 2 2 1 2 1
229
Messa di voce perceptual responses Listeners 11-14
LISTENER 11 LISTENER 12 LISTENER 13 LISTENER 14 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2 Answer1 Answer2
2 2 2 2 2 2 2 21 1 1 1 1 1 1 12 1 2 1 2 1 2 12 2 2 2 2 2 2 21 1 1 1 1 1 2 21 2 2 2 1 2 1 21 2 1 2 1 2 1 22 1 2 1 2 1 2 12 2 2 2 2 2 2 21 2 1 2 1 2 2 11 1 1 1 1 1 2 21 2 1 2 1 2 1 22 1 2 2 2 2 2 12 2 2 2 2 1 2 22 1 2 1 2 1 2 12 1 1 1 1 1 2 22 2 2 2 2 2 2 22 1 2 1 2 1 2 21 1 1 1 1 1 1 21 2 1 2 1 2 2 21 2 1 2 1 2 1 21 1 2 2 1 1 2 22 2 2 2 2 2 2 22 1 2 1 2 1 2 1
230
Messa di voce perceptual responses Listener 15
LISTENER 15 Answer1 Answer2
2 21 22 12 21 11 21 22 12 21 22 21 22 12 22 11 12 22 11 11 21 21 22 22 1
231
Appendix T: Perceptual judgments: song samples
Song sample perceptual responses Listeners 1-2
[Response code: O=Optimal, SO=Sub-optimal]
Song samples Listener 1 Listener 2 SAMPLE SINGER TASK CONDITION Response Rating Response Rating
1 6 Mozart SO SO 2 SO 1 2 1 Schubert O O 6 O 9 3 1 Mozart O O 9 O 8 4 5 Mozart O SO 6 O 9 5 6 Schubert SO O 5 SO 2 6 1 Mozart SO SO 4 SO 3 7 5 Schubert SO O 5 O 7 8 2 Mozart SO SO 4 SO 2 9 3 Schubert O SO 4 O 9
10 4 Mozart SO SO 3 SO 3 11 3 Mozart O O 5 O 9 12 4 Schubert SO O 5 SO 1 13 5 Schubert O O 9 O 10 14 5 Mozart SO SO 5 SO 1 15 2 Mozart O O 7 O 7 16 3 Schubert SO SO 5 SO 1 17 6 Schubert O O 7 O 5 18 2 Schubert O O 7 O 9 19 3 Mozart SO SO 4 SO 2 20 1 Schubert SO SO 3 SO 2 21 4 Schubert O SO 5 O 8 22 6 Mozart O SO 5 SO 3 23 2 Schubert SO O 6 SO 3 24 4 Mozart O O 7 O 8 25 4 Schubert O O 7 O 8 26 4 Schubert SO SO 4 SO 2 27 5 Mozart O O 9 O 9 28 6 Schubert SO O 5 SO 1 29 2 Mozart O O 5 O 7 30 2 Mozart SO SO 4 SO 1
233
Song sample perceptual responses Listeners 3-6
Listener 3 Listener 4 Listener 5 Listener 6 Response Rating Response Rating Response Rating Response Rating
SO 2 SO 3 SO 4 SO 4 O 5 O 8 O 7 O 7 O 7 O 8 O 6 O 7 O 7 O 6 SO 5 O 8
SO 2 SO 4 SO 5 SO 3 O 6 SO 5 SO 5 O 7 O 5 SO 6 SO 5 O 6
SO 3 SO 3 SO 5 SO 5 O 6 O 7 SO 6 O 7 O 3 SO 4 O 6 SO 5 O 8 O 9 O 7 O 8
SO 2 SO 2 O 5 SO 4 O 8 O 9 O 6 O 8
SO 4 SO 5 O 5 SO 6 O 6 O 5 O 7 O 7
SO 3 SO 3 SO 6 SO 4 O 3 SO 4 SO 6 O 7 O 7 O 7 O 8 O 7
SO 3 SO 3 SO 5 SO 5 SO 2 SO 2 SO 6 SO 6 O 7 O 9 O 7 O 8
