Pitch Matching Accuracy of Trained Singers,Untrained Subjects with Talented Singing Voices,

and Untrained Subjects with Nontalented Singing Voicesin Conditions of Varying Feedback

*Christopher Watts, *Jessica Murphy, and Kathryn Barnes-BurroughsMobile, Alabama

Summary: At a physiological level, the act of singing involves control andcoordination of several systems involved in the production of sound, includingrespiration, phonation, resonance, and afferent systems used to monitorproduction. The ability to produce a melodious singing voice (eg, in tune withaccurate pitch) is dependent on control over these motor and sensory systems.To test this position, trained singers and untrained subjects with and withoutexpressed singing talent were asked to match pitches of target pure tones. Theability to match pitch reflected the ability to accurately integrate sensoryperception with motor planning and execution. Pitch-matching accuracy wasmeasured at the onset of phonation (prephonatory set) before external feedbackcould be utilized to adjust the voiced source, during phonation when externalauditory feedback could be utilized, and during phonation when externalauditory feedback was masked. Results revealed trained singers and untrainedsubjects with singing talent were no different in their pitch-matching abilitieswhen measured before or after external feedback could be utilized. Theuntrained subjects with singing talent were also significantly more accurate thanthe trained singers when external auditory feedback was masked. Both groupswere significantly more accurate than the untrained subjects without singingtalent.

Key Words: Singing talentPitch controlFeedback.

Accepted for publication October 22, 2002.From the *Department of Speech Pathology and Audiology,

University of South Alabama, Mobile, Alabama; Departmentof Music, University of South Alabama, Mobile, Alabama.

Address correspondence and reprint requests to ChristopherR. Watts, Department of Speech Pathology and Audiology, Uni-versity of South Alabama, Mobile, AL 36688. E-mail: cwatts@usouthal.edu

Journal of Voice, Vol. 17, No. 2, pp. 185194 2003 The Voice Foundation0892-1997/2003 $30.000doi: 10.1016/S0892-1997(03)00023-7

185

INTRODUCTIONThe facility to express a skill in the performance

of a task is often credited to talent. Talent canbe thought of, in a general sense, as a special naturalability that has the potential to lead to a capacityfor achievement or success. Talent can be identifiedfor a number of different abilities, including theability to sing. Singing involves both physiologyand art. From a physiological perspective, singingtalent can be defined as the special natural ability toproduce musical modulations of the voice where the

CHRISTOPHER WATTS ET AL186

sounds vary over a wide range of frequencies andare in tune with each other, or where such soundsare melodious.

In the literature of music research, individualswith singing talent have been labeled as accuratesingers, whereas those without talent have beenlabeled as inaccurate singers, poor pitch singers,uncertain singers, or monotones.15 Goetze et al6identified variables that are requisite for accuratesinging, which included the ability to discriminatebetween pitches, vocalize over a wide range ofpitches, monitor vocal pitch, and desire to sing. Itis possible that these are the variables inherentlynecessary for one to express singing talent. Someindividuals, without any training, will express sing-ing talent, whereas others will not.

In the literature of voice science where physiolog-ical and perceptual variables related to the singingvoice are investigated, individuals with and withoutvoice training are often compared. The groups towhich these two populations, respectively, are as-signed, are usually labeled as trained and untrainedsingers, singers and nonsingers, or professional sing-ers and nonsingers. As noted above, within the popu-lation of untrained singers, some will express singingtalent whereas others will not. Although numerousstudies have found differences between individu-als with and without singing training, few studieshave separated untrained subjects into those whohave and do not have expressed singing talent.710As a result, there is little objective empirical evi-dence that sheds light on the factors responsible forexpressed singing talent. In addition, there is a lackof objective data that compare the abilities of trainedsingers to those of untrained subjects who possessa talented singing voice.

The ability to monitor vocal pitch has been identi-fied as one variable related to singing talent.6 Mon-itoring vocal pitch can be thought of as ongoingassessment of the accuracy of ones fundamentalfrequency (F0) and corresponding adjustments madetoward the goal of producing an intended pitch.11It has been suggested that the control of F0 is ofparamount importance in singing.12 The control of F0is accomplished via a combination of laryngeal ad-justments and changes in subglottal air pressure. Aninnate ability as well as training may be responsible

Journal of Voice, Vol. 17, No. 2, 2003

for singers exhibiting more accurate F0 controlthan nonsingers.13

Feedback plays an essential role in the control ofF0. Many studies have found that alterations to thevocal frequency being externally fed back to the earduring production of sound result in correspondingchanges to the F0.1418 Through such studies, a greatdeal of important information regarding the role ofexternal feedback on F0 control has been obtained.Less information is available, however, regardingthe role of internal feedback for the control of F0.For the purposes of this paper, internal feedbackinvolved in the control of F0 can include the bone-conducted signal to the cochlea (eg, internal palles-thetic feedback) in addition to the articular, myotatic,and mucosal reflexes stimulated by mechanorecep-tors located within the laryngeal joints, muscles, andsubglottic mucosa, respectively.19

In general, studies that have looked at the use ofinternal feedback for control of F0 have found that,when external feedback is not available, the accuracyof F0 is decreased. When both external (via outer ear)and internal (via bone conduction) auditory feedbackare not available, studies have found that F0 be-comes less stable and subjects are less accurate atmatching target pitches.11,20 Although such evidencesuggests that the use of external feedback and inter-nal feedback results in superior control of F0 com-pared to the use of internal feedback only, otherreports suggested a role of some importance servedby internal feedback for controlling the frequencyof the voice. For example, laryngeal perturbationsduring vocalization and laryngeal anesthesia haveboth been found to have an effect on the stabilityof F0.21,22

Anecdotal evidence that supports the use of inter-nal feedback for controlling F0 has also been re-ported. Di Carlo23 described the use of internalvoice sensitivities in opera singers. Internal voicesensitivities were described as the pallesthetic (vi-bratory) and kinesthetic (muscular) information thatsingers use to control their vocal productions. Hesuggested that opera singers used the vibrations inthe bones of the thorax and craniofacial bone struc-tures to assist in vocalizations, and he suggested thatsingers may rely on this type of feedback whenexternal auditory feedback is disrupted due to thereverberation effects that occur in some music halls.

PITCH-MATCHING ACCURACY 187

Experienced, trained singers have also anecdotallyreported that, in cases where external feedback wasnot available, they have relied on feeling notes toaccurately produce tones.

When producing an intended frequency, such asduring a pitch-matching task, there is a point in timewhen vocal production occurs before either externalor internal feedback can be utilized. This occursduring the first few milliseconds at the beginningof a vocalization. Leading up to this point in time,motor planning and programming has occurred thatwill govern the activation and position of laryngealmuscles and structures. The posturing of the laryn-geal structures prior to the initiation of vocaliza-tion has been referred to in the literature asprephonatory set or prephonatory tuning. The accu-racy of prephonatory tuning is measured in pitch-matching paradigms by calculating the period ofthe first measurable (eg, periodic) waveform of thevocalization, converting this value to the corres-ponding frequency, and comparing it against a targetfrequency. It has been found that singers are moreaccurate at prephonatory tuning than are nonsing-ers.13,24 The ability to accurately position the struc-tures related to vocalization for production of anintended frequency may be another variable relatedto singing talent.

A review of the literature has identified a lackof evidence related to objective measures of bothphysiological and perceptual abilities associatedwith vocal control in untrained subjects with andwithout singing talent. The purpose of this study wasto assess the abilities of these two groups, alongwith trained singers, to control F0 during a pitch-matching task, and to investigate whether these abili-ties were affected differentially when feedback wasand was not available. Specific research questionsincluded: (1) Are untrained subjects with expressedsinging talent able to utilize external auditoryfeedback to more accurately match pitch when com-pared to untrained subjects without expressed sing-ing talent and trained singers? (2) Are untrainedsubjects with expressed singing talent able to utilizeinternal feedback to more accurately match pitchwhen compared to untrained subjects without ex-pressed singing talent and trained singers? (3) Aretrained subjects able to pretune their vocal mecha-nism to more accurately match pitch when compared

Journal of Voice, Vol. 17, No. 2, 2003

to untrained subjects without expressed singingtalent and trained singers?

METHODSParticipants

Female volunteers from the student body of theUniversity of South Alabama and the local commu-nity were recruited. Fifteen subjects were assignedto three groups of five. The first group consisted offive female volunteers who were trained singers witha minimum of 3 years vocal training with a profes-sional voice teacher. This group was labeled thetrained singers (TS). The next group consistedof five female volunteers who had no history ofprofessional, individualized vocal training but wereidentified by professional voice teachers as exhib-iting expressed singing talent. This group was la-beled the untrained talented subjects (UTS). Thethird group consisted of five female volunteers whohad no history of vocal training and were identifiedby professional voice teachers as not exhibitingexpressed singing talent. This group was labeledthe nontalented subjects (NTS). The participantsranged in ages from 19 to 30 years, with a meanage of 23 years. In addition, the trained singershad a range of 3 to 6.5 years (mean of 5 years)of professional voice lessons. All participants hadnormal hearing at 15-dB HL for 500 through 8000Hz, no history of chronic vocal pathology, no previ-ous voice therapy, no history of drug or alcoholabuse, no history of smoking, and no current aller-gies or voice problems at the time of testing.

InstrumentationThe Kay Elemetrics Computerized Speech Lab

(CSL) (Kay Elemetrics, Lincoln Park, NJ) was usedas a signal generation and signal acquisition device.For auditory stimuli, the tone generator applicationof the CSL was used to generate periodic waveformsand noise stimuli. These stimuli were played throughthe CSL output via both Optimus Pro 40 circumaural headphones and a Fostex 6301B amplifiedloudspeaker (Fostex America, Anaheim, CA). Forsignal acquisition (singers voices), the CSL inputchannels were used. All speakers were wearing ahead-mounted AKG Acoustics microphone posi-tioned at a constant microphone-to-mouth distance

CHRISTOPHER WATTS ET AL188

of 1 inch, placed off-center at the left corner ofthe mouth. All signals captured via the CSL weredigitized at 44.1 KHz. Signals from the micro-phone were routed to the Kay Elemetrics externalCSL hardware to measure fundamental frequencyrange (via Kay Elemetrics Voice Range Profile[VRP] software) and to capture vocalizations. AnAlesis Masterlink hard-disk/CD recorder was incor-porated for storing singing samples from each sub-ject, and for playing tone stimuli. All participantswere tested in a double-walled sound booth. AnExtech sound level meter (SLM- model 407750)was used for visual feedback to assist subjects incontrolling the intensity of their vocal output.

StimuliTwo types of stimuli, pure tone waveforms with

and without additional aperiodic white noise, wereused during pitch-matching activities. In order toget a sampling of pitch-matching abilities at morethan one frequency, pure tones within 30% to 50%of each subjects frequency range were generated.Stimuli consisting of 4-second pure tones wereused during the external feedback condition andcalled the external tones. Also, stimuli consistingof the same pure tones followed immediately by 8seconds of white noise were used during the internalfeedback condition and called the internal tones.Each participant was presented with 30 externaltones, presented randomly in one testing condition,and 30 internal tones, presented randomly in asecond testing condition, so that a total of 60 toneswere presented after completion of testing.

ProceduresTesting consisted of two separate sessions. During

the first session, each participant read and signed aconsent form, completed a questionnaire, and under-went audiometric screening to ensure hearing waswithin the required limits for the study. They werethen recorded singing the first stanza of Americathe Beautiful. Before recording, all participantswere given a minimum of 5 minutes for vocal warm-up, including time to practice the stimulus song.Every participant reported that they were familiarwith the song. If any participant was unfamiliar withtasks used for vocal warm-up, they were given in-struction. After warm-up, participants were instructed

Journal of Voice, Vol. 17, No. 2, 2003

to sing the song using their best singing voice. Theserenditions took place in a double-walled sound boothwith the subject in isolation, and they were recordeddigitally via a direct line input from the head-mounted microphone. These samples were used toplace each participant into her respective groupbased on the judgments of the professional voiceteachers (see below). After singing, physiologicalfundamental frequency ranges (FFR) were acquiredusing the head-mounted microphone and VRP pro-gram with an automatic procedure. The automaticprocedure used in this study has been described indetail by Pabon and Plomp.25

In order to assign participants into an experimen-tal group, recordings of the stimulus songs wereplayed to two trained professional teachers of sing-ing, both of whom were faculty members of theUniversity of South Alabama School of Music. Bothjudges had over 10 years of professional experienceand a graduate-level degree in voice. Each judge lis-tened to the recordings independently. The identi-ties of the singers were not disclosed to the judges.The judges were provided the singing samples ona compact disk and asked to judge whether theperson singing expressed natural singing talent.Judgments were based on the intonation and qualityof their voice during singing. Only subjects whowere rated by both judges as either having or nothaving natural singing talent were assigned to oneof the two groups. Any subject who was rated differ-ently between the two judges was excluded fromthe study. A third professional voice teacher withsimilar experience independently judged 5 of theuntrained participants (3 that were judged as UTSand two judged as NTS) for reliability purposes.Agreement between this and the prior judges was100%.

During the second testing session, the subjectswere seated in the double-walled sound booth wear-ing a head-mounted microphone. The microphonehad a direct line input into the CSL. Subjectswere tested in two different conditions, called theexternal feedback condition and the internal feed-back condition. The external feedback conditionconsisted of the subjects matching the pitch of theexternal tones presented via loudspeaker at 70-dBSPL. The internal feedback condition consisted ofthe subjects matching the pitch of the pure tone

PITCH-MATCHING ACCURACY 189

segments of the internal tones while maskingnoise played through the headphones. The pure tonesegments of the internal tones were presented 70-dB SPL and followed by the white noise at a levelof 85-dB SPL. The purpose of the white noise wasto mask the external auditory sidetone, so that thesubjects had to rely on other feedback mechanisms,such as proprioceptive feedback, for matching thepitch of the target tones.

To control for possible effects of the maskingnoise on subsequent use of auditory feedback forF0 control, the external feedback condition wasalways presented first. In this condition, subjectswere presented with 30 external tones, which wereordered randomly. Subjects were instructed to matchthe pitch of the tone immediately after the presenta-tion of the target ended. Subjects viewed the SLMduring the matching task to keep the intensity oftheir acoustic response between 65- and 70-dB SPL,in order to control for the effects of vocal intensity onfundamental frequency. Acoustic response signalswere captured digitally using the CSL. The experi-menter started the recording of each subjects re-sponse 1 second prior to vocalization, so that theinitiation of phonation could be clearly delineatedwhen analyzing the acoustic signals.

During the internal feedback condition, subjectswore both the head-mounted microphone and circu-maural headphones. Subjects were presented with30 randomly ordered internal tones through theheadphones. Subjects were instructed to match thepitch of the tone immediately after the pure-tone seg-ment of the target ended, and during the presentationof white noise through the headphones, using theSLM to keep the intensity of their acousticresponse between 65- and 70-dB SPL. Acoustic re-sponses were captured and stored with the samemethod as that in the external condition.

AnalysisThree dependent variables were measured: pitch-

matching accuracy with prephonatory set, pitch-matching accuracy with external feedback, andpitch-matching accuracy with internal feedback.The prephonatory set was measured from the 30responses of each subject during the external feed-back condition. This measure could have been taken

Journal of Voice, Vol. 17, No. 2, 2003

from either of the experimental test conditions. How-ever, to control for possible effects of noise exposureon the prephonatory set, the measure was acquiredfrom the responses in the external feedback con-dition.

Pitch-matching abilities using the prephonatoryset were analyzed for each response by convertingthe period of the first measurable waveform into acorresponding frequency. This value was then sub-tracted from the frequency of the target tone, so thata difference score was obtained for each response.This difference score reflected pitch-matching accu-racy. A total of 30 difference scores were calculatedfor each subject in this measurement condition.Pitch-matching abilities in the external feedback andinternal feedback conditions were analyzed by mea-suring the period of the waveforms located 2 secondsinto the response. Difference scores were calculatedin the same manner as those for the prephonatory set.

Statistical analysis was completed using atwo-way (group condition) analysis of variance(ANOVA) with repeated measures applied to the datafor pitch-matching accuracy with the prephonatoryset, external feedback, and internal feedback. Anysignificant interaction effects were further investi-gated using separate one-way ANOVAs. Criterionfor significance was set at the 0.05 levels. Intrameas-urer reliability was calculated by having the experi-menter remeasure 15% of the waveforms. A secondtrained experimenter randomly remeasured 15% ofthe waveforms to assess intermeasurer reliability.The measurement values, in frequency, were com-pared using a Pearson product-moment correlation.Both intrameasurer (r 0.99) and intermeasurer re-liability (r 0.98) were high, and the comparisonrevealed a significant correlation for both (p 0.01).

