Singers’ and Nonsingers’ Perception of Vocal Vibrato A. Anita Reddy and Uma Subramanian, Bangalore, Karnataka, India Summary: Vibrato, a small, nevertheless an important component in the singing voice is known to enrich the overall singing voice quality. However, in the perception of overall performance, it is often neglected. Singing performance is often appreciated by a mixed audience of those who love music, but not necessarily sing and other singers who may or may not be teachers of singing. Objectives: The objectives of the present study were aimed at investigating singers’ and nonsingers’ perception of vocal vibrato and its effect on the ratings of singer’s overall performance. Method: Prerecorded audio samples of the chorus of a hymn (How Great Thou Art) as sung by 10 singers (both men and women) were played via a speaker to two groups of judges which consisted of three experienced singers and three experienced nonsingers. The singer judges (SJs) were vocal instructors in Western classical, music theater, pop, and contemporary styles. Seven parameters (presence of vibrato, rate, extent, conspicuousness, quality, periodicity, and type) related to vibrato were evaluated through auditory perception by these two groups of judges on a rating scale developed specifically for the study, and one parameter evaluated singer’s overall performance. Results: Cohen’s Kappa statistical analysis was used for inter-rater reliability within groups. Nonsinger judges (NSJs) within the group showed varied ratings as did SJs, yet SJs did have higher agreement than NSJs. Chi-square analysis was used across groups. Both groups were distinct from each other in their perception of vibrato. Ratings of singer’s overall performance were not affected for NSJs, but certainly affected for SJ. Conclusion: It could not be concluded that ratings on singer’s overall performance was affected as a result of vibrato. Since vibrato is often over-ridden by the singer’s voice. But a rare occasion can arise where a vibrato may not sound pleasant and can affect the listener’s perception of the singer’s performance. Often a feedback from listeners would help monitor singers’ performance. Key Words: Auditory–Perception–Vibrato–Singers and nonsingers–Singer’s overall performance. INTRODUCTION ‘‘Perception is the ability to identify, interpret, and attach meaning to sound’’ (McGraw Hill concise dictionary), and this could be through visual, auditory, or tactual mode. Sound perception is predominantly through auditory mode. There are certain characteristics that aid in perception of sound, such as loudness, pitch, quality, and the duration of sound and how it changes.13 The phenomenon vibrato ‘‘contributes to perception of pitch, intensity, and timbre of the vocal sound’’15; (pg. 117). Vibrato beautifies music and is perceived in both instrumental and vocal music. This phenomenon is considered a characteristic feature of Western operatic singing and is also quite apparent in other genres of music, such as pop, rock, gospel, country, Jazz, Indian-Hindustani music, and so on. Singers believe that vibrato adds an esthetic quality to the singing voice and ‘‘every great voice of all time has had a vibrato.’’20 It is used for expressive purposes15 to bring out emotion as it is said to ‘‘reflect genuine feeling by the singer.’’7 Vibrato has been described by Seashore as early as 196714 as ‘‘a pulsation of pitch, usually accompanied with synchronous pulsations of loudness and timbre, of such extent and rate as to give a pleasing flexibility, tenderness and richness to the tone.’’ Similarly, Zemlin21 describes these oscillaAccepted for publication September 16, 2014. From the Department of Speech and Language Studies, Dr.S.R.Chandrasekhar Institute of Speech and Hearing, Bangalore, Karnataka, India. Address correspondence and reprint requests to A. Anita Reddy, Department of Speech and Language Studies, Dr.S.R.Chandrasekhar Institute of Speech and Hearing, Hennur Main Road, Lingarajapuram, Bangalore 560084, Karnataka, India. E-mail: anita.reddy. [email protected] Journal of Voice, Vol. 29, No. 5, pp. 603-610 0892-1997/$36.00 Ó 2015 The Voice Foundation http://dx.doi.org/10.1016/j.jvoice.2014.09.022
tions as ‘‘small and rapid pitch and intensity changes that occur primarily during singing.’’ Stark16 mentioned two aspects for the perception of vibrato, one is its identification and the second is ‘‘distinguishing between different kinds of vibrato.’’ Rothman and Arroyo suggested that the parameters vital for the perception of vibrato are ‘‘the frequency and amplitude variations around their respective means’’ (cited by Stark16) They could be either frequency or amplitude variations or both. Acoustically, vocal vibrato has been described by Sundberg18 in four parameters as rate, extent, regularity, and waveform. Ideally, these pulsations for a good singer (on an average) should be between 5 and 8 Hz/second, and pulsations slower than this would be perceived by the listener as unpleasant.4 Other researchers found vibrato rate to be between 5.8 Hz and 7.3 Hz12 or 4.5 Hz and 7.2 Hz.11 The extent of vibrato is acceptable to the human ear if it is within 0.5 to 1.0 semitone or even up to 2 semitones.17 However, the terms ‘‘rate’’ and ‘‘extent’’ have been widely accepted and are used by vocal trainers to define vibrato. If the vibrato is too fast, it is called a ‘‘bleat.’’9 If it is too slow or exaggerated, it is called a ‘‘wobble.’’1,9,21 Available literature on past research discusses acoustic parameters of vibrato such as rate (how rapid the modulations take place), extent (how large or small the modulations are),6,11,12 and periodicity.6 Some researchers2,7,8 have attempted at comparing the perceived attributes of vibrato with acoustic measures. Howes et al7 chose to study parameters such as vibrato onset, rate, and extent and other parameters involved with the obtrusiveness of vibrato. The first three parameters were measured on a bipolar visual analog scale. Vibrato onset was measured between ‘‘delayed’’ and ‘‘‘‘immediate,’’ vibrato rate showed ‘‘too slow’’ on the left hand and
604 ‘‘too fast’’ on the right hand and vibrato extent showed ‘‘too wide’’ and ‘‘too narrow’’ on the left and right hand sides, respectively. The judges who evaluated vibrato were 24 qualified teachers of singing. Their findings showed that, although they were similar in their perception in certain parameters, two of the judges found it difficult to assess vibrato, and some of them found it difficult to separate vibrato assessment from overall voice quality. Listeners develop their perceptual abilities without effort and often underestimate their ability to discriminate between different instruments and between different performers playing the same instrument.10 Lopez reported that inexperienced listeners were able to recognise trained singers by their speaking voices (cited by Brown et al3). Kenny and Mitchell8 stated that ‘‘expert listeners’’ can be ‘‘reliable in their judgments of overall quality’’ (pg.55). There is a ‘‘prevailing confusion regarding vibrato that it is never heard as it really is in the actual tone. It is modified by a number of illusions which result in our hearing something entirely different from that which is performed’’ (Seashore, cited by Stark16; pg.147). ‘‘The