Pitch Discrimination and Pitch Matching Abilitieswith Vocal and Nonvocal Stimuli

*Robert E. Moore, *Julie Estis, *Susan Gordon-Hickey, and Christopher Watts

*Mobile, Alabama and Harrisburg, Virginia

Summary: Various stimulus types have been investigated in pitch discrimi-nation and pitch matching tasks. However, previous studies have not exploredthe use of recorded samples of an individuals own voice in performing thesetwo tasks. The purpose of this study was to investigate pitch discriminationand pitch matching abilities using three stimuli conditions (participantsown voice, a neutral female voice, and nonvocal complex tones) to determineif pitch discrimination and/or pitch matching abilities are influenced by thetype of stimuli presented. Results of the pitch discrimination tasks yieldedno significant difference in discrimination ability for the three stimuli. Forthe pitch matching tasks, a significant difference was found for the partici-pants voice versus neutral female voice and the participants voice versustonal stimuli. There was no significant difference in pitch matching ability be-tween the neutral female voice and the tonal stimuli. There was no significantcorrelation between pitch discrimination and pitch matching abilities for anyof the three stimuli types. These results suggest that it is easier to match thepitch of ones own voice than to match the pitch of a neutral female voice andnonvocal complex tones, although no difference was found for pitch discrim-ination abilities. One possible implication of this study is that differences inmatching the pitch of ones own voice compared to matching other stimulitypes may help to differentiate the source of singing inaccuracy (motor vs dis-crimination skills).

Key Words: Pitch matchingPitch discriminationVocal stimuliNon-vocal stimuliAuditory feedback.Accepted for publication October 27, 2006.From the *Department of Speech Pathology & Audiology,

University of South Alabama, Mobile, Alabama; and theDepartment of Communication Sciences & Disorders, JamesMadison University, Harrisburg, Virginia.

Address correspondence and reprint requests to Robert E.Moore, Department of Speech Pathology & Audiology, Uni-versity of South Alabama, 2000 UCOM, Mobile, AL 36688-0002. E-mail: rmoore@usouthal.edu

Journal of Voice, Vol. 22, No. 4, pp. 3994070892-1997/$34.00 2008 The Voice Foundationdoi:10.1016/j.jvoice.2006.10.013399INTRODUCTION

The ability to accurately discriminate and matchpitches varies across individuals. Singers and trainedmusicians must exhibit accurate pitch discrimina-tion and pitch matching skills to adequately performmusical tasks. Some individuals are unable to sing intune or to hear differences among pitches. The na-ture of these differences in pitch discriminationand pitch matching abilities across individuals isuncertain. Such differences may result from inherentability, from structural or functional differences, orfrom exposure to music and musical training.1,2

mailto:rmoore@usouthal.edu

400 ROBERT E. MOORE ET ALTo accurately match pitch with the voice, indi-viduals must use accurate pitch discriminationand precise control of the vocal mechanism. Accu-rate pitch discrimination first involves hearing thepresented stimuli; therefore, normal auditory func-tion is required. Then, the pitch must be held inpitch memory while being compared.3 Higher-levelprocesses that may be involved include classifica-tion of stimuli types, identification of stimuli rela-tionships, and allocation of attention resources.Similarly, accurate pitch matching involves hearingthe target stimulus, forming an internal representa-tion of that stimulus in pitch memory, plus the plan-ning and coordination of the vocal mechanism toproduce a sound that matches the target stimulus.

One of the processes likely to be involved inpitch matching is auditory feedback. Auditory feed-back is a primary means of sensory feedback forspeech.46 Speakers appear to monitor their audi-tory feedback to produce speech that is intelligibleand in agreement with the linguistic and supraseg-mental intent of the message. Researchers haveshown that individuals attempt to modify the pitchor loudness of their voice if it does not match theauditory feedback.5,7,8 For example, when the vocalfundamental frequency (F0) being fed back to anindividual is altered, the individual responds by alter-ing the pitch of their voice in either a positive or neg-ative direction.9 While producing a pitch to matcha target stimulus, the individual monitors the pitchof their voice and compares it to the perceived pitchof the target. Adjustments to vocal production aremade based on that feedback. It is likely that individ-uals who are accurate pitch matchers possess percep-tual systems that are more finely tuned, or usedifferent strategies for monitoring auditory feedbackof their own vocal production, compared to inaccu-rate pitch matchers. This occurs although perceptionof our own voice is different when heard during pro-duction as compared to when heard after being re-corded and played through an external transducer.The perception of our voice during production is im-pacted by the fact that the sound is heard via boneconduction and air conduction, whereas when heardthrough an external transducer, the sound is heardalmost exclusively via air conduction.

Individuals with good pitch matching skills tendto have good pitch discrimination skills. ThisJournal of Voice, Vol. 22, No. 4, 2008relationship between pitch discrimination and pitchmatching has been investigated in a variety of pop-ulations. These include school-aged children, non-vocal musicians, trained singers, and individualswith no formal music training.1013 Trained musi-cians and singers often excel at pitch matchingand pitch discrimination tasks. In addition, some in-dividuals with no formal music training exhibit ac-curate pitch discrimination and pitch matchingabilities similar to those of trained vocal musicians.Watts et al13,14 reported that these individuals werejudged as being accurate singers, whereas thosewith poor pitch discrimination and pitch matchingskills tended to be judged as inaccurate singers.Studying inaccurate adult singers, Bradshaw andMcHenry15 found that all participants performedpoorly on a pitch matching task. They also reportedno correlation between pitch discrimination andpitch matching abilities with this population.

For pitch discrimination and pitch matching tasks,various stimulus types have been explored. Theseinclude pure tones, computer-generated complextones, complex tones generated by musical instru-ments, and vocal samples.1619 Previous researchinvolving pitch matching tasks have not used re-corded samples of participants own voices. In par-ticular, pitch matching accuracy for ones ownvoice has not been compared to pitch matching accu-racy with other stimuli types, such as neutral voicesamples and nonvocal complex tones. The accuracyof matching the pitch of ones own voice may holdclinical relevance, as a traditional means of therapyhas been to improve a patients perceptual capacity,via exercises or alterations in vocal F0 used torestore a normal balance of laryngeal muscularactivity.2022 If stimulus type influences discrimina-tion and reproduction accuracy relative to vocal pitchtargets, then stimulus type could be used to enhancetherapeutic management. Additionally, singers mustmonitor their own voice to achieve accurate pitchproduction. During performances singers listen totheir own voice through monitors and adjust the pitchaccordingly. The purpose of this study was to exam-ine pitch matching abilities with three types of stim-uli (viz, participants own voice, a neutral femalevoice, and nonvocal complex tones) to determine ifpitch matching abilities are influenced by the typeof stimuli presented. The research questions were

401PITCH MATCHING/DISCRIMINATION VOCAL/NONVOCAL STIMULI1. Does stimulus type influence pitch discrimi-nation and pitch matching accuracy in un-trained individuals?

2. Are there differences in pitch discriminationand pitch matching accuracy in untrained in-dividuals who are accurate and inaccuratepitch matchers, when presented with differentstimuli types?

It was expected that pitch discrimination accu-racy and pitch matching accuracy would be influ-enced by stimulus type. Specifically, it wasexpected that discrimination and pitch matchingwould be more accurate when presented with an in-dividuals own voice as the target for discriminationand reproduction. Additionally, it was expected thatindividuals who display accurate pitch matchingabilities to nonvocal complex tones would be ableto both discriminate and match pitches of varyingstimulus types consistently more accurately com-pared to inaccurate pitch matchers.

METHODS

ParticipantsParticipants included 20 females recruited from

the University of South Alabama and community.Because there appears to be an effect of age on fre-quency discrimination, only individuals between 20and 30 years of age participated.23 All participantshad hearing thresholds of 25 dB hearing level orbetter at 500, 1000, 2000, and 4000 Hz.24 Each par-ticipant completed a questionnaire to ensure thatthere was no history of voice, speech, language,and/or hearing disorders. Only potential partici-pants who indicated on the questionnaire that theyhad not received formal vocal musical trainingwere included in the study.