RESULTSThe pitch-matching accuracy of the three groups

using the prephonatory set, external feedback, andinternal feedback is displayed in Figure 1. As canbe seen, the UTS participants, as a group, alwaysperformed better than both the TS and NTSgroups. Average difference scores for the UTS groupwere 19 Hz, 8 Hz, and 6 Hz across those threemeasurement conditions, respectively, whereasthose of the TS group were 24 Hz, 15 Hz, and 17

CHRISTOPHER WATTS ET AL190

FIGURE 1. Average pitch-matching accuracy of the trained, untrained talented, and untrainednontalented groups when using prephonatory set, external feedback, and internal feedback.

Hz, respectively, and for the NTS group 62 Hz,53 Hz, and 45 Hz, respectively. It can also be seenthat the TS group was always more accurate than theNTS group.

Across conditions, both the UTS and NTS groupswere least accurate in matching pitch when mea-sured at the prephonatory set, and most accuratewhen using internal feedback. The TS differed fromthis trend only in that they were more accurate whenusing external feedback compared to when they wereusing internal feedback.

Figures 2 to 4 show box plots of pitch-matchingaccuracy for each group in the three measurementconditions. These plots are of interest as they showthat the UTS group was not only more accurate atmatching pitch than the TS in all conditions, but alsowas less variable within the group, especially inthe two conditions where feedback could have beenutilized. The NTS group was always more variablewith greater spread around the mean compared withthe other two groups.

Journal of Voice, Vol. 17, No. 2, 2003

A two-way (group condition) ANOVA with re-peated measures was applied to the pitch-matchingdata. Results revealed a significant main effect forcondition (F 70.8, p 0.01) and group (F 98.7,p 0.01), with a significant interaction effect(F 5.1, p 0.01). Because of the interaction,group means were compared in each measurementcondition using three separate one-way ANOVAs.

Results of the one-way ANOVAs revealed a sig-nificant effect for group when using the prephona-tory set (F 76.6, p 0.01), external feedback(F 99.6, p 0.01), and internal feedback (F 69.7,p 0.01). Posthoc testing utilizing a Scheffe testrevealed that when using the prephonatory set andexternal feedback, the TS and UTS groups weresignificantly more accurate at matching pitch thanwas the NTS group (p 0.01 and p 0.01, respec-tively), but not different from each other. When usinginternal feedback, the UTS group was significantlymore accurate than were the TS group (p 0.01)and the NTS group (p 0.01). Additionally, the TS

PITCH-MATCHING ACCURACY 191

FIGURE 2. Box plot of pitch-matching accuracy using prephonatory set for thetrained, untrained talented, and untrained nontalented groups.

group was significantly more accurate than was theNTS group (p 0.01).

DISCUSSIONThis study investigated the pitch-matching abili-

ties of trained singers (TS), untrained subjects withexpressed singing talent (UTS), and untrained sub-jects without expressed singing talent (NTS). Theseabilities were tested when using prephonatory set,external feedback, and internal feedback. A smallsample size was utilized for group comparisons, andas such, any generalizations based on the resultsshould be made with caution. Analysis of the datarevealed a number of interesting findings. One wasthat the TS and UTS groups demonstrated signifi-cantly greater pitch-matching accuracy in all mea-sured conditions compared to the NTS group.Additionally, the UTS group was significantly moreaccurate than was the TS group when they had torely on internal feedback.

The UTS group could match pitch equally as wellas the TS group when measured at the initial onset

Journal of Voice, Vol. 17, No. 2, 2003

of phonation and when utilizing external feedback.The ability to match pitch accurately when using ex-ternal feedback supports the idea, as Goetze et al.6suggested, that accurate monitoring of pitch is afactor related to singing talent. This might explainwhy the TS and UTS groups did not differ in apitch-matching task where external feedback wasavailable. However, Goetze et als suggestion mightbe expanded, based on the results from measuringthe prephonatory set, to include the ability to bothaccurately perceive the pitch of a tone and coordinatethat perception with the motor planning, program-ming, and execution needed to reproduce it accu-rately at the onset of phonation. This ability wouldresult in less reliance on the external signal forfeedback purposes, which would be of value in sing-ing, as room acoustics can greatly affect the auditorysignal that is fed back into the ears. It is suggestedby the authors that natural singing talent is related, atleast in part, to this ability to accurately monitorpitch and coordinate perception with motor systemsfor reproduction of a pitch at the very beginningof phonation.

CHRISTOPHER WATTS ET AL192

FIGURE 3. Box plot of pitch-matching accuracy using external feedback for thetrained, untrained talented, and untrained nontalented groups.

Di Carlo23 suggested that opera singers are ableto rely heavily on their internal voice sensitivities,which are the vibratory and muscular feedback avail-able to the person during singing. She concludedthat although auditory feedback plays a role in pitchaccuracy, singers judge the quality of their vocaliza-tions by utilizing ongoing assessment of propriocep-tive feedback. If this is true, this ability may comefrom many years of training or practice. The trainedsubjects in this study had a mean of 5 years ofprofessional voice training. However, they were notas accurate at matching pitch compared to the UTSgroup when external feedback was not available.Interestingly, the raw data indicated that the TS sub-jects, within the group, were less accurate in theinternal feedback condition compared to the othertwo conditions, whereas both the UTS and NTSgroups were more accurate in this condition.

One possible explanation for this finding is thatduring early voice training, emphasis may be placedon training the ear to monitor the sound of the voice.The method of early voice training varies from

Journal of Voice, Vol. 17, No. 2, 2003

teacher to teacher, and some may not focus onear training at all. However, anecdotal reports seemto indicate that training the ear to the auditory side-tone is an important element of early voice trainingutilized by some teachers of singing. This emphasiscould then possibly result in learning to utilize exter-nal feedback of the sidetone while deemphasizinginternal information as a consequence. With moreadvanced voice training and experience, the utili-zation of internal feedback may increase. Moredetailed experimental designs are required to accu-rately test this theory.

The finding that there was no difference in pitch-matching abilities between untrained singers withexpressed singing talent and trained singers whenusing the prephonatory set or external feedback isimportant as it differs somewhat from findings ofprevious investigations that have compared trainedand untrained individuals. Previous studies havefound differences for pitch-matching accuracybetween trained and untrained subjects.13,26 How-ever, these and most other studies that have com-pared abilities related to singing in trained and

PITCH-MATCHING ACCURACY 193

FIGURE 4. Box plot of pitch-matching accuracy using internal feedback for thetrained, untrained talented, and untrained nontalented groups.

untrained subjects have not factored out the abilitiesof subjects in the untrained groups who have singingtalent. As a result, findings from these previousinvestigations may not represent the abilities of alluntrained singers, specifically those who expresssinging talent.

The present investigation found that, when match-ing pitch either at the onset of phonation or whenusing external feedback, untrained subjects with ex-pressed singing talent are equally as accurate astrained singers. This gives credence to the idea thatthe ability to match pitch accurately is a prerequisitefor singing talent. There may also be additional abili-ties related to voice production that are similar intrained singers and untrained subjects with ex-pressed singing talent. This idea was in some waysupported by Holliens27 view that a golden voicewas a product of both innate talent and training. Itis most often the case that a few of those untrainedsubjects with expressed singing talent are the sameindividuals who go on to seek further training, viaprofessional instruction, to expand and refine theirsinging skills for professional development.

Journal of Voice, Vol. 17, No. 2, 2003

The findings from this investigation point to aneed for more research that investigates the sub-strates of expressed singing talent. Such investiga-tions could provide objective evidence to answer thequestion, why can some people, without training,produce a good singing voice while some cannot?To date, anecdotal evidence exists to answer thisquestion. Larger empirical studies are needed, how-ever, to substantiate opinions and ideas related tothose answers.

REFERENCES1. Yarbrough C, Green G, Benson W, Bowers J. Inaccurate

singers: An exploratory study of variables affecting pitch-matching. Bulletin of the Council for Research in MusicEducation. 1991;105:2333.

2. Davies A, Roberts E. Poor pitch singing: A survey of itsincidence in school children. Psychol Music. 1981;3:2436.

3. Flowers P, Dunne-Sousa D. Pitch-pattern accuracy, tonality,and vocal range in preschool childrens singing. J Res MusicEduc. 1990;38:102114.

4. Price H. Interval matching by undergraduate nonmusicmajors. J Res Music Educ. 2000;48:360372.

CHRISTOPHER WATTS ET AL194

5. Joyner D. The monotone problem. J Res Music Educ. 1969;17:115124.

6. Goetze M, Cooper N, Brown C. Recent research on singingin the general music classroom. Bull Council Res MusicEduc. 1990;104:1637.

7. Brown W, Rothman H, Sapienza C. Perceptual and acous-tic study of professional trained versus untrained voices.J Voice. 2000;3:301309.

8. Sulter A, Schutte H, Miller D. Differences in phonetogramfeatures between male and female subjects with and withoutvocal training. J Voice. 1995;9:363377.

9. Akerlund L, Gamming P. Average loudness level, meanfundamental frequency, and subglottal pressure: compari-son between female singers and nonsingers. J Voice. 1994;8:263270.

10. Sundberg J. Maximum speed of pitch changes in singersand untrained subjects. J Phonet. 1979;7:7179.

11. Elliott L, Niemoeller A. The role of hearing in controllingvoice fundamental frequency. Int Audiol. 1970;9:4752.

12. Titze I. Principles of Voice Production. Upper Saddle River,NJ: Prentice Hall; 1994:191217.

13. Murry T. Pitch-matching accuracy in singers and nonsing-ers. J Voice. 1990;4:317321.

14. Larson C, Burnett T, Kiran S, Hain T. Effects of pitch-shiftonset velocity on voice FO responses. J Acoust Soc Am.2000;107:559564.

15. Burnett T, Senner J, Larson C. Voice FO responses to pitch-shifted auditory feedback: a preliminary study. J Voice.1997;11:202211.

16. Burnett T, Freedland M, Larson C, Hain T. Voice FO re-sponses to manipulations in pitch feedback. J Acoust SocAm. 1998;103:31533161.

Journal of Voice, Vol. 17, No. 2, 2003

17. Laukkanen A. Artificial pitch changing in auditory feedbackas a possible method in voice training and therapy. FoliaPhoniatrica et Logopaedica. 1994;46:8696.

18. Elman J. Effects of frequency-shifted feedback on the pitchof vocal productions. J Acoust Soc Am. 1981;70:4551.

19. Wyke B. Laryngeal neuromuscular control in singing. FoliaPhoniatrica et Logopaedica. 1974;26:295306.

20. Ward W, Burns E. Singing without auditory feedback. JRes Singing. 1978;1:2444.

21. Sapir S, Baker K, Larson C, Ramig L. Short-latencychanges in Fo and neck surface EMG induced by mechani-cal perturbations of the larynx during sustained vowel pho-nation. J Speech Lang Hear Res. 2000;43:268276.

22. Sorensen D, Horii Y, Leonard R. Effects of laryngeal topicalanesthesia on voice fundamental frequency perturbation. JSpeech Hear Res. 1980;23:274283.

23. Di Carlo N. Internal voice sensitivities in opera singers.Folia Phoniatrica et Logopaedica. 1994;46:7985.

24. Leonard R, Ringel R. Vocal shadowing under conditionsof normal and altered laryngeal sensation. J Speech HearRes. 1979;22:794817.

25. Pabon J, Plomp R. Automatic phonetogram recordingsupplemented with acoustical voice-quality parameters. JSpeech Hear Res. 1988;31:710722.

26. Michel J, Dawes P. Accuracy of pre-phonatory laryngealadjustments. In: Lawrence V, ed. Transcripts of the seventhsymposium on care of the professional voice. New York:The Voice Foundation; 1978:8587.

27. Hollien H. That golden voicetalent or training? J Voice.1993;7:195205.

Pitch Matching Accuracy of Trained Singers, Untrained Subjects with Talented Singing Voices, and Untrained Subjects with NontINTRODUCTIONMETHODSParticipantsInstrumentationStimuliProceduresAnalysis

RESULTSDISCUSSIONREFERENCES