ability to perceive vibrato differs from one individual to another.’’16 He also stated that some listeners cannot discern it at all. Even if they did, ‘‘Repulsiveness of the bad vibrato has put many a musician on edge against all vibratos . Thus, the good vibrato is ignored through the offense of the bad,’’ (Seashore cited by Stark16; pg.147). For the most part, research has been directed toward acoustic measurements and perceptual–acoustic correlations of vibrato in trained singers and lesser toward perceptual studies alone. Singing is predominantly a perceptual phenomenon as it is tailored for human perception.19 There have been fewer attempts at investigating the differences or similarities of how experienced singers and nonsingers would perceive the same event and in this case, vocal vibrato. Singers rely on the listener’s judgment of their singing. Because vibrato is essential to the singer, how do nonsingers perceive this feature? Would vibrato affect the listener’s perception of the singers’ overall performance? METHOD The purpose of this study was to investigate singers’ perception of vibrato, nonsingers’ perception of vibrato, and if singers’ and nonsingers’ perception of vibrato would affect their ratings of ‘‘singers’ overall performance.’’ In the present study, we hypothesized that there would be no difference between singers’ and nonsingers’ judgments of vocal vibrato. Second, nonsingers’ ratings of singer’s overall performance would not be affected by their perception of vibrato as much as it would affect singers’ ratings on the singer’s overall performance. We reasoned that nonsingers’ exposure to a variety of musical styles may cause them to accommodate different types of vibrato. Materials Definition list (Appendices), recordings, rating scale (Appendices), desktop computer, and a speaker belonging to Computerized Speech Lab software (model 4500; Kay Pentax, Lincoln Park, NJ).
Journal of Voice, Vol. 29, No. 5, 2015
Recordings Ten prerecorded audio samples (recorded on Praat software; Praat, version 5.3.05; developed by Paul Boersma & David Weenink) of professional singers (five male and five female) who had sung the chorus of a hymn (How Great Thou Art) were used for the present study. ‘‘Then sings my soul. My Saviour God to thee. How great Thou art. How Great Thou art. (repeat entire chorus)’’
The professional singers had performed on stage for a minimum of 5 years and had sung contemporary music. Of the 10 professional singers, five (three female and two male) of them were trained and five (two female and three male) were untrained. The untrained singers did not have any formal training in the past and they are well known in the city of Bangalore. The trained singers had 3–5 years of vocal training at the time of recording. They were trained in Western classical music and contemporary style and were still pursuing higher levels in vocals. All singers at the time of recording had been instructed to sing in their comfortable pitch and loudness levels to ascertain a natural production. The men’s voices ranged from baritone to tenor. The female voices ranged from contralto to soprano. Participants The participants who were willing to serve as judges were three professional singers (S1, S2, and S3) and three nonsingers (NS1, NS2, and NS3). The singer judges (SJ) were vocal pedagogs themselves who had 20 years experience in singing and singing instruction of Western classical, Music Theater, and contemporary styles (pop, gospel, and jazz) of music. The nonsinger judges (NSJs) were those who enjoyed listening to Western music on a daily basis over a long period (approximately 20 years) of time. The participant judges were divided into two groups: one singer group and one nonsinger group. Each group consisted of three judges. RATING SCALE Definitions Presence of vibrato was evaluated between present or absent (never). Rate (speed of oscillations that were perceived in the singing voice) was evaluated on ‘‘too slow,’’ ‘‘just right,’’ and ‘‘too fast.’’ Extent (how large or small the oscillations6 are perceived in the singing voice) was evaluated on ‘‘too wide,’’ ‘‘just right,’’ ‘‘too narrow.’’ Conspicuousness of vibrato (if it drew undesired attention to itself on perception) rated on ‘‘too much,’’ ‘‘just right,’’ ‘‘too less.’’ Quality was rated on the ‘‘pleasantness’’ of vibrato on perception. Periodicity referred to the stability of vibrato when it was produced.6 Type of vibrato was judged between ‘‘free’’ or ‘‘forced,’’ where free refers to an effortless production and forced refers to a manual attempt (either throat or diaphragmatic region) to bring the effect of vibrato during singing. Finally, singer’s overall performance
A. Anita Reddy and Uma Subramanian
605
Perception of Vocal Vibrato
referred to how each singer’s performance was perceived by the singer group and nonsinger group (Appendix II). The rating scale was specifically developed for the purpose of this study. It was given to three singing teachers for their opinion on the parameters of vibrato. Once the approval of the singing teachers was taken on the parameters used in the rating scale, it was used for the present study. Parameters such as presence, quality, and periodicity were measured on a five-point rating scale were rated on always, frequently, sometimes, rarely, never, where ‘‘always’’ represented 100% of the time, ‘‘frequently’’75%, ‘‘sometimes’’–50%, ‘‘rarely’’–25%, and ‘‘never’’ represented 0% of the time. Singer’s overall performance was also rated on a five-point rating scale as very good, good, fair, poor, and very poor. Parameters such as rate, extent, and conspicuousness were rated on a three-point rating scale. Type of vibrato was rated on a two-point rating scale as free or forced (Appendix I). Administration procedures The 10 prerecorded audio samples of the chorus of a hymn (How great Thou Art) were randomly arranged and played via a speaker (JBL Professional, China) belonging to CSL Model 4500 in a quiet room. There was no orchestral accompaniment in the background, thereby allowing a clear perception of vibrato. Each singer had sung the entire chorus of the song. The judges were asked to listen to the samples and evaluate each sung sample using the rating scale provided. The group of NSJs was given an explanation of vocal vibrato. Both groups were given a rating scale consisting of seven parameters of vocal vibrato and one parameter related to the singer’s overall performance. They were given an explanation of each parameter that had to be evaluated. All participants were na€ıve to the experiment and had not evaluated vibrato (in particular) earlier. The parameters considered for evaluation were presence, rate, extent, conspicuousness, quality, periodicity, type of vibrato, and singers’ overall performance. However, the definition list was not given to the judges, but each judge was given the same explanation of the parameters that needed clarification. This was carried out to maintain uniformity in their understanding of the requirement of each parameter. Data analysis Once the SJs and NSJs had completed rating the samples, the ratings obtained on each parameter were entered on an Excel
sheet and subjected to statistical analysis. SPSS software version 15 (IBM, Armonk, NY) was used for statistical analysis.
RESULTS The present study investigated the perception of vocal vibrato by SJs and NSJs; also if the SJs and NSJs’ perception of vibrato would affect their ratings on singers’ overall performance. Cohen’s kappa statistics was computed using SPSS software (version 15) for inter-rater reliability within each group, and chi-square tests were computed for measuring agreement across groups (nonsinger vs singer).
Inter-rater reliability of NSJs The ratings of each judge were compared against each other as NS1-NS2; NS1-NS3; and NS2-NS3. All eight parameters of the rating scale showed varied ratings among all three NSJs. On the parameter, ‘‘presence’’ of vibrato, NS1 rated the 10 singers’ samples across four points (always, frequently, sometimes, and rarely) on the rating scale, NS2 rated across three points (always, frequently, and sometimes), and NS3 rated across two points (always and frequently). Because different categories on the rating scale were chosen, the agreement statistics could not be computed between any two judges. The rate of vibrato was rated on three points, such as ‘‘too slow,’’ ‘‘just right,’’ and ‘‘too fast.’’ Ratings varied between judges, yet, among the few singers’ samples that were agreed upon, most of the ratings were given for a ‘‘just right’’ vibrato rate and the other ratings were distributed between too slow and too fast. Kappa values were 0.123 (slight agreement) and 0.310 (fair agreement) and 0.206 (slight agreement) for NS1NS2, NS1-NS3, and NS2-NS3, respectively (Table 1). Therefore, the kappa values obtained for rate of vibrato showed only slight agreement between either group of judges. The extent of vibrato was judged on three parameters such as ‘‘too wide,’’ ‘‘just right’’ and ‘‘too narrow.’’ NS1-NS2 rated the extent of vibrato as ‘‘just right’’ on nine singers’ samples and one as ‘‘too narrow’’ thus having 100% agreement (kappa ¼ 1.000). Their ratings for other parameters were scattered along the length of the rating scale. However, NS1-NS3 and NS2-NS3 agreed on six singers’ samples, where five were rated as ‘‘just right’’ and one was rated as ‘‘too narrow.’’
TABLE 1. Inter-Rater Agreement Between Nonsinger Judges (NS1, NS2, and NS3) on the Number of Singers for Each Parameter Parameter Presence Rate Extent Conspicuousness Quality Periodicity Type Singer’s overall performance
NS1-NS2
NS1-NS3
NS2-NS3
5 5 (K ¼ 0.123) 10 (K ¼ 1.000) 7 2 2 5 4
2 6 (K ¼ 0.310) 6 3 4 3 5 (K ¼ 0.194) 6
3 5 (K ¼ 0.206) 6 2 1 4 7 4
606 Because the categories selected for rating varied, there was no significant agreement among them. Conspicuousness of vibrato was rated on ‘‘too much,’’ ‘‘just right,’’ and ‘‘too less.’’ NS1 and NS2 agreed on seven singers’ samples, where two were perceived as ‘‘too less’’ and five were perceived as ‘‘just right,’’ whereas NS1 and NS3 agreed only on three singers’ samples as being ‘‘just right.’’ NS2 and NS3 agreed only on two samples, where one sample was perceived as having a vibrato that was ‘‘too much’’ vibrato and another sample of vibrato was perceived as ‘‘just right.’’ Rankings differed among them; hence, agreement statistics could not be computed. Quality of vibrato. This parameter was rated on a five-point rating scale on how pleasant the vibrato sounded for each singer (always [100% of the time], frequently [75%], sometimes [50%], rarely [25%], and never [0% of the time]). With reference to quality of vibrato, NS1 and NS2 agreed only on two singers’ samples, and hence, there was a poor agreement among them. NS1 and NS3 agreed on four samples of which one singer’s sample was rated as having a pleasant vibrato ‘‘always’’ (100% of the time), another as having a pleasant vibrato ‘‘frequently’’ (75% of the time) and the rest as having a pleasant vibrato only ‘‘sometimes’’ (50% of the time). NS2 and NS3 agreed on only one case as having a pleasant vibrato ‘‘sometimes’’ (50% of the time). Because judges varied in their rankings, agreement statistics could not be computed. Periodicity (stability) of vibrato. The periodicity of vibrato was rated on whether vibrato was stable always (100%), frequently (75%), sometimes (50%), rarely (25%), or never (0% of the time), while singing. NS1 and NS2 agreed on only two singers’ samples. NS1 and NS3 agreed on the rating of periodicity on three samples, where two of the singers were rated as having a periodic vibrato ‘‘always’’ (100% of the time) and one was ‘‘rarely’’ periodic or stable (25% of the time). NS2 and NS3 agreed on four samples where three of them were rated as having a stable vibrato only ‘‘sometimes’’ (50% of the time) while singing. Because the rankings largely varied among the judges, agreement statistics could not be computed. Type of vibrato. The type of vibrato was rated across two points between ‘‘free’’ and ‘‘forced.’’ NS1 and NS2 agreed on five singers’ samples, of which three were rated as ‘‘forced’’ and two were rated as ‘‘free.’’ NS2 rated one sample as ‘‘cannot say.’’ Because their rankings varied, no significant agreement was found among them. NS1 and NS3 agreed on five samples, of which three were rated as ‘‘forced’’ and two were rated as ‘‘free.’’ Because the categories selected for rating were similar for the five samples, they had a slight agreement (kappa ¼ 0.194). Although NS2 and NS3 agreed on seven samples (70%), of which two were rated as ‘‘forced’’ and five were rated as ‘‘free,’’ (as mentioned earlier, NS2 had rated one sample as ‘‘cannot say’’) no significant agreement was found among them. Overall performance. Each singer’s overall performance was rated on five points, such as very good, good, fair, poor, and very poor. NS1 and NS2 agreed only on four singers (40%) as having a good performance. NS1 and NS3 agreed on the ratings of overall performance on six singers (60%), of