SO 3 SO 4 SO 6 O 7 SO 4 O 6 O 7 SO 4 O 6 O 9 O 6 O 8 O 9 O 9 O 7 O 8
SO 2 SO 2 SO 5 SO 4 O 5 O 9 O 8 O 8
SO 3 SO 4 SO 6 SO 4 O 3 O 7 O 7 O 7
SO 3 SO 3 SO 5 SO 4
234
Song sample perceptual responses Listeners 7-10
Listener 7 Listener 8 Listener 9 Listener 10 Response Rating Response Rating Response Rating Response Rating
SO 2 SO 5 SO 3 SO 3 O 4 O 8 O 6 O 7 O 5 O 7 O 5 O 7 O 5 O 8 O 5 O 7
SO 3 SO 5 SO 5 SO 6 O 5 O 6 O 7 SO 4
SO 2 O 8 SO 5 O 8 SO 3 SO 6 SO 4 SO 5 SO 3 O 6 O 5 O 7 SO 3 SO 6 SO 3 SO 2 SO 2 O 7 O 5 O 8 O 5 O 7 SO 3 SO 2
SO 2 O 8 O 7 O 8 O 5 SO 6 SO 6 SO 2
SO 3 O 7 SO 4 O 6 O 4 O 6 SO 3 SO 2
SO 2 O 7 SO 5 SO 6 O 2 SO 6 O 6 O 8 O 5 SO 6 SO 4 SO 2
SO 1 SO 6 SO 5 SO 4 SO 2 O 7 O 5 O 8 O 5 O 7 O 6 SO 2
SO 3 SO 6 O 7 SO 6 O 3 O 8 O 7 SO 5 O 5 O 7 O 6 O 7
SO 1 SO 6 SO 4 SO 2 O 5 O 8 O 7 O 9 O 3 SO 6 SO 4 SO 5 O 4 SO 6 O 7 SO 6
SO 2 SO 6 SO 5 SO 2
235
Song sample perceptual responses Listeners 11-14
Listener 11 Listener 12 Listener 13 Listener 14 Response Rating Response Rating Response Rating Response Rating
SO 1 SO 2 SO 3 SO 1 O 7 O 9 O 8 SO 5
SO 5 O 10 O 7 O 7 O 7 O 10 SO 8 O 8
SO 3 SO 3 SO 5 SO 3 SO 6 SO 6 SO 6 SO 4 SO 2 O 8 O 7 SO 4 SO 2 SO 2 SO 4 SO 1 O 7 O 7 O 8 SO 3
SO 2 SO 3 SO 4 SO 2 O 7 O 10 O 9 O 4
SO 3 SO 2 SO 5 SO 4 O 8 O 9 O 8 O 8
SO 4 SO 6 SO 5 SO 2 O 7 SO 5 O 7 SO 4
SO 5 SO 3 O 7 SO 5 SO 5 SO 6 O 8 SO 6 SO 6 O 7 O 9 SO 5 SO 2 SO 3 SO 5 SO 3 SO 4 SO 3 SO 6 SO 2 O 6 O 9 O 9 O 5 O 5 SO SO 4 O 4
SO 4 SO 6 O 7 SO 3 SO 6 O 10 O 7 O 7 O 7 O 9 O 9 O 6
SO 4 SO 2 SO 4 SO 2 O 7 O 10 O 8 O 8
SO 6 SO 3 SO 3 SO 4 O 5 SO 5 O 7 SO 4
SO 3 SO 3 SO 3 SO 2
236
Song sample perceptual responses Listeners 15
Listener 15 Response Rating
SO 4 O 8 O 6 O 7
SO 4 SO 4 SO 3 SO 3 O 9
SO 3 O 6
SO 3 O 6
SO 4 O 5
SO 5 SO 4 O 6
SO 3 SO 4 O 7
SO 3 SO 6 O 4 O 5
SO 3 O 8
SO 4 SO 5 SO 3
237
Appendix U: Perceptual LTAS
Perceptual Sample LTAS Subjects 1-2
Hz SO1 MO1 SSO1 MSO1 SO 2 MO2 SSO2 MSO2 0
125 -61.17 -59.84 -62.37 -60.4 -62.27 -60.8 -62.62 -61.48 250 -49.71 -47.57 -54.42 -49.66 -58.75 -49.06 -64.02 -51.69 375 -42 -39.1 -46.2 -42.05 -46.16 -40.3 -51.43 -43.43 500 -35.92 -36.35 -40.17 -40.06 -36.38 -37.84 -44.29 -41.36 625 -32.47 -39.1 -36.84 -42.93 -30.18 -39.45 -39.59 -44.02 750 -34.28 -42.4 -37.99 -42.35 -28.98 -41.79 -39.04 -46.3 875 -40.29 -39.57 -40.4 -38.88 -32.91 -43.34 -43.88 -47.11
1000 -40.89 -39.43 -40.62 -39.56 -43 -43.82 -55.18 -49.15 1125 -38.18 -42.79 -41.73 -44.83 -52.96 -47.72 -62.39 -54.27 1250 -36.32 -47.99 -43.28 -49.88 -53.96 -55.09 -62.91 -61.24 1375 -36.98 -48.13 -45.28 -50.4 -54.99 -52.31 -62.35 -55.17 1500 -42.08 -47.62 -49.12 -51.16 -55.86 -49.88 -61.85 -52.37 1625 -49.74 -49.15 -51.79 -51.54 -56.82 -51.28 -60.84 -55.3 1750 -50.79 -53.81 -50.59 -54.23 -59.85 -54.12 -62.54 -61.01 1875 -54.15 -56.95 -52.63 -57.17 -62.45 -55.87 -68.29 -62.12 2000 -58.42 -55.59 -54.29 -54.17 -61.29 -57.74 -71.99 -60.28 2125 -61.94 -56.35 -56.81 -55.32 -60.67 -57.12 -73.76 -60.53 2250 -62.12 -61.09 -61.2 -62.48 -62.35 -58.08 -76.5 -64.35 2375 -60.13 -64.62 -62.11 -65.53 -63.26 -61.2 -77.94 -65.78 2500 -57.56 -60.08 -61.11 -62.43 -63.12 -61.34 -77.83 -63.56 2625 -53.93 -56.16 -57.33 -62.31 -64.23 -60.63 -78.03 -65.9 2750 -52.01 -55.05 -55.92 -61.24 -63.85 -63.14 -76.95 -71.15 2875 -50.98 -55.49 -55.67 -56.85 -62.17 -62.04 -76.96 -67.16 3000 -51.13 -55.24 -57.29 -54.98 -58.31 -58.1 -75.14 -62.54 3125 -52.35 -55.3 -59.81 -56.78 -54.83 -55.97 -72.24 -62.72 3250 -51.51 -57.29 -59.74 -61.39 -54.63 -56.67 -73.82 -65.05 3375 -51.86 -59.86 -60.95 -63.49 -57.21 -57.64 -76.2 -65.72 3500 -55.39 -60.14 -62.26 -62.18 -59.59 -57.75 -77.1 -64.06 3625 -59.88 -60.25 -63.65 -62.92 -60.03 -58.7 -78.47 -63.89 3750 -65.1 -63.35 -68.62 -66.58 -60.18 -62.22 -77.54 -67.38 3875 -70.55 -69.59 -76.42 -70.36 -62.06 -66.47 -79.03 -66.46 4000 -74.65 -76 -78.82 -73.92 -64.72 -67.14 -79.77 -63.96 4125 -78.67 -78.5 -78.8 -79.18 -68.02 -68.64 -80.94 -65.33 4250 -83.62 -81.52 -80.96 -85.97 -72.51 -72.87 -84.79 -70.1 4375 -88.71 -88.37 -87.1 -92.64 -75.41 -75.82 -85.95 -74.65
238
4500 -89.92 -94.55 -88.6 -96.65 -78.69 -79.5 -87.13 -77.27 4625 -88.08 -97.18 -87.27 -98.21 -81.63 -84.34 -86.28 -78.46 4750 -87.64 -98.95 -87.13 -97.6 -82.61 -89.32 -85.77 -84.49 4875 -90.19 -99.39 -85.16 -95.68 -82.64 -94.1 -89.66 -94.3 5000 -93.54 -100.04 -86.34 -95.3 -83.82 -95.37 -96.55 -98.04 5125 -94.27 -99.37 -86.82 -95.89 -85.7 -95.18 -100.35 -100.34 5250 -92.06 -96.62 -85.34 -97.22 -87.23 -96.16 -103.28 -101.52 5375 -90.41 -94.78 -85.04 -98.14 -88.72 -95.87 -104.36 -101.53 5500 -91.57 -94.99 -83.61 -97.4 -89.58 -94.69 -104.48 -99.2 5625 -93.46 -96.59 -83.38 -97.52 -88.45 -94.37 -104.54 -97.76 5750 -93.48 -97.2 -85.25 -97.8 -88.41 -93.68 -104.07 -97.93 5875 -92.53 -96.76 -86.57 -96.71 -88.81 -93.62 -103.61 -97.35 6000 -90.82 -96.91 -89.64 -96.86 -88.77 -92.48 -103.18 -96.52 6125 -88.98 -96.59 -94.15 -98.06 -87.77 -90.19 -101.94 -94.4 6250 -86.61 -94.47 -95.21 -97.7 -85.64 -88.54 -101.09 -92.24 6375 -85.67 -92.14 -94.82 -96.45 -85.1 -86.98 -98.8 -91.93 6500 -85.46 -90.09 -94.08 -96.46 -84.65 -86.56 -96.88 -91.59 6625 -85.18 -88.81 -94.2 -96.32 -84.56 -86.44 -97.19 -89.87 6750 -85.83 -89.08 -95.1 -95.41 -85.66 -85.44 -96.96 -90.36 6875 -87.06 -90.5 -95.69 -94.93 -86.61 -84.53 -96.54 -90.83 7000 -87.52 -91.56 -96.68 -94.78 -88.19 -84.66 -88.17 -90.64
239
Perceptual Sample LTAS Subjects 3-4
Hz SO3 MO3 SSO 3 MSO3 SO4 MO4 SSO4 MSO4 0
125 -62.1 -59.16 -62.43 -60.93 -62.02 -59.19 -62.72 -61.24 250 -55.37 -45.17 -57.12 -48.73 -54.1 -47.36 -58.56 -50.5 375 -43.55 -36.96 -45.69 -40.51 -42.94 -38.83 -46.99 -42.81 500 -35.68 -34.48 -38.51 -38.36 -35.6 -35.95 -40 -41.02 625 -28.24 -36.54 -32.04 -40.38 -30.12 -36.98 -31.88 -44.05 750 -24.9 -39.49 -29.15 -42.62 -27.35 -38.38 -27.43 -44.7 875 -27.17 -38.25 -31.72 -40.09 -28.67 -39.28 -28.41 -40.67
1000 -35.91 -36.14 -40.79 -38.04 -35.54 -40.5 -35.93 -39.9 1125 -43.43 -38.44 -45.96 -40.91 -41.91 -44.63 -44.05 -43.77 1250 -43.41 -45.55 -45.64 -48.15 -42.56 -51.4 -44.88 -52.44 1375 -42.52 -49.49 -45.45 -47.06 -46.29 -54.68 -47.07 -53.02 1500 -40.35 -51.62 -44.27 -49.61 -50.7 -53.73 -49.8 -52.47 1625 -41.09 -51.39 -44.35 -53.23 -52.4 -53.58 -50.04 -53.18 1750 -46.68 -51.58 -50.24 -51.83 -53.93 -56.78 -52.98 -54.41 1875 -55.6 -53.99 -59.13 -51.85 -54.28 -59.02 -59.56 -56.72 2000 -59.02 -54.12 -58.49 -52.07 -53.03 -61.66 -56.6 -56.22 2125 -57.95 -53.17 -59.62 -53.16 -54.43 -63.58 -53.85 -56.81 2250 -55.97 -55.8 -59.95 -57.36 -58.2 -66.8 -55.27 -62.15 2375 -54.2 -59.46 -61.11 -60.49 -59.37 -68.54 -54.52 -66.25 2500 -53.35 -56.73 -63.99 -56.95 -58.94 -66.03 -56.12 -62.82 2625 -55.1 -52.56 -60.87 -53.76 -58.07 -62.81 -60.28 -62.69 2750 -53.15 -49.29 -58.54 -55.06 -56.33 -59.39 -56.46 -67.78 2875 -49.35 -47.48 -56 -57.58 -54.5 -56.74 -55.71 -66.86 3000 -47.24 -47.61 -52.91 -52.23 -53.15 -54.8 -57.49 -60.52 3125 -47.69 -47.52 -53.2 -49.1 -51.87 -53.6 -53.51 -59.55 3250 -51 -48.16 -54.91 -50.94 -51.62 -54.54 -52.76 -63.16 3375 -54.22 -50.1 -55.66 -55.2 -52.15 -56.82 -55.22 -65.26 3500 -55.69 -51.76 -58.12 -55.36 -53.02 -56.06 -57.52 -62.95 3625 -57.7 -53.74 -59.45 -54.53 -55.61 -56.27 -59.66 -63.08 3750 -62.57 -58.48 -62.02 -58.01 -58.77 -59.23 -62.32 -66.36 3875 -69.75 -66.79 -65.95 -65.84 -61.9 -63.38 -63.74 -65.44 4000 -76.07 -76.15 -68.42 -71.9 -64.8 -67.7 -65.82 -62.32 4125 -77.93 -81.19 -73.4 -76.15 -66.67 -70.76 -68.27 -62.7 4250 -78.98 -82.14 -79.31 -77.67 -69.84 -73.21 -70.17 -67.27 4375 -80.31 -83.11 -81.8 -78.47 -75.15 -77.16 -75.06 -72.56 4500 -80.64 -82.68 -82.4 -78.71 -79.8 -84.34 -81.04 -74.49 4625 -77.65 -81.66 -81.92 -77.86 -80.76 -88.36 -79.85 -75.32 4750 -76.6 -81.57 -82.71 -77.94 -79.2 -88.06 -76.26 -79.46
240
4875 -78.61 -80.12 -84.17 -80.51 -76.73 -86.75 -75.66 -83.28 5000 -80.83 -79.37 -85.89 -81.82 -76.26 -85.82 -77.19 -83.31 5125 -82.8 -81.78 -87.03 -81.6 -79.05 -86.89 -80.42 -83.85 5250 -83.85 -85.42 -86.95 -83.63 -83.08 -89.34 -86.39 -87.64 5375 -82.1 -86.62 -87.6 -86.59 -85.35 -90.17 -85.9 -93.74 5500 -81.36 -85.15 -89.22 -86.4 -83.82 -89.04 -81.05 -95.72 5625 -79.59 -82.67 -89.61 -85.42 -82.32 -87.51 -80.36 -95.35 5750 -76.37 -81.26 -89.5 -84.38 -79.61 -86.41 -82.56 -95.41 5875 -74.17 -80.64 -89.43 -84.72 -76.27 -87.07 -86.13 -94.18 6000 -75.28 -80.23 -88.13 -83.28 -75.05 -88.66 -88.01 -91.97 6125 -80.01 -79.97 -86.11 -79.17 -76.64 -89.93 -85.45 -90.58 6250 -82.46 -80.78 -84.63 -77.18 -81.59 -90.08 -82.76 -90.93 6375 -82.47 -82.01 -83.66 -77.68 -86.29 -87.93 -83.51 -91.65 6500 -82.92 -82.52 -84.23 -77.67 -85.39 -87.05 -86.98 -90.58 6625 -82.84 -82.81 -85.7 -76.34 -85.04 -88.96 -88.08 -89.76 6750 -83.19 -82.99 -86.81 -76.98 -86.34 -90.98 -88.47 -90.64 6875 -84.22 -83.56 -88.6 -79.65 -84.91 -90.13 -90.77 -91.1 7000 -85.11 -85.35 -91.87 -82.4 -84.56 -88.5 -90.27 -90.56
241
Perceptual Sample LTAS Subjects 5-6
Hz SO5 MO5 SSO5 MSO5 SO6 MO6 SSO6 MSO6 0
125 -58.47 -59.07 -59.58 -60.91 -61.65 -60.98 -62.53 -62.22 250 -48.01 -45.47 -48.44 -49.57 -51.97 -49.17 -54.25 -52.73 375 -38.79 -37.05 -41.82 -41.82 -41.12 -40.93 -44.33 -43.11 500 -31.68 -34.55 -36.36 -40.07 -31.33 -38.71 -35.1 -39.95 625 -30.85 -36.85 -35.27 -43.13 -27.29 -41.73 -31.01 -41.92 750 -37.1 -41.56 -40.62 -45.24 -29.35 -45.77 -32.78 -45.13 875 -49.05 -45.96 -47.57 -43.24 -38.87 -45.02 -41.32 -44.41
1000 -49.81 -46.46 -48.9 -43.8 -50.45 -46.47 -49.52 -44.65 1125 -51.72 -49.14 -51.34 -48.88 -48.15 -51.87 -48.37 -49 1250 -55.56 -54.35 -53.71 -57.01 -48.61 -52.91 -49.51 -55.37 1375 -54.12 -51.4 -52.93 -53.85 -51.67 -49.85 -49.31 -52.75 1500 -46.47 -46.64 -48.34 -50.53 -48.56 -49.82 -49.76 -50.95 1625 -43.12 -46.28 -43.56 -51.02 -42.75 -51.28 -48.8 -51.12 1750 -45.05 -50.93 -41.84 -53 -39.96 -54.29 -48.17 -54.96 1875 -51.87 -54.29 -45.89 -56.16 -40.87 -59.72 -51.03 -57.82 2000 -58.58 -54.38 -55.89 -56.47 -45.87 -59.23 -57.06 -54.25 2125 -60.71 -56.58 -63.04 -57.56 -53.67 -58.41 -62.27 -51.17 2250 -58.81 -61.16 -64.68 -61.43 -60.08 -62.62 -63.23 -53.65 2375 -56.95 -63.1 -61.86 -63.82 -60.51 -67.94 -63.14 -59.79 2500 -56.51 -58.37 -62.12 -64.04 -58.28 -65.01 -65.13 -59.7 2625 -55.02 -55.41 -64.34 -64.48 -55.64 -65.48 -63.6 -56.84 2750 -52.81 -55.85 -58.53 -67.6 -56.63 -68.82 -63.06 -58.69 2875 -52.36 -55.99 -53.4 -62.36 -58.66 -65.28 -63.67 -63.39 3000 -52.94 -54.53 -51.86 -57.9 -55.66 -60.52 -62.77 -57.35 3125 -53.99 -54.5 -53.48 -58.5 -50.5 -58.52 -59.2 -51.96 3250 -52.66 -55.78 -58.05 -62.67 -46.49 -58.89 -54.68 -52.04 3375 -51.13 -55.39 -58.22 -62.6 -46.19 -58.96 -52.57 -56.39 3500 -51.8 -54.06 -56.31 -60.19 -48.81 -58.58 -53.78 -57.77 3625 -53.87 -54.64 -56.64 -61.54 -50.99 -59.07 -56.47 -56.22 3750 -57.52 -56.96 -58.41 -64.99 -53.48 -60.73 -59.22 -57.4 3875 -60.41 -61.04 -59.24 -62.76 -54.3 -62.22 -60.39 -61.26 4000 -61.76 -64.62 -58.44 -59.95 -55.25 -63.47 -60.16 -62.02 4125 -61.64 -67.57 -59.06 -60.46 -57.7 -65.58 -61.85 -62.44 4250 -63.22 -70.69 -62.9 -63.44 -61.35 -69.06 -65.01 -63.79 4375 -68.98 -75.34 -68.23 -65.92 -66.14 -73.24 -68.61 -67.87 4500 -79.22 -80.65 -70.55 -67.84 -67.87 -74.28 -73.34 -73.79 4625 -86.68 -82.76 -71.91 -69.62 -68.02 -74.9 -72.19 -74.58 4750 -87.32 -85 -74.56 -73.62 -70.21 -76.91 -71.22 -73.32
242
4875 -89.44 -86.85 -78.82 -78.02 -74.53 -78.75 -74.19 -74.84 5000 -92.05 -87.52 -81.13 -80.6 -78.73 -80.4 -78.43 -77.64 5125 -91.88 -86.33 -80.51 -83.51 -80.45 -83.54 -82.44 -78.49 5250 -88.91 -85.51 -81.12 -87.53 -81.1 -88.09 -85.1 -79.68 5375 -82.68 -85.49 -82.54 -88.9 -77.92 -92 -86.59 -84.31 5500 -78.63 -85.59 -82.53 -85.46 -75.96 -93.22 -88.91 -92.08 5625 -78.31 -85.1 -80.96 -84.64 -76.48 -92.39 -91.53 -92.19 5750 -78.57 -85.53 -81.33 -85.02 -74.99 -90.49 -93.28 -90.57 5875 -79.53 -86.56 -82.3 -84.42 -73.05 -89.43 -93.36 -90.68 6000 -82.65 -88.04 -83.02 -84.03 -73.11 -88.72 -91.01 -91.29 6125 -85.46 -87.56 -84.54 -85.38 -75.03 -87.8 -89.1 -89.69 6250 -85.95 -85.72 -86.42 -88.39 -75.63 -86.8 -88.6 -86.24 6375 -85.78 -85 -90.1 -90.04 -74.56 -86.85 -87.03 -83.55 6500 -85.05 -83.84 -93.17 -90.36 -73.84 -88.15 -84.56 -82.96 6625 -84.55 -82.45 -92.41 -91.32 -74.24 -88.82 -82.73 -84.37 6750 -85.73 -82.35 -91.72 -92.14 -76.95 -87.98 -82.14 -85.2 6875 -87.51 -82.16 -91.44 -91.21 -79.99 -87.23 -82.32 -84.5 7000 -88.1 -82.06 -91.12 -90.86 -79.28 -87.49 -81.89 -85.36
243
Appendix V: ICMPC8 Abstract
APPRAISING VOCAL BEAUTY BY EAR AND EYE: CAN WE BELIEVE OUR SENSES? Dianna T Kenny and Helen F Mitchell
Background
What qualities of the singing voice define vocal beauty? This question has come into
sharp focus since the advent of acoustic analysis. There is now both a vast literature on
vocal acoustic properties and many unanswered questions about what conclusions can
be drawn from such analyses. These questions are both methodological and conceptual.
There are many ways that the voice can be represented acoustically, but the most
common is the measurement of energy production. Long-term average spectra (LTAS)
are used to represent singers’ sound and its different vocal qualities based on energy
changes that occur during different vocal tasks. Studies of voice quality have taken two
main forms. In the first type of study, LTAS is used to “validate” perceptual cues. In the
second type of study, acoustic findings are “validated” by perceptual ratings.
Aims
In this study, we investigated pedagogues’ rating of female classical singers using
maximum open throat technique and reduced open throat technique in their singing of a
classical song and a romantic lied. We matched perceptual ratings to acoustic measures
of each perceptual sample to determine whether acoustic analysis matched perceptual
judgements of overall timbre. The aim of the study was to assess the degree of
agreement between acoustic analysis and perceptual ratings of vocal beauty.
244
Method
Fifteen expert judges rated 24 samples with six repeats of six advanced singing students
under two conditions: ‘optimal’ (O), representing maximal open throat and ‘suboptimal’
(SO), representing reduced open throat. LTAS were performed on each singing sample
and sound pressure ratio (SPR) and energy ratio (ER) were calculated on each LTAS.
Perceptual scores, SPR and ER were rank ordered. We then compared perceptual
rankings with rankings of acoustic measures (SPR and ER) to assess whether the
acoustic characteristics matched the perceptual judgments of overall timbre.
The study design was a repeated measures [24 ratings per listener] randomized complete
block with a 2 [Task (Mozart vs Schubert)] x 2 [condition (Optimal and Sub-optimal]
factorial structure. Main effects for task and condition and interaction effects (contrasts)
were calculated for each dependent measure.
Results
Intraclass correlations with two-way random effects model of absolute agreement
(ICC(2,1)) indicated that judges were reliable in their ratings, matching their exact rating
in the first rendition and the repeat in 5 of 6 cases.
Perceptual ratings were significantly higher for O than for SO in both the Mozart and
Schubert tasks. There was no interaction between musical task and condition.
245
While there was a significant relationship between SPR and ER, there was no
relationship between perceptual ratings of vocal samples or singers based on SPR or
ER.
Conclusions
These findings suggest that since LTAS measures are not consistent with perceptual
ratings of vocal beauty, they cannot be used to define a beautiful voice.
Topic Areas
Aesthetic perception and response
Key words: LTAS, perceptual ratings, vocal beauty
246
Appendix W: Summary of publications and presentations arising from the thesis
Peer-reviewed journal articles Mitchell, H. F., & Kenny, D. T. (in review). Open throat: acoustic and perceptual
support for pedagogical practice. Journal of Singing.
Mitchell, H. F., Kenny, D. T., Ryan, M., & Davis, P. J. (2003). Defining open throat
through content analysis of experts' pedagogical practices. Logopedics
Phoniatrics Vocology, 28(4), 167-180.
Mitchell, H. F., & Kenny, D. T. (2004). The effects of “open throat” technique on long
term average spectra (LTAS) of female classical voice. Logopedics Phoniatrics
Vocology, 29(3), 99-118.
Mitchell, H. F., & Kenny, D. T. (2004). The impact of “open throat” technique on
vibrato rate, extent and onset in classical singing. Logopedics Phoniatrics
Vocology 29(4), 171-182.
Kenny, D. T. & Mitchell, H. F., (in press). Acoustic and perceptual appraisal of vocal
gestures in the female classical voice. Journal of Voice.
Mitchell, H. F., & Kenny, D. T., (in press). Can experts identify “open throat” technique
as a perceptual phenomenon? Musicae Scientiae.
Conference Papers Mitchell, H. F., Kenny, D. T., Ryan, M. Open throat: pedagogical perceptions and
practices. ANATS National Conference, October 2002, Melbourne.
Mitchell, H. F., & Kenny, D. T. Evaluation of open throat technique on classical
singing. Music and Gesture [University of East Anglia in association with the
European Society for the Cognitive Sciences of Music (ESCOM), the Society
247
for Music Analysis (SMA), and the Society for Education, Music, and
Psychology Research (SEMPRE)], August 2003, East Anglia, UK.
Kenny, D.T., & Mitchell, H.F., Visual and auditory perception of vocal beauty: conflict
or concurrence? In: Lipscomb S, Ashley R, Gjerdingen R, Webster P, editors.
8th International Conference on Music Perception & Cognition (ICMPC8);
August 2004; Evanston, IL USA: Causal Productions; 2004. p. 171-174.
248