Equipment and stimuliAll preliminary procedures, pitch discrimination,

and pitch matching tasks were performed ina sound-attenuated room, meeting specificationsfor permissible ambient noise levels.25 The audio-metric screening was done using an audiometer,calibrated to ANSI standards.26 The pure tone stim-uli were presented through TDH-50 headphones(Teledyne, New York, NY).Stimuli for pitch discrimination and pitch match-ing tasks were created using Adobe Audition (Version1.5; Adobe Systems Inc., San Jose, CA) sound edit-ing software and the Computerized Speech Labora-tory (CSL 4500; Kay Elemetrics Corporation,Lincoln Park, NJ). Three types of stimuli were cre-ated, namely, nonvocal complex tones, neutralfemale voice samples, and samples of each partic-ipants voice. Five nonvocal complex tones weregenerated with an F0 of 212, 218, 224, 206, and200 Hz. With 212 Hz being the reference fre-quency, the other frequencies represent 50 cents,100 cents, 50 cents, and 100 cents, respec-tively. These complex tones were composed of fourequal amplitude harmonics, which were added inphase. Each tone had a total duration of 350 millisec-onds, and was gated on and off with 10 millisecondslinear amplitude ramps. All nonvocal complex toneswere recorded at equal amplitudes.

For the neutral female voice condition, a femalevolunteer, who did not participate as a listener, sus-tained the vowel ah at a comfortable pitch. Thissample was recorded using the CSL 4500 at a sam-pling rate of 44.1 kHz. Then, this sample was digi-tally manipulated using Adobe Audition to have anF0 of 212 Hz and duration of 350 milliseconds. Us-ing the same software, four additional voice sam-ples were created having the same frequenciesand durations as the nonvocal complex tones.

In a similar manner, each participants voice wasrecorded as they sustained production of the vowelah at a comfortable pitch. Acoustic analysis wasconducted via the CSL 4500 and MultidimensionalVoice Range Profile-Advanced (MDVP-A2.7.0; KayPENTAX, Lincoln Park, NJ) software to ensure thatcycle-to-cycle variations in frequency and ampli-tude (jitter and shimmer, respectively), and noise-to-harmonic ratios were within normal limits.This recording was manipulated to have the sameduration as the nonvocal complex tones and femalevoice samples. For this condition, the 50 cent,100 cent, 50 cent, and 100 cent sampleswere referenced to each participants individualF0, thus varying for each individual. The individ-uals F0 and additional tones are shown in Table 1.

For the pitch discrimination task, the stimuliwere grouped into sequential sets of three withineach condition. For each set, there were fourJournal of Voice, Vol. 22, No. 4, 2008

402 ROBERT E. MOORE ET ALpossible combinations: (1) all three stimuli havingthe same F0, (2) the F0 of the first stimulus differingfrom the second and third stimuli, (3) the F0 of thesecond stimulus differing from the first and thirdstimuli, and (4) the F0 of the third stimulus differ-ing from the first and second stimuli. A total of65 stimuli sets were generated for each of the threestimuli conditions (Appendix A).

The stimuli for the pitch matching task consistedof nonvocal complex tones, neutral female voicesamples, and samples of the participants voice atthe five generated pitches, for a total of 15 stimuli.For this task, each stimulus was presented in soundfield via a Tucker-Davis System 3 psychoacousticworkstation.

ProceduresFor the pitch discrimination task, participants

were seated at a desk inside the sound-attenuatedroom, with a computer monitor and mouse. The in-vestigator monitored the participants responsesand progress on a networked computer from

TABLE 1. Fundamental Frequencies for EachParticipants Own Voice Sample and Digitally

Derived Samples Used for Pitch Discriminationand Pitch Matching Tasks

Subject F0

F0 L 100cents

F0 L 50cents

F0 D 50cents

F0 D 100cents

1 216.72 203.87 210.15 222.35 228.272 232.96 219.34 225.53 239.77 246.543 230.75 218.30 224.26 237.53 245.544 288.45 272.51 280.81 297.39 305.985 258.00 243.76 249.70 266.39 273.506 231.00 218.85 224.70 237.22 245.007 266.00 251.14 258.00 273.67 282.488 246.57 233.22 239.80 254.42 261.009 239.77 225.80 233.09 246.75 254.0010 235.97 239.70 246.10 261.25 269.3711 265.80 251.37 258.26 272.96 281.7212 221.00 207.84 213.84 227.56 234.4513 276.60 261.14 269.20 285.16 294.3014 239.00 225.39 231.77 246.30 253.9415 240.00 227.27 233.68 248.29 255.6216 203.17 191.57 197.23 209.34 215.8017 296.11 279.00 287.39 305.20 313.5418 207.90 196.10 201.86 213.92 221.7119 219.06 206.77 212.79 225.58 231.8920 207.92 196.20 201.87 214.31 220.72Journal of Voice, Vol. 22, No. 4, 2008outside the room. The investigator was also ableto see the participant using a closed-circuit cameraand monitor system. Presentation of each series ofstimuli was controlled by ECos/Win software(AVAAZ Innovations, Ontario, Canada). The stim-uli were presented at 75 dB Sound pressure level(SPL) in sound field through the same speakerand equipment used in the pitch matching task.

Participants were instructed to listen to eachthree-tone set and indicate if the stimuli were thesame pitch or if one of the stimuli was differentin pitch. If one of the stimuli was different in pitch,the participant was instructed to identify that stim-ulus as the first, second, or third stimulus. Partici-pants recorded their judgment of the three tonesby clicking on the appropriate icon displayed onthe computer monitor with a computer mouse. Astimulus set was not presented until the participanthad made a judgment regarding the preceding set.Each of the 65 stimuli sets was presented two timesin random order. The order of the presentations wasrandomized by the ECos/Win software (AVAAZ In-novations). All participant responses were saved tothe computer hard drive for further analysis.

For the pitch matching task, each of the 15 tar-gets was randomly presented two times via soundfield through a Fostex 7301B3E amplified speaker(Fostex Corporation, Tokyo, Japan) at 75 dB SPLwith the participant seated 1.5 m from the speaker.A total of 30 targets were presented in two trials.Participants were instructed to listen to the targetpresented, and then to vocally match the pitch ofeach target with the vowel ah for 6 seconds.The investigator timed the 6-second interval. Partic-ipants pitch matching responses were recorded us-ing a head-mounted microphone routed to the CSL4500, digitized at a sampling rate of 44.1 kHz. Eachparticipants response was saved in an individualfile to the hard drive of the computer for futureanalysis.

AnalysisAfter data collection, participants were assigned

groups based on pitch matching accuracy to thenonvocal complex tones. Participants who had anaverage semitone difference score in the nonvocalstimulus condition less than 1.0 semitones wereconsidered accurate pitch matchers. All other

403PITCH MATCHING/DISCRIMINATION VOCAL/NONVOCAL STIMULIparticipants were considered inaccurate pitchmatchers, because a 1.0 semitone difference orgreater constitutes a half tone or more differenceon the musical scale.

Two dependent variables were measured in thisstudy: pitch discrimination accuracy and pitchmatching accuracy. Performance accuracy for thepitch discrimination task was measured by calcu-lating the percentage of correct score for each par-ticipant from a total of 130 stimulus presentationsfor each of the three stimuli types. Pitch matchingaccuracy was calculated by analyzing the middle2 seconds of each participants vocal match. Thistime frame was chosen to avoid including onsetand offset of vocal production, which could leadto variability in the measurement. Short-term vari-ation in frequency, which could also impact overallF0 measurement, was within normal limits for allparticipants. The F0 of each sample was calculatedby the CSL 4500 using the MDVP-A software. Anaverage of the two vocal attempts for each targetwas calculated. The frequency of the target and av-erage frequency of the pitch matching attemptswere used to calculate the difference in semitones.An average semitone difference was calculated inthis manner for each of the three stimuli condi-tions. Absolute values were used for all measuresto accurately reflect F0 differences. The use of ac-tual differences with positive numbers indicatingproductions higher in frequency than the targetstimuli and negative numbers indicating produc-tions lower in frequency than the target stimulicould have resulted in misrepresentation of an inac-curate pitch matcher as an accurate pitch matcher.A smaller score indicated more precise pitchmatching accuracy.

A 2 (group) 3 (condition) repeated measuresANOVA was performed to compare the pitch dis-crimination score means of accurate and inaccuratepitch matchers in the three stimulus conditions. Asecond 2 (group) 3 (condition) repeated measuresANOVA was performed to compare the pitchmatching response means of accurate and inaccu-rate pitch matchers in the three stimulus conditions.The presence and strength of the relationship be-tween pitch discrimination and pitch matching forthe three conditions was also evaluated using Pear-sons product-moment correlations.RESULTS

Based on pitch matching accuracy to nonvocalstimulus tones, 16 participants were accurate pitchmatchers (average semitone differences rangingfrom 0.07 to 0.61 semitones) and four were inaccu-rate pitch matchers (average semitone differencesranging from 1.04 to 5.05 semitones). For the pitchdiscrimination task, both groups were most accu-rate for their own voice. The neutral female voicesamples were most difficult to discriminate forthe accurate pitch matchers, whereas the nonvocalcomplex tones were most difficult to discriminatefor the inaccurate pitch matchers.

Individual means and standard deviations forpitch matching are presented in Table 2. In bothgroups, pitch matching of their own voice wasmost accurate for most participants. The neutral fe-male voice samples were more difficult to matchthan the nonvocal complex tones. Group meansand standard deviations for pitch matching tasksare presented in Figure 1.

TABLE 2. Individual Means and StandardDeviations of Absolute Semitone Differences forthe Three Conditions of the Pitch Matching Task

SubjectOwnVoice

FemaleVoice

NonvocalTone Average

1 0.18 (0.20) 0.07 (0.07) 0.09 (0.06) 0.152 0.89 (0.65) 2.05 (0.61) 2.00 (0.70) 1.833 0.11 (0.09) 0.19 (0.18) 0.61 (0.39) 0.494 0.18 (0.04) 0.09 (0.09) 0.07 (0.02) 0.105 0.04 (0.02) 0.13 (0.07) 0.09 (0.05) 0.096 0.75 (0.50) 0.67 (0.31) 0.31 (0.18) 0.517 0.65 (0.37) 2.21 (0.67) 1.04 (0.45) 1.388 0.09 (0.06) 0.38 (0.08) 0.40 (0.26) 0.339 0.07 (0.04) 0.14 (0.07) 0.09 (0.11) 0.1510 0.15 (0.11) 0.26 (0.13) 0.24 (0.09) 0.2311 3.01 (1.02) 5.90 (2.64) 5.05 (3.09) 5.6112 0.31 (0.25) 0.72 (0.35) 0.47 (0.46) 0.5613 0.44 (0.28) 2.31 (0.81) 1.85 (0.50) 1.6414 0.09 (0.07) 0.10 (0.09) 0.14 (0.14) 0.1215 0.10 (0.08) 0.21 (0.11) 0.18 (0.08) 0.1716 0.14 (0.15) 0.28 (0.15) 0.20 (0.10) 0.2117 0.22 (0.15) 0.40 (0.12) 0.37 (0.16) 0.4818 0.51 (0.44) 0.29 (0.18) 0.31 (0.31) 0.3419 0.27 (0.07) 0.16 (0.10) 0.21 (0.18) 0.1820 1.53 (1.04) 0.70 (0.68) 0.41 (0.34) 1.13

Note: Standard deviations are in parenthesis. Boldface typeindicates the inaccurate pitch matching group.Journal of Voice, Vol. 22, No. 4, 2008

404 ROBERT E. MOORE ET ALFor the pitch discrimination task, a repeatedmeasures ANOVA indicated no significant main ef-fects for stimulus type and no interaction betweenstimulus type and pitch matching accuracy. Multi-ple bivariate correlation analyses using Pearsonproduct-moment correlations were performed to in-vestigate the relationship between pitch discrimina-tion and pitch matching for each stimuluscondition. For each of the three stimuli conditions,correlations were nonsignificant.

For the pitch matching task, a repeated measuresANOVA was used to compare pitch matching per-formance across groups and among conditions.Mauchlys Test of Sphericity was significant(W 5 0.481); therefore, sphericity was not as-sumed, and Huynh-Feldt corrected tests were usedto measure significance. Results of the ANOVA in-dicated a significant main effect for stimulus type,F 5 31.626, P ! 0.05, and a significant interactionbetween stimulus type and pitch matching accu-racy, F 5 29.217, P ! 0.05. The significant differ-ence for stimulus type was further studied bythree paired comparisons: own voice versus neutralfemale voice, own voice versus nonvocal complextone, and neutral female voice versus nonvocalcomplex tone. To control for type I errors, theHolms sequential Bonferroni procedure was usedto adjust the alpha level. Pairwise comparisons in-dicated a significant difference between the partic-ipants matching of their own voice and the other

FIGURE 1. Mean semitone difference scores for individualswith accurate and inaccurate pitch matching abilities in fourstimuli conditions.Journal of Voice, Vol. 22, No. 4, 2008two stimuli conditions. There was no significantdifference between the neutral female voice condi-tion and the nonvocal complex tone condition. Sta-tistical summary of paired comparisons acrossgroups and stimuli for pitch matching is presentedin Table 3.

DISCUSSION AND CONCLUSION

The purpose of this study was to investigatewhether stimulus type influences the accuracy ofpitch discrimination and vocal pitch matching.Three stimulus types were compared: (1) nonvocalcomplex tones, (2) a neutral female voice, and (3)ones own voice. In addition, this study sought todetermine if pitch discrimination and pitch match-ing abilities were differentially influenced by stim-ulus type in individuals who were accurate pitchmatchers and those who were inaccurate pitchmatchers.

The ability to perceptually discriminate the pitchof comparison tones was not significantly influ-enced by stimulus type. However, the results indi-cated that individuals are more accurate atmatching the pitch of their own voice comparedto another female voice or a nonvocal complextone. One reason for this result may have to dowith the frequency of the stimuli. The complextone and the female voice were at set frequencies,whereas the frequencies for the participants ownvoice varied across participants. The stimuli createdfrom the participants own voice were produced byhaving the participant sustain ah at their mostnatural pitch. The F0 of the sample was used asthe base frequency for each participant. From

TABLE 3. Statistical Summary of PairedComparisons Across Stimuli Conditions

for Pitch Matching

PairedComparisons

MeanDifference

StandardError Significance

Own voicevs female voice

0.957 0.115 0.000*

Own voicevs complex tone

0.746 0.164 0.001*

Female voicevs complex tone

0.210 0.088 0.082

Note: * indicates significant difference between conditions.

405PITCH MATCHING/DISCRIMINATION VOCAL/NONVOCAL STIMULIeach sample obtained, five stimuli were created.These were at F0, F0 100 cents, F0 50 cents,F0 50 cents, and F0 100 cents. Thus, the F0of the samples with the participants own voice dif-fered from the frequency of the complex tone andfemale voice stimuli. The fact that the own voicestimuli were at or near the natural pitch of the par-ticipant may have made pitch matching easier. Thisis emphasized by comparing the accurate and inac-curate pitch matching groups. The accurate pitchmatchers were generally accurate for all three stim-uli. However, the inaccurate pitch matchers weremost accurate for their own voice.

Based on listening to ones own voice from arecording device, it might seem surprising that anindividual is able to match their own voice moreaccurately. The reason ones own voice sounds dif-ferent from a recording device is that when heardnaturally, the voice is heard by both bone conduc-tion and air conduction. When heard through a re-cording device it is heard through air conductiononly. It appears from the findings of this study thatF0 is maintained when heard through both bone con-duction and air conduction. The difference we hearis due to changes in the spectrum of the sound.

If individuals show good pitch discriminationand matching abilities, their mechanism for complet-ing these tasks appears to be more flexible as they areable to accurately match pitch regardless of the stim-ulus type. However, for individuals who are less ac-curate and less flexible, they may have difficulty withstimuli that differ from their natural range of produc-tion. In other words, when presented with a stimulushaving a pitch different from their own natural pitch,they will tend to produce a sound near their naturalpitch. Continued exploration of the impact of onesown voice as the stimulus for pitch matching is war-ranted. Specifically, comparing the individuals per-formance at or near their natural F0 with other typesof vocal and tonal stimuli will provide insight regard-ing the nature of the improved performance seen inthe present study.

In the present study, no correlation was foundbetween pitch discrimination ability and pitchmatching ability. This is not in agreement with pre-vious work in this area, which has shown a signifi-cant correlation between pitch discriminationability and pitch matching ability.13 One possiblereason for this difference is that in previous re-search participants were asked to determine if twostimuli presented were the same or different inpitch. In this study, participants were asked tomake pitch discrimination judgments between threestimuli. Participants indicated if the three stimuliwere the same in pitch or identified the stimuli ofthe three with the odd pitch. In reviewing the datafrom this study and previous studies, it would ap-pear that the three stimuli paradigm was easierthan the two stimuli paradigm even though fre-quency differences between stimuli were similar(eg, within one semitone or less). In general, partic-ipants in this study were more accurate pitchdiscriminators than in previous studies. All partici-pants in this study achieved more than 80% accu-racy on the pitch discrimination task. It may bethat hearing two stimuli of the same frequency so-lidified the representation of the pitch in workingmemory, thereby making it easier to correctly dis-criminate the pitch of the differing stimulus. Futureresearch should investigate differences in pitch dis-crimination performance by comparing a two stim-uli paradigm to a three stimuli paradigm.

Another possible factor for the lack of agreementin correlation of pitch discrimination and pitchmatching findings between this study and previousresearch relates to different or multiple profiles of in-accurate singers. Bradshaw and McHenry15 studiedadults who sing inaccurately and did not find a signif-icant correlation between pitch matching and pitchdiscrimination. Their results indicated two profilesof inaccurate singers. The two profiles were thosewho were accurate pitch discriminators and poorpitch matchers and those who performed poorly atboth tasks. They suggested that those who wereaccurate pitch discriminators and inaccurate pitchmatchers lacked coordination and flexibility of thevocal mechanism. Because pitch discriminationunderlies pitch matching, the limited pitch discrim-ination abilities of those who were inaccurate forboth tasks hindered their pitch matching ability. Ofthe four participants who were inaccurate pitchmatchers, two participants were accurate pitch dis-criminators (less than 13 errors in 390 trials). Thefocus of future research should include a greaternumber of participants having inaccurate pitchmatching and discrimination abilities.Journal of Voice, Vol. 22, No. 4, 2008

406 ROBERT E. MOORE ET ALOne implication of this study is that using sam-ples of an individuals own voice for pitch matchingtasks may help differentiate the source of singinginaccuracy (motor skills vs discrimination skills).If an individual possessing accurate pitch discrimi-nation ability performs better on pitch matchingtasks with their own voice than with tonal stimulus,it may indicate a lack of flexibility or coordinationof the vocal mechanism required for accurate sing-ing. For these individuals, teachers of singing mightfocus on improving the motor aspects of singingrather than improving pitch discrimination abilities.Also, these individuals may have more potential forsinging improvement than individuals who performpoorly with both types of stimuli. For those individ-uals with poor discrimination, regardless of stimulitype, teachers of singing might first focus onimproving discrimination abilities because pitchdiscrimination is a foundation of pitch matchingability.

REFERENCES

1. Brashears S, Morlet T, Berlin C, Hood L. Olivocochlearefferent suppression in classical musicians. J Am AcadAudiol. 2003;14:314324.

2. Gaab N, Gaser C, Zaehle T, Jancke L, Schlaug G. Functionalanatomy of pitch memoryan fMRI study with sparse tem-poral sampling. Neuroimage. 2003;19:14171426.

3. Deutsch D. The deterioration of pitch information in mem-ory [unpublished doctoral dissertation]. San Diego, CA:University of California at San Diego; 1970.

4. Fairbanks G. Selective vocal effects on delayed auditoryfeedback. J Speech Hear Disord. 1955;20:333346.

5. Siegle G, Pick H. Auditory feedback in the regulation ofvoice. J Acoust Soc Am. 1974;56:16181624.

6. Siegle G, Kennard K. Lombard and sidetone amplificationeffects in normal and misarticulating children. J SpeechHear Res. 1984;27:5662.

7. Fairbanks G. Systematic research in experimental phonet-ics. I. A theory of the speech mechanism as a servosystem.J Speech Hear Disord. 1954;19:133140.

8. Howell P, El-Yaniv N, Powell D. Factors affecting fluencyin stutters. In: Peters HFM, Hulstijin W, eds. Speech MotorDynamics in Stuttering. New York: Springer-Verlag; 1987:361369.

9. Elman J. Effects of frequency-shifted feedback on the pitchof vocal productions. J Acoust Soc Am. 1981;70:4550.Journal of Voice, Vol. 22, No. 4, 200810. Brody VA. The emergence of song. Music Educ J. 1949;36:2224.

11. Amir O, Amir N, Kishon-Rabin L. The effect of superiorauditory skills on vocal accuracy. J Acoust Soc Am.2003;113:11021108.

12. Pitt M. Perception of pitch and timbre by musically trainedand untrained listeners. J Exp Psychol. 1994;20:976986.

13. Watts C, Moore R, McCaghren K. The relationship be-tween vocal pitch-matching skills and pitch discriminationskills in untrained accurate and inaccurate singers. J Voice.2005;19:534543.

14. Watts C, Murphy J, Barnes-Burroughs K. Pitch-matchingaccuracy of trained singers, untrained participants withtalented singing voices, and untrained participants withnon-talented singing voices in conditions of varyingfeedback. J Voice. 2003;17:187196.

15. Bradshaw E, McHenry M. Pitch discrimination and pitchmatching abilities of adults who sing inaccurately. J Voice.2005;19:431439.

16. Moore B, Glasberg B, Proctor G. Accuracy of pitch match-ing for pure tones and for complex tones with overlappingand non-overlapping harmonics. J Acoust Soc Am. 1992;91:34433450.

17. Singh P, Hirsh I. Influence of spectral locus and F0 changeson the pitch and timbre of complex tones. J Acoust SocAm. 1992;92:26502662.

18. Jarvelainen H, Verma T, Valimaki V. Perception andadjustment of pitch in inharmonic string instrument tones.J New Music Res. 2002;31:311319.

19. Schueller M, Fucci D, Bond Z. Perceptual judgment ofvoice pitch during pitch matching tasks. Percept MotSkills. 2002;94:967974.

20. Boone D, McFarlane S, Von Berg S. The Voice and VoiceTherapy. Boston: Allyn & Bacon; 2005. 201206.

21. Stemple J. Clinical Voice Pathology: Theory and Manage-ment. San Diego: Singular; 2000. 304310.

22. Mathieson L. The Voice and Its Disorders. London:Whurr; 2001. 512554.

23. He N, Dubno J, Mills J. Frequency and intensity discrim-ination measured in a maximum likelihood procedure fromyoung and aged normal-hearing subjects. J Acoust Soc Am.1998;103:553565.

24. American Speech-Language-Hearing Association. Guide-lines for manual pure-tone audiometry. ASHA. 1978;20:297301.

25. American National Standards Institute. Maximum Permis-sible Ambient Noise Levels for Audiometric Test Rooms.ANSI S3.1-1991. New York: American National StandardsInstitute; 1991.

26. American National Standards Institute. Specifications foraudiometers. ANSI S3.6-1996. New York: AmericanNational Standards Institute; 1996.

407PITCH MATCHING/DISCRIMINATION VOCAL/NONVOCAL STIMULIAPPENDIX A. SIXTY-FIVE COMBINATIONS OF PITCH DISCRIMINATIONTASK STIMULI SETS

F0/F0/F0 50/50/50 100//100/100 50/50/50 100/100/100F0/50/F0 50/F0/50 100/F0/100 50/F0/50 100/F0/100F0/100/F0 50/100/50 100/50/100 50/50/50 100/50/100F0/50/F0 50/50/50 100/50/100 50/100/50 100/100/100F0/100/F0 50/100/50 100/100/100 50/100/50 100/50/100F0/F0/50 50/50/F0 100/100/F0 50/50/F0 100/100/F0F0/F0/100 50/50/100 100/100/50 50/50/50 100/100/50F0/F0/50 50/50/50 100/100/50 50/50/100 100/100/100F0/F0/100 50/50/100 100/100/100 50/50/100 100/100/5050/F0/F0 F0/50/50 F0/100/100 F0/50/50 F0/100/100100/F0/F0 100/50/50 50/100/100 50/50/50 50/100/10050/F0/F0 50/50/50 50/100/100 100/50/50 100/100/100100/F0/F0 100/50/50 100/100/100 100/50/50 50/100/100

Notes: Each of the combinations was presented two times for each of three conditions (nonvocal complex tone, neutral female voice,and own voice). 50 550 cents relative to F0; 100 5100 cents relative to F0; 50 550 cents relative to F0; 100 5100cents relative to F0.Journal of Voice, Vol. 22, No. 4, 2008

Pitch Discrimination and Pitch Matching Abilities with Vocal and Nonvocal StimuliIntroductionMethodsParticipantsEquipment and stimuliProceduresAnalysis

ResultsDiscussion and conclusionReferencesSixty-five combinations of pitch discrimination task stimuli sets