Journal of Voice, Vol. 29, No. 5, 2015
whom two singers were rated as fair, three as good, and one as very good; however, there was no significant agreement among them. NS2 and NS3 agreed on a good performance for four singers (40%). Because they varied in their rankings, agreement statistics could not be computed. Parameters such as ‘‘presence,’’ ‘‘conspicuousness,’’ ‘‘quality,’’ ‘‘periodicity,’’ ‘‘type,’’ and ‘‘singer’s overall performance’’ ratings showed poor agreement between all three NSJs (kappa values that were obtained wherever possible are shown in Table 1). Inter-rater reliability of SJs Three experienced singers (S1, S2, and S3) participated as judges. They also had to rate singers’ vibrato on the same parameters. The singer judges were grouped as S1-S2, S1-S3, and S2-S3. Although all the three singer judges agreed on vibrato being present in all 10 singers yet, they varied on the five categories (always, frequently, sometimes, rarely, and never) selected for rating. Hence, agreement statistics on presence of vibrato could not be computed (between any two groups of SJs). Rate of vibrato showed only slight agreement (kappa value ¼ 0.231) between S1 and S3. These judges (S1 and S3) agreed that of 10 samples, two samples had a vibrato that was ‘‘too slow’’ and three that were ‘‘just right.’’ However, only S3 perceived one sample to have a fast vibrato between the two SJs. Although the other two groups of SJs (S1-S2 and S2-S3) agreed on seven samples (ie, S1-S2) and three samples (S2-S3), respectively, the categories (too slow, just right, and too fast) selected for rating varied; hence, agreement statistics could not be computed (Table 2). For the extent of vibrato, S1 and S3 perceived two samples as ‘‘too wide,’’ three as ‘‘just right’’ and two as ‘‘too narrow,’’ leading to a significant agreement between them on a total of seven samples (kappa value ¼ 0.545/moderate agreement; approximate significance ¼ 0.010). The other two groups of SJs (S1-S2 and S2-S3) did not show adequate agreement between their respective groups. Hence, kappa scores could not be obtained (Table 2). Only S1 and S3 showed moderate agreement (seven samples; kappa value ¼ 0.455) on conspicuousness of vibrato (Table 2). Here they perceived one singer to have a very conspicuous (too much) vibrato, five singers whose vibrato was ‘‘just right’’ and one singer whose vibrato was ‘‘too less.’’ The other two groups of judges varied in their rankings, and hence, agreement statistics could not be computed. With reference to the quality of vibrato, there was moderate agreement (kappa value ¼ 0.474) on seven samples between S2 and S3. They perceived two samples to have a pleasant vibrato 75% (frequently pleasant) of the time, four samples to have a pleasant vibrato 50% (sometimes pleasant) of the time and one sample with a pleasant vibrato 25% (rarely pleasant) of the time. The other two groups of SJs (S1-S3 and S1-S2) showed poor agreement; hence, agreement statistics could not be computed (Table 2). There was substantial agreement (kappa ¼ 0.655) between S2 and S3 over periodicity of vibrato on eight samples. They agreed that one sample had a stable vibrato 100% of the time, three samples had a stable vibrato 75% of the time, and four samples had a
A. Anita Reddy and Uma Subramanian
607
Perception of Vocal Vibrato
TABLE 2. Inter-Rater Agreement Between Singer Judges (S1, S2, and S3) on the Number of Singers Parameter Presence Rate Extent Conspicuousness Quality Periodicity Type Singer’s overall performance
S1 and S2
S1 and S3
S2 and S3
7 7 5 6 3 3 7 6 (K ¼ 0.385)
6 5 (K ¼ 0.231) 7 (K ¼ 0.545) 7 (K ¼ 0.455) 6 3 6 3
6 3 8 9 7 (K ¼ 0.474) 8 (K ¼ 0.655) 7 4
stable vibrato 50% of the time (Table 2). Again, the other two groups of judges (S1-S3 and S1-S2) varied in the categories selected for rating, and hence, kappa scores could not be obtained. Type of vibrato showed ratings on equal number of samples (seven singers) between S1-S2 and S2-S3 (Table 2) nevertheless, S1 mentioned that she could not tell if one of the singers’ vibrato was free or forced and S3 mentioned that one singer’s vibrato was ‘‘in between’’ free and forced (S1 rated sample ‘‘H’’ as ‘‘cant say’’ and S2 rated sample ‘‘G’’ as ‘‘in between’’). Because they did not agree on similar categories, agreement statistics could not be computed. On singer’s overall performance, there was fair agreement (kappa ¼ 0.385) between S1 and S2, where they perceived two singers to have a ‘‘fair’’ performance, three singers as ‘‘good,’’ performance and one singer as ‘‘very good’’ performance. The other groups of SJs did not agree on similar categories, and hence, agreement statistics could not be computed (Table 2). In summary, judges S1 and S3 had higher agreement than the other groups of SJs. Although S1 and S3 showed agreement on lesser number of singers for parameters such as rate, extent, and conspicuousness, yet they seemed to agree on similar categories on each rating scale. The kappa agreement varied between fair (K ¼ 0.231) to moderate agreement (K ¼ 0.545) between them (Table 2). Although S2-S3 showed agreement only on two parameters (quality and periodicity), it ranged from moderate to substantial agreement. Comparison of ratings between NSJs and SJs A statistical analysis of singers and nonsingers’ ratings on all eight parameters was done using chi-square tests to examine if there were differences in the rankings between singers and nonsingers. Singers and nonsingers’ ratings on the presence of vibrato were compared. Both SJ and NSJ had similar ratings of vibrato occurring only 50% of the time. NSJs agreed on five samples and SJs agreed on six samples. Only one singer was perceived to have a vibrato 100% of the time (always) during singing by two SJs and one NSJ. For this reason, there was no significant difference in their ratings (Table 3). Rate of vibrato was perceived as ‘‘just right’’ by NSJs who showed a slightly higher percentage (50%) of rating than SJs (44.4%). The rest of the ratings were distributed between ‘‘too fast’’ and ‘‘too slow.’’ Therefore, both NSJs and SJs did not show a significant difference in their ratings. Extent of
vibrato was rated on ‘‘too wide,’’ ‘‘just right,’’ and ‘‘too narrow.’’ SJs agreed on four samples (40%) whose vibrato extent was perceived as ‘‘just right’’ and four other samples (40%) as ‘‘too narrow.’’ Nonsingers on the other hand, agreed on nine samples (90%) whose vibrato extent was perceived as ‘‘just right’’ and on one sample as ‘‘too narrow.’’ Thus, there was a significantly higher percentage of rating among NSJs compared with SJs. Although the ratings did show a higher percentage among NSJs, it was not statistically significant (Table 3). Conspicuousness of vibrato was rated on ‘‘too much,’’ ‘‘just right,’’ and ‘‘too less.’’ SJs had higher ratings on ‘‘just right’’ (eight samples; 80%), ‘‘too much’’ (one sample; 10%), and ‘‘too less’’ (one sample; 10%) than NSJs, whose ratings were distributed between ‘‘just right’’ (four samples; 40%) and ‘‘too less’’ (five samples; 50%) and ‘‘too much’’ vibrato (one sample; 10%). Thus, there was a slightly higher percentage of rating among SJs on the conspicuousness of vibrato being just right. Yet, there was no significant difference between both groups (Table 3). Quality of vibrato had equal ratings by both SJs and NSJs, and they seemed to agree that vibrato was pleasant only 50% (sometimes) of the time in five samples and 75% of the time in four other samples. There was a difference in their rating only by one sample, where NSJs rated one sample as ‘‘always’’ pleasant (100% of the time) and SJs rated one sample as ‘‘rarely’’ pleasant. Thus, similar ratings were found between SJs and NSJs. Hence the chi-square analysis did not show a significant difference (Table 3). With reference to periodicity of vibrato, SJs had rated six samples as sometimes stable, three as frequently stable, and only one as always stable. But NSJs rated only three as TABLE 3. Comparison of ratings between NSJs and SJs Parameter Presence Rate Extent Conspicuousness Quality (pleasantness) Periodicity (stability) Type Singer’s overall performance
Singer (%) Nonsinger (%) 60 44.4 40 80 50 60 90 44.5
50 50 90 50 50 30 55.6 70
608 ‘‘sometimes’’ stable, two as ‘‘frequently’’ stable, four as ‘‘always’’ stable, and one as ‘‘rarely’’ stable. There was no significant difference among the two groups of judges (Table 3). SJs rated nine samples (90%), as having a ‘‘free’’ vibrato and only one sample as ‘‘forced’’ vibrato. So, there was significantly higher percentage of rating (90%) among SJs on a ‘‘free’’ vibrato than NSJs, whose ratings were distributed between ‘‘free vibrato’’ (five samples; 55.6%) and ‘‘forced’’ vibrato (44.4%). One sample was judged by NS2 as ‘‘can’t say.’’ Although SJs had a higher percentage of rating, there was no significant difference between both groups (Table 3). Overall performance (of singing) of each singer was rated on a five-point rating scale (very good, good, fair, poor, and very poor). NSJs rated 70% of singer’s overall performance as being ‘‘good,’’ 20% as ‘‘fair’’ and 10% as ‘‘very good.’’ On the other hand, SJs agreed on 44.5% as ‘‘good’’ performance and 44.4% as ‘‘fair’’ performance and 11.1% as ‘‘very good’’ performance. Thus, NSJs had higher ratings than SJs but showed no statistical significance (Table 3). DISCUSSION This study aimed at investigating how singers’ and nonsingers’ perceived vibrato and if perception of vocal vibrato would affect their ratings of singer’s overall performance. Rating vocal vibrato was challenging for both singers and nonsingers. Because the judges had to focus only on vibrato while listening to the entire music excerpt, they found it intricate to rate vibrato on each of the eight parameters. It was interesting to note that NSJs perceived most singers to have an appropriate rate and extent for their vibrato, but SJs did not perceive it the same way, the ratings seemed to be divided between appropriate and either too slow or fast. It seemed that SJs perceived vibrato in a more detailed manner for almost all parameters. Rate and extent of vibrato did not seem to bother the NSJs as much as it did for SJs, and this may be attributed to the nonsingers having exposure to not just different genres of music besides classical and pop music but also to singers who may not have had formal vocal training and yet have begun their singing career. This is where they may have heard vibratos ranging from a narrow to a wide extent. It also may be possible that trained singers get accustomed to hearing a certain rate and extent of vibrato and anything above or below a certain range would be considered for correction. If the rate and extent of vibrato was at either extreme (too fast or too slow and too narrow or too wide, respectively) it would still sound unpleasant (which is mostly seen during the early vocal training period). Vocal pedagogs who participated in this study mentioned that vibrato is never taught specifically but the student may be instructed to correct it if it drew attention to itself. By and large, vocal pedagogs allow the fluctuations of pitch in the student’s voice and allow vibrato to gain its own speed (Van Christy, 1967). With reference to conspicuousness of vibrato, NSJs could not conclude at a particular category of rating but SJs were more consistent, where vibrato was perceived to be ‘‘just right’’ for most of the samples (Table 2). Singers who are vocal pedagogs
Journal of Voice, Vol. 29, No. 5, 2015
have practical knowledge of vibrato, whereas nonsingers have a more psychoacoustical outlook toward vibrato. The group of NSJs did mention that it was a difficult task as they had not perceived vibrato in such detail in the past as it is generally overridden by the performer’s voice and the orchestra. In general, no one focuses on listening to a vibrato rather than the song (unless it is done intentionally). The manner by which SJs perceived vibrato seemed to reflect how each of these parameters, such as presence of vibrato, conspicuousness, periodicity, and type of vibrato, were connected. The influence of other parameters on the quality of vibrato was evident. This was perceived by both SJs and NSJs. Perhaps, it may be assumed that ‘‘presence’’ is the foremost feature for any musical sound and ‘‘quality’’ is its final feature. Even if one of these parameters, such as rate, extent, conspicuousness, or periodicity, embedded in between these two features (presence and quality) was not appropriate, the final output (quality) would be affected. For example, if extent was too wide, it would still sound unpleasant, or if vibrato was not periodic/stable, it would still affect the quality. Diaz and Rothman5 studied the acoustic comparison between good and poor vibratos in singers and found that the good vibratos were the most periodic ones. Vibrato has also been reported to have an influence on overall quality of singing voice8 and its presence is what determines a ‘‘trained sound.’’22 With reference to type of vibrato, whether free or forced, a free vibrato would be the result of a free vocal tone which is again the result of ‘‘freedom and good technique.’’7 But a forced vibrato would often be produced by manipulating the throat (in an up and down manner) or diaphragmatic region (by pulsating it voluntarily) or through constricted vocal cords.1 One SJ expressed that she could not tell if vibrato was free or forced, for one of the samples. Another SJ also did mention that one sample of vibrato was ‘‘in between’’ free and forced (it was not the same sample). Nevertheless, SJs had a higher percentage of rating on a free vibrato rather than forced vibrato unlike NSJs. SJs did remark on the overall performance of singers and also on the vibrato. Remarks such as good vocal tone, controlled/ uncontrolled vibrato, too much pitch variation, inadequate breath support, good vocal agility, and weak/good vocal resonance had been written at the end of the rating scale, in the space below. SJs had distributed ratings between good and fair performance. NSJs also made similar remarks with regard to singers’ pitches, vibrato rate decreasing along with lowering of pitch, control of voice. Despite remarks of such kind, the singer’s overall performance ratings ranged from fair to very good performance by the NSJs. But when SJs’ ratings were compared with each other within their respective group, they had a wider range, ranging from poor to very good. Perhaps, it can be attributed to their (SJs) experience in teaching and training students on their performance that their expectations of overall performance would have been higher. Finally, it was observed that third NSJ (NS3) had similar ratings as SJs (S1 and S2). The varied ratings among NSJs may be justified for this reason.
A. Anita Reddy and Uma Subramanian
Perception of Vocal Vibrato
The present study brought forth three aspects: (1) SJs perceived vibrato on practical aspects such as presence, conspicuousness, periodicity, and type of vibrato, which is not surprising as they are more mindful of what kind and how much of vibrato is required. On the other hand, NSJs did not seem to have too much concern with regard to rate and extent of vibrato. (2) The perception of the quality of vibrato seemed to be similar for both SJs and NSJs. (3) Perceptual ratings on singer’s overall performance seemed to be affected for the SJs but it did not seem to affect the ratings of NSJs. However, it cannot be said that the ratings of singer’s overall performance were affected because of vibrato. It could be attributed to the tone, breath support, quality of voice, and resonance, based on how both groups of judges had mentioned as remarks. Furthermore, it can also be a possibility that the location of the parameter singer’s overall performance at the very end of the score sheet may have been a possible reason for the ratings on singer’s overall performance being affected (for SJs). CONCLUSION Limitations of the study Intra-rater reliability was not measured for both groups of judges as it would have given a better representation of each individual’s perception of vibrato. Perception differs from person to person; nevertheless, singers rely on their listeners’ feedback. Furthermore, this rating scale provides detailed features of vocal vibrato, which may be used by singing teachers to explain to their students where their vibrato may need correction. Because production of vibrato depends on good singing technique, this rating scale may suggest on improvement of singing techniques for the young student. REFERENCES 1. Bennet G. From Rock to Opera: The Basic Technique of Voice. 2nd ed. Milwaukee, WI: Hal Leonard Corporation; 2001:65.
609
2. de Almeida Bezerra A, Cukier-Blaj S, Duprat A, Camargo Z, Granato L. The characterization of the vibrato in Lyric and Sertanejo singing styles: acoustic and perceptual auditory aspects. J Voice. 2009;23:666–670. 3. Brown WS Jr, Rothman HB, Sapienza CM. Perceptual and acoustic study of professionally trained versus untrained voices. J Voice. 2000;14:301–309. 4. Bunch MD. Dynamics of Singing Voice. 5th ed. New York, NY: Springer Wien; 2009:108–109. 5. Diaz JA, Rothman HB. Acoustical comparison between samples of good and poor vibrato in singers. J Voice. 2003;17:179–184. 6. Dromey C, Carter N, Hopkin A. Vibrato rate adjustment. J Voice. 2003;17: 168–178. 7. Howes P, Callaghan J, Davis P, Kenny D, Thorpe W. The relationship between measured vibrato characteristics and perception in western operatic singing. J Voice. 2004;18:216–230. 8. Kenny DT, Mitchell HF. Acoustical and perceptual appraisal of vocal gestures in the female classical voice. J Voice. 2006;20:55–70. 9. McKinney CJ. The Diagnosis and Correction of Vocal Faults. Revised ed. Naashville, TN: Genevox Music Group; 1994. 10. Mitchell HF, MacDonald RAR. Non-singers as spectators? Audio visual integration improves music performer identification. Psychol Music. 2012;42:112–127. 11. Murbe D, Zahnert T, Kuhlisch E, Sundberg J. Effects of professional singing education on vocal vibrato- a longitudinal study. J Voice. 2007; 21:683–688. 12. Prame E. Measurement of vibrato rate of ten singers. J Acoust Soc Am. 1994;96:1979–1984. 13. Sanders AD. Auditory Perception of Speech: An Introduction to Principles and Problems. Englewood, NJ: Prentice Hall Inc; 1977:14. 14. Seashore CE. Psychology of Music. New York, NY: Dover Publications Inc; 1967:33. 15. Sell K. The Disciplines of Vocal Pedagogy: Towards an Holistic Approach. Aldershot, UK: Ashgate Publishing Co; 2005:117–118. 16. Stark AJ. Bel Canto: A History of Vocal Pedagogy. Toronto, Canada: University of Toronto Press; 1999:146–147. 17. Sundberg J. The Science of the Singing Voice. Dekalb, IL: Northern Illinois University Press; 1987:163–164. 18. Sundberg J. Acoustic and psychoacoustic aspects of vocal vibrato. In: Dejoncjeres PH, Hirano M, Sundberg J, eds. Vibrato. San Diego, CA: Singular publishing; 1995:39. 19. Sundberg J. The perception of singing. In: Deutsch D, ed. The Psychology of Music. 2nd ed. San Diego, CA: Academic Press Series in Cognition and Perception; 1999:171–214. 20. Van Christy A. Expressive Singing. Dubuque, IA: Wm.C. Brown Company Publishers; 1972:42–43. 21. Zemlin RW. Speech and Hearing Science: Anatomy and Physiology. 4th ed. Englewood Cliffs, NJ: Allyn and Bacon; 1998:170–171. 22. Sataloff RT. Professional Voice: The Art and Science of Clinical Care. 3rd Edn. Vol. 2. San Diego, CA: Plural Publishing; 2005.
610
Journal of Voice, Vol. 29, No. 5, 2015
APPENDIX I Score sheet Instructions to judges: parameters such as presence, quality, periodicity, and singer’s overall performance are rated on a
Parameter
Always
five-point rating scale, where always ¼ 100%; almost always ¼ 75%; sometimes ¼ 50%; rarely ¼ 25%; and never ¼ 0% of the time. Parameters such as rate, extent, and conspicuousness are rated on a three-point rating scale, and parameters such as type are rated on just two points. Kindly tick wherever appropriate.
Presence of vibrato Frequently Sometimes
Rarely
Never
Presence of vibrato Too slow
Rate of vibrato Just right
Too fast
Too wide
Extent of vibrato Just right
Too narrow
Too much
Conspicuousness (obvious) of vibrato Just right
Too less
Parameter
Always
Quality of vibrato Frequently Sometimes
Rarely
Never
Always
Periodicity (stability) of vibrato Frequently Sometimes
Rarely
Never
Pleasant Parameter Stable Type of vibrato Free/forced Singer’s overall performance: very good, good, fair, poor, and very poor. Specify as to why it was one of the five.
APPENDIX II Definitions Presence of vibrato is evaluated between present or absent (never) on a five-point rating, where presence can range from ‘‘always,’’ frequently, sometimes, and rarely. Rate of vibrato refers to the speed of oscillations that are perceived in the singing voice and is evaluated on ‘‘too slow,’’; ‘‘just right,’’ and ‘‘too fast.’’ Extent of vibrato: how large or small the oscillations are perceived in the singing voice and is evaluated on ‘‘too wide,’’ ‘‘just right,’’ and ‘‘too narrow.’’
Conspicuousness of vibrato: if it draws undesired attention to itself on perception and is rated on ‘‘too much,’’ ‘‘just right’’, and ‘‘too less.’’ Quality refers to the ‘‘pleasantness’’ of vibrato on perception. It can range from ‘‘always,’’ ‘‘frequently,’’ ‘‘sometimes,’’ ‘‘rarely’’ and ‘‘never.’’ Periodicity refers to the stability of vibrato oscillations on a sustained musical note. Type of vibrato is judged between ‘‘free’’ or ‘‘forced,’’ where free refers to an effortless production and forced refers to an attempt to bring the effect of vibrato during singing. Singer’s overall performance refers to how each singer’s performance is perceived based on parameters such as ‘‘very good,’’ ‘‘good,’’ ‘‘fair,’’ ‘‘poor,’’ and ‘‘very poor.’’
tions as ‘‘small and rapid pitch and intensity changes that occur primarily during singing.’’ Stark16 mentioned two aspects for the perception of vibrato, one is its identification and the second is ‘‘distinguishing between different kinds of vibrato.’’ Rothman and Arroyo suggested that the parameters vital for the perception of vibrato are ‘‘the frequency and amplitude variations around their respective means’’ (cited by Stark16) They could be either frequency or amplitude variations or both. Acoustically, vocal vibrato has been described by Sundberg18 in four parameters as rate, extent, regularity, and waveform. Ideally, these pulsations for a good singer (on an average) should be between 5 and 8 Hz/second, and pulsations slower than this would be perceived by the listener as unpleasant.4 Other researchers found vibrato rate to be between 5.8 Hz and 7.3 Hz12 or 4.5 Hz and 7.2 Hz.11 The extent of vibrato is acceptable to the human ear if it is within 0.5 to 1.0 semitone or even up to 2 semitones.17 However, the terms ‘‘rate’’ and ‘‘extent’’ have been widely accepted and are used by vocal trainers to define vibrato. If the vibrato is too fast, it is called a ‘‘bleat.’’9 If it is too slow or exaggerated, it is called a ‘‘wobble.’’1,9,21 Available literature on past research discusses acoustic parameters of vibrato such as rate (how rapid the modulations take place), extent (how large or small the modulations are),6,11,12 and periodicity.6 Some researchers2,7,8 have attempted at comparing the perceived attributes of vibrato with acoustic measures. Howes et al7 chose to study parameters such as vibrato onset, rate, and extent and other parameters involved with the obtrusiveness of vibrato. The first three parameters were measured on a bipolar visual analog scale. Vibrato onset was measured between ‘‘delayed’’ and ‘‘‘‘immediate,’’ vibrato rate showed ‘‘too slow’’ on the left hand and
604 ‘‘too fast’’ on the right hand and vibrato extent showed ‘‘too wide’’ and ‘‘too narrow’’ on the left and right hand sides, respectively. The judges who evaluated vibrato were 24 qualified teachers of singing. Their findings showed that, although they were similar in their perception in certain parameters, two of the judges found it difficult to assess vibrato, and some of them found it difficult to separate vibrato assessment from overall voice quality. Listeners develop their perceptual abilities without effort and often underestimate their ability to discriminate between different instruments and between different performers playing the same instrument.10 Lopez reported that inexperienced listeners were able to recognise trained singers by their speaking voices (cited by Brown et al3). Kenny and Mitchell8 stated that ‘‘expert listeners’’ can be ‘‘reliable in their judgments of overall quality’’ (pg.55). There is a ‘‘prevailing confusion regarding vibrato that it is never heard as it really is in the actual tone. It is modified by a number of illusions which result in our hearing something entirely different from that which is performed’’ (Seashore, cited by Stark16; pg.147). ‘‘The ability to perceive vibrato differs from one individual to another.’’16 He also stated that some listeners cannot discern it at all. Even if they did, ‘‘Repulsiveness of the bad vibrato has put many a musician on edge against all vibratos . Thus, the good vibrato is ignored through the offense of the bad,’’ (Seashore cited by Stark16; pg.147). For the most part, research has been directed toward acoustic measurements and perceptual–acoustic correlations of vibrato in trained singers and lesser toward perceptual studies alone. Singing is predominantly a perceptual phenomenon as it is tailored for human perception.19 There have been fewer attempts at investigating the differences or similarities of how experienced singers and nonsingers would perceive the same event and in this case, vocal vibrato. Singers rely on the listener’s judgment of their singing. Because vibrato is essential to the singer, how do nonsingers perceive this feature? Would vibrato affect the listener’s perception of the singers’ overall performance? METHOD The purpose of this study was to investigate singers’ perception of vibrato, nonsingers’ perception of vibrato, and if singers’ and nonsingers’ perception of vibrato would affect their ratings of ‘‘singers’ overall performance.’’ In the present study, we hypothesized that there would be no difference between singers’ and nonsingers’ judgments of vocal vibrato. Second, nonsingers’ ratings of singer’s overall performance would not be affected by their perception of vibrato as much as it would affect singers’ ratings on the singer’s overall performance. We reasoned that nonsingers’ exposure to a variety of musical styles may cause them to accommodate different types of vibrato. Materials Definition list (Appendices), recordings, rating scale (Appendices), desktop computer, and a speaker belonging to Computerized Speech Lab software (model 4500; Kay Pentax, Lincoln Park, NJ).
Journal of Voice, Vol. 29, No. 5, 2015
Recordings Ten prerecorded audio samples (recorded on Praat software; Praat, version 5.3.05; developed by Paul Boersma & David Weenink) of professional singers (five male and five female) who had sung the chorus of a hymn (How Great Thou Art) were used for the present study. ‘‘Then sings my soul. My Saviour God to thee. How great Thou art. How Great Thou art. (repeat entire chorus)’’
The professional singers had performed on stage for a minimum of 5 years and had sung contemporary music. Of the 10 professional singers, five (three female and two male) of them were trained and five (two female and three male) were untrained. The untrained singers did not have any formal training in the past and they are well known in the city of Bangalore. The trained singers had 3–5 years of vocal training at the time of recording. They were trained in Western classical music and contemporary style and were still pursuing higher levels in vocals. All singers at the time of recording had been instructed to sing in their comfortable pitch and loudness levels to ascertain a natural production. The men’s voices ranged from baritone to tenor. The female voices ranged from contralto to soprano. Participants The participants who were willing to serve as judges were three professional singers (S1, S2, and S3) and three nonsingers (NS1, NS2, and NS3). The singer judges (SJ) were vocal pedagogs themselves who had 20 years experience in singing and singing instruction of Western classical, Music Theater, and contemporary styles (pop, gospel, and jazz) of music. The nonsinger judges (NSJs) were those who enjoyed listening to Western music on a daily basis over a long period (approximately 20 years) of time. The participant judges were divided into two groups: one singer group and one nonsinger group. Each group consisted of three judges. RATING SCALE Definitions Presence of vibrato was evaluated between present or absent (never). Rate (speed of oscillations that were perceived in the singing voice) was evaluated on ‘‘too slow,’’ ‘‘just right,’’ and ‘‘too fast.’’ Extent (how large or small the oscillations6 are perceived in the singing voice) was evaluated on ‘‘too wide,’’ ‘‘just right,’’ ