A Comparison of Trained and Untrained Vocalists on

the Dysphonia Severity Index

*Shaheen N. Awan and †Anysia J. Ensslen, *Bloomsburg, Pennsylvania, yLexington, Kentucky

Summary: The purposes of this study were (1) to compare trained and untrained singers on the Dysphonia Severity

AccepPortion

fessionalFrom

Pennsylvences, Co

Addreology &ond Stree

Journa0892-1� 201doi:10

Index (DSI) and its component measures, and (2) to contribute to normative DSI data for trained singers. This studyincluded 36 untrained participants (15 males and 21 females) and 30 participants (15 males and 15 females) with singingexperience between the ages of 18 and 30 years. Measures of maximum phonation time (MPT), highest phonationalfrequency, lowest intensity, and jitter were obtained for each subject and incorporated into the previously published mul-tivariate DSI formula. Results indicated that vocally trained subjects have significantly higher DSI scores than untrainedsubjects (mean DSI: 6.48 vs 4.00, respectively), with significant differences observed between trained and untrainedgroups for three of the four components of the DSI (F0 high; I low; jitter). The findings of this study are consistentwith previous reports that indicate significant increases in the DSI with vocal training, and with various studies thathave observed increased vocal capability in trained singers versus their untrained counterparts. The results of this studyindicate that alternative normative expectations for the DSI may need to be taken into account when using the DSI withpatients who have participated in directed vocal training, such as choral participation and voice/singing lessons.Key Words: Dysphonia Severity Index–Singers–Maximum phonation time–Jitter–Phonational frequency range.

INTRODUCTION

The perceptual evaluation of voice is considered to be an essen-tial aspect of the conventional voice diagnostic that has rele-vance to most voice-disordered patients and provides a globalmeasure of vocal performance readily available to all clini-cians.1 Although perceptual evaluation of voice has obvious im-portance, there are several limitations associated with thismethod of assessment that clearly influence its clinical utility.These limitations include problems with scale validity and reli-ability, particularly for midscale (ie, mild to moderate) patho-logical voices; lack of credibility for medical-legal purposes;poorly defined and/or shifting definitions of severity; and the in-trusive effects of voice and speech characteristics other than thequality dimension that is meant to be judged.1–3 Many of theselimitations stem from the attempt to describe the voice viaa temporary auditory impression of the acoustic signal. As a re-sponse, voice clinicians and researchers have added to the per-ceptual assessment of voice quality with other methods thatprovide a permanent record of the vocal behavior and allowfor a more objective analysis of the patient’s voice quality.Acoustic methods of voice analysis have been primary toolsof both the clinician and the researcher for many years. Thesemethods have become widely used in both research and clinicalsituations since the advent of relatively low-cost personal com-puters and analog-to-digital acquisition hardware in the early1990s, and have the benefits of being noninvasive; readily avail-able at relatively low cost compared with other methods ofvoice analysis; applicable to treatment and diagnosis; and are

ted for publication April 2, 2009.s of this article were presented at the 37th Annual Symposium: Care of the Pro-Voice, Philadelphia, PA, in May 2008.

the *Department of Audiology & Speech Pathology, Bloomsburg University ofania, Bloomsburg, Pennsylvania; and the yDepartment of Rehabilitation Sci-llege of Health Sciences, University of Kentucky, Lexington, Kentucky.

ss correspondence and reprint requests to Shaheen N. Awan, Department of Audi-Speech Pathology, Bloomsburg University of PA, Centennial Hall, 400, East Sec-t, Bloomsburg, PA 17815-1301. E-mail: sawan@bloomu.edul of Voice, Vol. 24, No. 6, pp. 661-666997/$36.00

0 The Voice Foundation.1016/j.jvoice.2009.04.001

supported by a substantial body of literature.4 Multiparameteracoustic models, which may be used to characterize voice func-tion and quantify the severity of dysphonia, have been pre-sented by Michaelis et al,5 Callan et al,6 Frohlich et al,7 andAwan and Roy.8–10

An alternative method for encompassing the multidimen-sional nature of the normal versus disordered voice is the Dys-phonia Severity Index (DSI). The DSI was developed by Wuytset al,11 with the purpose of developing an index that would bothobjectively and quantitatively correlate with perceived voicequality. The DSI makes use of a combination of several voicemeasures that may be obtained from voice-assessment proce-dures, such as the voice range profile and basic aerodynamicand acoustic analyses: the highest phonational frequency (F0

high in Hz), lowest intensity (I low in dB), maximum phonationtime (MPT in seconds), and jitter (%). Because voice has beendescribed as a multiparameter behavior, it appears reasonablethat a multiparameter model, such as that presented in theDSI, may be useful in describing vocal function.12 The compo-nents of the DSI form a specific combination of acoustic voicemeasures that may aid in characterizing various types of vocaldysfunction. Wuyts et al11 stated that, when extra mass is evenlydistributed along the true vocal fold(s), the higher vibratoryrates become dampened. The result is a decrease in the upperreaches of the phonational frequency range. Structural changesto the true vocal folds, such as distributed or organized mass le-sions, may increase glottal resistance such that greater subglot-tal pressures will be necessary to initiate and maintain vocalfold vibration. Consequently, the lowest phonational intensitywill often be increased. In addition to changes in vocal intensity,vocal fold pathology (eg, unilateral or bilateral organized le-sions or distributed tissue change; organic pathology affectingthe ability to effectively tense or approximate the folds duringphonation) often results in disturbances in the periodicity ofphonation. These disturbances may be described in terms of jit-ter, a measure of short-term instability that quantifies cycle-to-cycle variations in frequency and has been used to assess thedegree of perturbation in the voice signal. Finally, MPT has

Journal of Voice, Vol. 24, No. 6, 2010662

been regarded as a general measure of phonatory ability13 thatreflects the function of several mechanisms necessary for voiceproduction, such as respiratory capacity and control, subglotticpressure, airflow resistance, and closure of the vocal folds. TheDSI was initially described and validated in a study by Wuytset al,11 in which the authors obtained various acoustic and com-monly used aerodynamic measures from 387 subjects (68 nor-mal controls vs 319 voice-disordered subjects). In addition,each patient’s voice was perceptually rated using the grade,roughness, breathiness, asthenia, strain (GRBAS) scale.13 TheDSI was obtained via multiple regression analyses and consistsof four weighted variables in the equation: DSI¼ 0.13 3 MPT(seconds) + 0.0053 3 F0 high (Hz)� 0.26 3 I low (dB)�1.18 3 jitter (%) + 12.4. Results of the study indicated an in-verse relationship between the DSI and the grade (overall sever-ity) of dysphonia, as well as between the DSI and the VoiceHandicap Index (VHI). The DSI was transformed such thata DSI¼ +5 corresponded to G0 (normal voice), and a DSI¼�5corresponded to G3 (severe dysphonia). It was noted by theseauthors that the DSI is not necessarily restricted to the +5 to�5 range. A DSI of +1.6 was determined to be the cutoff forperceptually normal voices.

Several studies have used the DSI to objectively describe nor-mal versus disordered voice characteristics and change in voiceover time. Timmermans et al14 evaluated voice quality changein 68 students (49 received voice training; 19 served as an un-trained control group). The vocally trained group was providedwith instruction regarding relaxation, posture, breathing pat-tern, and active articulation. Results showed a significant in-crease in the DSI for the trained group (from 2.3 to 4.5)versus no significant change in the untrained group. In a secondstudy, Timmermans et al15 used the DSI as a method of evalu-ating the effectiveness of a voice-training program for 23 pro-fessional voice users. The DSI scores were again observed tosignificantly increase from the time of training onset (meanDSI¼ 2.0) to 9 months post-training onset (mean DSI¼ 3.7)and to 18 months post-onset (mean DSI¼ 4.6), with the mostprominent voice-characteristic change being increases in F0

high. Hakkesteegt et al16 investigated the possible influenceof age and gender on the DSI (69 females; 49 males between20 and 79 years of age). Although significant differences be-tween males and females were observed for F0 high andMPT, the mean DSI between the genders was not significantlydifferent (mean DSI: females¼ 4.3; males¼ 3.8). The DSI wasobserved to decrease significantly with age in both genders, pri-marily because of reductions in F0 high and low intensity. Theseauthors inferred that the DSI of a particular voice patient shouldbe compared with appropriate normative data from comparableage and gender subjects. Woisard et al17 examined the possiblecorrelation between a French version of the VHI and quantita-tive methods of describing voice, including the DSI. In contrastto Wuyts et al,11 no significant correlation between the VHI andthe DSI was observed. Woisard et al17 indicated that the DSIshould be seen as a source of clinical information independentof the VHI. Hakkesteegt et al18 reported on the interobserverand test-retest variability of the DSI. Thirty normal subjectswere measured by two speech pathologists on three different

days (approximate interval of 1 week between measurements).The interobserver variability in measurement was observed tobe low with very little influence on DSI variability. In addition,any differences in DSI measurement between the observerswere reported as nonsignificant. These authors determinedthat intrasubject DSI changes needed to exceed 2.49 to be sig-nificant. Hakkesteegt et al19 investigated the possible relation-ship between the GRBAS scale and the DSI. The subjectsincluded 294 voice-disordered patients and 118 normal con-trols. Significantly lower DSI and higher grade (overall sever-ity) scores were observed in disordered versus control groups.In addition, a DSI score of 3.0 was observed to discriminate be-tween control and disordered groups with sensitivity¼ 0.72 andspecificity¼ 0.75. It was observed that DSI scores may be re-duced even for patients with overall grade/severity scores¼ 0,indicating that the voice complaints of some patients may notbe solely quality based. Most recently, Hakkesteegt et al20 ex-amined the possible relationship between the DSI and theVHI. Pre- and postintervention measures were obtained from171 voice-disordered patients. The subjects were divided intovoice therapy, surgical intervention, and no intervention groups.Consistent with Woisard et al,17 results indicated that the DSIand VHI measure different aspects of a voice disorder, withthe VHI being a measure of patient perception and the DSIa measure of vocal performance/capability. Although bothmethods were able to show differences between pre- and post-intervention groups, these authors indicated that DSI and VHIare not necessarily related.

The DSI has been reported to be a valuable clinical tool forthe quantitative description of normal versus disordered voice.However, as indicated by Hakkesteegt et al,16 extended norma-tive data that will provide focused comparisons for subgroupsof the normal population are necessary for appropriate clinicalinterpretations. A subgroup of the normal population that maybe seen for voice complaints is that of the trained singer. Voicetraining has been said to influence the morphology and the con-trol over the voice source.21 Although inconclusive, several re-search studies have reported that trained singers may haveincreased respiratory capacities and different respiratory pos-tures as compared with untrained subjects;22–25 may havegreater F0 ranges than the normal untrained population;26–28

and greater dynamic range capability.27,29,30 In addition, Sulteret al29 speculated that trained singers may have developed im-proved breath control during voicing, resulting in the ability toproduce vocal fold oscillation at lower subglottal pressures.Whether these reported differences between trained and un-trained singers are specifically the result of training itself; theregular participation in experiences such as choral singing; orto inherent physiological differences in those who choose toregularly participate in directed singing, is unclear. However,with these possible differences in mind, it would appear thatDSI values for trained singers may be substantially differentfrom those reported for untrained participants. Because of pos-sible increased vocal capabilities, it may be that a trained singerwho is experiencing some aspect of vocal dysfunction couldstill obtain a DSI score within the expected values reportedby Wuyts et al.11 The availability of data from trained singers

Shaheen N. Awan and Anysia J. Ensslen Comparing Trained and Untrained Vocalists on DSI 663

will make normative expectations for this index more applica-ble to a greater base of potential patients, and allow for moreappropriate clinical decisions regarding this population. There-fore, the purposes of this study were (1) to compare trained ver-sus untrained singers on the DSI and its component measures,and (2) to contribute to normative DSI data for trained singers.

METHODOLOGY

Subjects

The methodology used in this study adhered to the basic ethicalconsiderations for the protection of human participants in re-search and was approved by the Bloomsburg University Institu-tional Review Board. This study included 66 subjects (36untrained participants [15 males and 21 females]; 30 partici-pants [15 males and 15 females] with singing experience) be-tween the ages of 18 and 30 years. An untrained participanthad no formal voice/singing lessons and no directed singing ex-perience (choral experience, etc) beyond middle school age. Atrained participant had participated in directed singing experi-ence since middle school age. Directed singing experience in-cluded private voice lessons and/or regular participation ina choir or musical theater under the guidance of a trained mu-sical director. The criteria used to gauge singing experiencein this study were similar to those used by Sulter and Peters31

and McCrae and Morris.32 The trained male and female sub-jects reported receiving an average of 3.50 and 4.89 years of pri-vate voice lessons, respectively (Table 1). At the time of testing,all trained subjects were either (1) currently participating invoice lessons, (2) were participating in choral singing or musi-cal theater, or (3) had participated in choral singing or musicaltheater within 2 years before testing. Seventeen of the 30 trainedsubjects were simultaneously participating in voice lessons andchoir and/or musical theater.

Each subject was asked to complete a brief questionnaire,which included the subjects’ contact information, age, height,and a description of previous vocal training and vocal experi-ence (choirs, theater, etc) since middle school age. All subjectsreported no history of significant injury or trauma to the vocalfolds or larynx, passed a hearing screening of 20 dB at 0.5, 1, 2,and 4 kHz, and were perceived as having normal voice charac-teristics of pitch, loudness, and quality. Subject demographic

TABLE 1.

Means of Subjects’ Age and Height, As Well As Mean Number o

the Trained Subjects

Demographics Trained Males Trained Fe

Age (y) 20.40 (1.76) 21.46 (2

Height (cm) 175.43 (5.03) 164.25 (9

Voice lessons (y) 3.50* (1.84) 4.89* (4

Participation in choir,

musical theater (y)

7.93y (3.98) 8.00y (2

SDs are provided in parentheses.

Abbreviation: N/A, not applicable; y, years.

* Average of 2.0 h/wk.y Average of 3.5 h/wk.

data for age, height, and singing experience are provided inTable 1.

Procedures

Each subject was asked to complete the following tasks:

1. Highest phonational frequency (F0 high): Each subjectwas asked to go up a scale (using the vowel / /) untilthey reached their highest pitch level without losing con-trol of the voice (ie, no pitch or phonation breaks). Thehighest pitch level was digitally recorded at 44.1 kHz,16 bits of resolution using Sound Forge v. 4.033 and lateranalyzed using the TF32 speech/voice analysis pro-gram34 for the highest frequency level (in Hz). Three tri-als were elicited.

2. Lowest intensity level (I low): Each subject was asked tosustain the vowel /a/ at a comfortable pitch as quietly aspossible. Three trials were elicited. Each production wasanalyzed for the intensity level (in dB) using AerophoneII.35 Calibration of all key components of the AerophoneII system was conducted according to the manufacturer’sinstructions. According to the manual, the microphoneincluded in the Aerophone II system (AKG typeCK77P-3L electret condenser microphone, AKG Acous-tics GmbH, Vienna, Austria) is precalibrated by means ofa Bruel and Kjaer Integrating Precision Sound LevelMeter (Type 2230, Bruel & Kjaer Sound and VibrationMeasurement A/S, Naerum, Denmark). A calibrationfactor is provided with the system and entered when in-stalling the system. The mouth-to-microphone distancewas 15 cm. To convert this to the commonly used 30-cm mouth-to-microphone distance, 6 dB may be sub-tracted from the 15-cm mouth-to-microphone distanceintensity level. Sound intensity is expected to decreaseby 6 dB when the distance from the source is doubled.36

3. Jitter: To obtain a measure of jitter, the subject was in-structed to chant ‘‘1, 2, 3, 4’’ at a comfortable pitch andloudness and then sustain the vowel / / for 2–3 secondsat that similar pitch level. This elicitation method wasused to obtain a vowel sample that closely approximatesthe subjects’ habitual speaking pitch.4 Three trials were

f Years of Participation in Directed Singing Experiences for

males Untrained Males Untrained Females

.92) 21.66 (2.47) 22.57 (4.03)

.44) 179.15 (6.91) 164.13 (5.53)

.01) N/A N/A

.93) N/A N/A

TABLE 2.

Means and Standard Deviations for the DSI and Its

Component Measures for the Trained vs. Untrained

Groups (Males and Females Combined)

Variable

Trained Vocalists

n¼ 30

Untrained Vocalists

n¼ 36

MPT (s) 20.43 (5.30) 21.67 (8.01)

F0 High (Hz) 805.01 (227.01) 591.86 (185.64)

Journal of Voice, Vol. 24, No. 6, 2010664

elicited. Each production was digitally recorded at44.1 kHz, 16 bits of resolution using Sound Forge v.4.0. The central 1 second of each sustained vowel waslater analyzed for jitter (%) using the TF32 program.

4. MPT: When measuring MPT, the subject was asked tohold out a sustained vowel / / for as long as possible aftera maximum inhalation. Three trials were elicited. Eachtrial was timed with a digital stopwatch to calculate theduration (in seconds).

I low (dB) 47.97 (5.99) 53.34 (4.26)

Jitter (%) 0.31 (0.11) 0.41 (0.20)

DSI 6.48 (2.19) 4.00 (2.04)

SDs are provided in parentheses.

Dysphonia severity index formula

As previously mentioned, the DSI formula is derived froma weighted combination of the following vocal parameters:highest frequency (Hz), lowest intensity (dB), MPT (seconds),and jitter (%). For each subject, the DSI was calculated usingthe maximum performances for F0 high and MPT, the lowest in-tensity, and the lowest jitter.18 The results were entered into thefollowing formula:

DSI ¼ 0:133MPT ðsecondsÞ þ 0:00533F0 high ðHzÞ� 0:263I low ðdBÞ � 1:183jitter ð%Þ þ 12:4

RESULTS

All statistical procedures were conducted using SPSS v. 15.0 forWindows.37 Where necessary, significant interaction effectswere analyzed using Tukey Honestly Significant Different(HSD) means comparison tests.

Analysis of maximum phonation time (Seconds)

A two-way analysis of variance (ANOVA) was conducted toassess the possible differences in MPT (two levels of group—trained vs untrained subjects; two levels of gender). Resultsindicated no significant difference between the trained versusuntrained groups (F (1, 62)¼ 0.76; P¼ 0.38). Although therewas a tendency for males to produce longer MPTs than females(22.78 seconds [standard deviation (SD)¼ 6.94] vs 19.71 sec-onds [SD¼ 6.62], respectively), no significant gender effect(F (1, 62)¼ 3.56; P¼ .06) and no significant interaction effects(F (1, 62)¼ 0.02; P¼ 0.90) were observed. Means and SDs forthe MPTs of the trained versus untrained subjects, as well as formale and female subgroups are provided in Tables 2 and 3.

Analysis of maximum frequency (Hz)

A two-way ANOVAwas conducted to assess the possible differ-ences in maximum phonational frequency (F0 high). Resultsindicated significant main effects of group (F (1, 62)¼ 26.71;P < 0.0001), and gender (F (1, 62)¼ 24.31; P < 0.0001). Resultsindicated that the mean F0 high for the trained subjects wassignificantly greater than that observed for untrained subjects(805.01 vs 591.86 Hz, respectively). As expected, female sub-jects produced significantly greater mean F0 high than male sub-jects (776.01 Hz [SD¼ 226.69] vs 584.03 Hz [SD¼ 190.15],respectively). No significant interaction of group and gender

was observed (F (1, 62)¼ 1.85; P¼ 0.18). Means and SDs forF0 high are provided in Tables 2 and 3.

Analysis of low intensity (dB)

A two-way ANOVAwas conducted to assess the possible differ-ences in low intensity phonation (I low). Results indicated a sig-nificant main effect of group (F (1, 62)¼ 26.02; P < 0.0001),with trained subjects observed to produce significantly lower-intensity productions than untrained subjects (47.97 vs53.34 dB, respectively—see Table 2). No significant effect ofgender was observed (F (1, 62)¼ 0.01; P¼ 0.98). However,a significant interaction of group and gender was observed(F (1, 62)¼ 18.94; P < 0.0001). Post hoc investigation of thesignificant interaction effect using Tukey HSD means compari-son tests indicated that untrained females produced significantlylower intensity phonations compared with untrained males. Incontrast, trained males produced significantly lower-intensityphonations than trained females. In addition, trained males pro-duced significantly lower-intensity phonations than untrainedmales. No significant difference was observed between un-trained and trained females (Table 3).

Analysis of jitter (%)

A two-way ANOVAwas conducted to assess the possible differ-ences in jitter (%). Results indicated a significant main effect ofgroup (F (1, 62)¼ 4.81; P¼ 0.03), with results indicating thattrained subjects had significantly lower jitter than untrainedsubjects (0.31% vs 0.41%, respectively—see Table 2). No sig-nificant effects of gender (F (1, 62)¼ 0.25; P¼ 0.62) or groupand gender interaction (F (1, 62)¼ 0.52; P¼ 0.47) were ob-served (Table 3).

Analysis of the dysphonia severity index

A two-way ANOVAwas conducted to assess the possible differ-ences in the DSI. Results indicated a significant main effect ofgroup (F (1, 62)¼ 26.15; P < 0.0001), with the trained subjectsobserved to have significantly higher DSI scores than untrainedsubjects (6.48 vs 4.00, respectively—see Table 2). No signifi-cant gender effect was observed (F (1, 62)¼ 1.85; P¼ 0.18).Although a tendency was observed for untrained females tohave greater DSI scores than untrained males, no significant

TABLE 3.

Means and SDs for the DSI and Its Component Measures for the Trained Versus Untrained Males and Females

Variable

Trained Males

(n¼ 15)

Trained Females

(n¼ 15)

Untrained Males

(n¼ 15)

Untrained Females

(n¼ 21)

MPT (s) 22.14 (5.53) 18.73 (4.61) 23.42 (8.26) 20.42 (7.78)

F0 high (Hz) 667.31 (197.89) 942.71 (164.16) 500.75 (144.47) 656.94 (187.15)

I low (dB) 45.49 (6.63) 50.45 (4.17) 56.19 (4.24) 51.30 (2.94)

Jitter (%) 0.32 (0.12) 0.31 (0.10) 0.38 (0.13) 0.43 (0.24)

DSI 6.61 (2.52) 6.35 (1.90) 3.04 (1.98) 4.69 (1.83)

SDs are provided in parentheses.

Shaheen N. Awan and Anysia J. Ensslen Comparing Trained and Untrained Vocalists on DSI 665

group and gender interaction was observed (F (1, 62)¼ 3.49;P¼ 0.07—see Table 3).

DISCUSSION

The results of this study indicate that vocally trained subjectshave significantly higher DSI scores than untrained subjects.Significant differences were observed between trained and un-trained groups for three of the four components of the DSI (F0

high, I low, jitter). The findings of this study are consistent withthe reports of Timmermans et al,14,15 which indicated signifi-cant increases in the DSI with vocal training, and with variousstudies which have observed increased vocal capability intrained singers versus their untrained counterparts.23–30 Al-though it is unclear from the results of this study as to whetherthe significantly increased DSI scores for the trained subjectsare specifically the result of training, consistent participationin singing/choral activities, innately increased vocal capabil-ities in these subjects, or some combination of these factors,the difference in the mean DSI scores (6.48 for trained subjectsvs 4.00 for untrained) is substantial and primarily related to in-creased frequency and dynamic range.

In light of previous reports of increased vital capacity intrained singers, it was somewhat surprising that there was nosignificant difference between the trained singers and the un-trained subjects on measures of MPT. However, this findinghas some similarity to the observations of Sulter and Meijer,21

in which MPT was observed to decrease post-vocal training.These authors attributed a decrease in phonation time to a de-crease in glottal resistance and a more relaxed vocal fold pos-ture. It is also speculated that the trained subjects may nothave performed to their maximum potential on this particulartask, despite prompting during the elicitation of all DSI compo-nent measures, because they were hesitant to ‘‘stress’’ their vo-cal mechanism. In addition, it may have seemed unnatural forthe trained subjects to continue sustained phonation even asbreath support became substantially reduced.

Several studies have indicated that there may be increasedfrequency ranges and high frequency production in trainedsingers.26–28 In addition, in studies by Mendes et al38 and LeB-orgne and Weinrich,39 significant increases in frequency rangewhen compared before and after vocal training are reported. AsWestern singing training generally includes exercises and/or

warm-ups that may involve pitch matching, pitch glides, andpitch (musical) scales, it is of no surprise that trained subjectsmay be able to produce greater frequency ranges and maximumfrequencies than untrained subjects.38 The observation of sig-nificantly higher maximum frequency (F0 high) production infemales versus males is expected, and consistent with previousfindings by Wuyts et al11 and Hakkesteegt et al.16 Even thoughfemales characteristically produce higher maximum frequencyproductions as compared with males, this difference isgenerally offset by greater MPTs in males than females. Theresult is that the DSI scores for the genders tend not to differsignificantly.11

The trained singers were observed to have significantly lowerminimum intensity (I low) productions than untrained subjects.This finding is consistent with literature that has indicated in-creased phonatory control and a lowering of the minimum in-tensity capability for subjects who have undergone focusedvocal training.21,27,39 Vocal-training exercises that promote in-creased vocal fold thickness and increase breath support andcontrol tend to be conducive to lower minimum intensity pro-ductions.21 In addition, Western singing training often includesvarious respiratory exercises/warm-ups that focus on the properuse and strengthening of the abdominal muscles and dia-phragm,38 as well as a focus on optimal posture and respiratorysupport during both singing and speech tasks, which may en-able those with vocal training to produce greater ranges of vocalintensities.

The results of this study showed that trained subjects had sig-nificantly lower jitter than untrained subjects. Although a signif-icant difference in jitter between the trained and untrainedgroups was observed, both groups produced jitter values wellwithin normative expectations (generally expected to be sub-stantially less than 1%4,40). In addition, the normal jitter resultsfor both groups are consistent with the perception of normalvoice quality of all the subjects included in this study. It isknown that subjects can have normal voice quality and mea-sures of jitter, and yet differ in terms of the DSI. This resultclearly emphasizes that the DSI should be interpreted as a mea-sure of vocal function/performance, and is not necessarily a cor-relate of perceived or measured vocal quality.19

The DSI has the potential to be a useful diagnostic and clin-ical tool in regard to a variety of patients who may be experienc-ing voice difficulties. Although the DSI results of untrained

Journal of Voice, Vol. 24, No. 6, 2010666

vocalists reported in this study are consistent with the data ofuntrained vocalists reported in previous literature,11,16,19 the re-sults of this study indicate that alternative normative expecta-tions for the DSI may need to be taken into account whenusing the DSI with patients who have participated in directedvocal training or singing experience. The results of this studyclearly indicate that DSI scores for subjects with vocal/singingtraining may substantially exceed previously reported meanDSI values for normal untrained subjects. Although the currentliterature includes results showing changes in DSI for thosewho undergo vocal training, specific norms for subjects whohave continued their participation in vocal training or singingtasks over multiple years of experience are lacking. The resultsof this study provide preliminary DSI expectations for more ex-perienced vocalists. Future studies should extend these prelim-inary norms to include larger samples and alternative forms ofvocal experience beyond the choral/musical theater backgroundfor the subjects in this study.

Acknowledgment

The authors of this study would like to thank Dr. P. Milenkovicfor the download and use of the TF32 speech/voice analysisprogram.

REFERENCES1. Orlikoff RF, Dejonckere PH, Dembowski J, et al. The perceived role of

voice perception in clinical practice. Phonoscope. 1999;2:89–104.

2. Kreiman J, Gerratt B, Kempster GB, Erman A, Berke GS. Perceptual

evaluation of voice quality: review, tutorial, and a framework for future re-

search. J Speech Hear Res. 1993;36:21–40.

3. de Krom G. Consistency and reliability of voice quality ratings for different

types of speech fragments. J Speech Hear Res. 1994;37:985–1000.

4. Awan SN. The Voice Diagnostic Protocol: A Practical Guide to the

Diagnosis of Voice Disorders. Austin: Pro-Ed Inc.; 2001.

5. Michaelis D, Gramss T, Strube HW. Glottal-to-noise excitation ratio: a new

measure for describing pathological voices. Acustica/Acta Acustica.

1997;83:700–706.

6. Callan DE, Kent RD, Roy N, Tasko SM. Self-organizing maps for the

classification of normal and disordered female voices. J Speech Lang

Hear Res. 1999;42:355–366.

7. Frolich M, Michaelis D, Strube HW, Kruse E. Acoustic voice analysis by

means of the hoarseness diagram. J Speech Lang Hear Res. 2000;43:

706–720.

8. Awan SN, Roy N. Acoustic prediction of voice type in adult females with

functional dysphonia. J Voice. 2005;19:268–282.

9. Awan SN, Roy N. Toward the development of an objective index of dyspho-

nia severity: a four-factor model. Clin Linguist Phonet. 2006;20:35–49.

10. Awan SN, Roy N. Outcomes measurement in voice disorders: application

of an acoustic index of dysphonia severity. J Speech Lang Hear Res.

2009;52:482–499.

11. Wuyts FL, De Bodt MS, Molenberghs G, et al. The Dysphonia Severity

Index: an objective measure of vocal quality based on a multiparameter

approach. J Speech Lang Hear Res. 2000;43:796–809.

12. Awan SN. Instrumental analysis of phonation and voice. In: Ball M,

Perkins M, Muller N, Howard S, eds. Handbook of Clinical Linguistics.

Oxford, UK: Blackwell Publishing; 2008:344–359.

13. Hirano M. Clinical Examination of Voice. New York: Springer Verlag;

1981.

14. Timmermans B, De Bodt M, Wuyts F, Van de Heyning P. Voice quality

change in future professional voice users after 9 months of voice training.

Eur Arch Otorhinolaryngol. 2004;261:1–5.

15. Timmermans B, De Bodt MS, Wuyts FL, Van de Heyning P. Analysis and

evaluation of a voice-training program in future professional voice users.

J Voice. 2005;19:202–210.

16. Hakkesteegt MM, Brocaar MP, Wieringa MH, Feenstra L. Influence of age

and gender on the dysphonia severity index: a study of normative values.

Folia Phoniatr Logopaed. 2006;58:264–273.

17. Woisard V, Bodin S, Yardeni E, Peuch M. The Voice Handicap Index: cor-

relation between subjective patient response and quantitative assessment of

voice. J Voice. 2007;21:623–631.

18. Hakkesteegt MM, Wieringa MH, Brocaar MP, Mulder PGH, Feenstra L.

The interobserver and test-retest variability of the Dysphonia Severity

Index. Folia Phoniatr. 2008;60:86–90.

19. Hakkesteegt MM, Brocaar MP, Wieringa MH, Feenstra L. The relationship

between perceptual evaluation and objective multiparametric evaluation in

dysphonia severity. J Voice. 2008;22:138–145.

20. Hakkesteegt MM, Brocaar MP, Wieringa MH. The applicability of the Dys-

phonia Severity Index and the Voice Handicap Index in evaluating effects of

voice therapy and phonosurgery. J Voice. 2010;24:199–205.

21. Sulter AM, Meijer JM. Effects of voice training on phonetograms and max-

imum phonation times in female speech therapy students. In: Sulter AM, ed.

Variations of Voice Quality Features and Aspects of Voice Training in Males

and Females. Groningen: Rijkuniversiteit Groningen; 1996:133–147.

22. Large J. Observations on the vocal capacity of singers. NATS Bull. 1971;28:

34–35.

23. Gould W. The effect of voice training on lung volumes in singers and the

possible relationship to the damping factor of Pressman. J Res Sing.

1977;1:3–15.

24. Hixon T, Hoffman C. Chest wall shape in singing. In: Lawrence V, ed. Tran-

scripts of the Seventh Symposium Care of the Professional Voice, Part I:

Scientific Papers. New York: The Voice Foundation; 1979:9–10.

25. Watson B, Hixon T. Respiratory kinematics in classical (opera) singing.

J Speech Hear Res. 1985;28:104–122.

26. Troup G. The physics of the singing voice. J Res Sing. 1982;6:1–26.

27. Awan SN. Phonetographic profiles and frequency-SPL characteristics of

untrained versus trained vocal groups. J Voice. 1991;5:41–50.

28. Titze I. Principles of the Singing Voice. Englewood Cliffs: Prentice Hall;

1994.

29. Sulter AM, Schutte HK, Miller DG. Differences in phonetogram features

between male and female subjects with and without vocal training. J Voice.

1995;9:363–377.

30. Murbe D, Sundberg J, Iwarsson J, Pabst F. Phonetogram measurements of

singers. Before and after solo singer education. TMH-QPSR. 1996;37:

49–56.

31. Sulter AM, Peters HF. Perceptive characteristics of speech of untrained and

trained subjects, and influences of gender and age. In: Sulter AM, ed. Var-

iations of Voice Quality Features and Aspects of Voice Training in Males

and Females. Groningen: Rijkuniversiteit Groningen; 1996:73–105.

32. McCrae CR, Morris RJ. Voice onset time for female trained and untrained

singers during speech and singing. J Commun Disord. 2007;40:418–431.

33. Sound Forge v. 4.0. Madison, WI: Sonic Foundry Inc.; 1998 [Computer

Software].

34. TF32. Madison, WI: Milenkovic, P.; 2002 [Computer Software].

35. Aerophone II Model 6800. Lincoln Park: Kay Elemetrics Corp.; 1995

[Computer Software].

36. Asplund A. How loud was it? A calibration system for voice recording in

clinical and research applications. Available at: http://www.umu.se/

clin_sci/logopedi/personal/AndersA0902.pdf (Accessed June 2006).

37. SPSS 15.0 for Windows. Chicago: SPSS, Inc; 2006 [Computer Program].

38. Mendes A, Rothman H, Sapienza C, Brown WS. Acoustic effects of vocal

training on the singing voice. J Voice. 2003;17:529–543.

39. LeBorgne WD, Weinrich BD. Phonetogram changes for trained singers

over a nine-month period of vocal training. J Voice. 2002;16:37–43.

40. Dworkin JP, Meleca RJ. Vocal Pathologies: Diagnosis, Treatment, and

Case Studies. San Diego: Singular Publishing Group, Inc.; 1997.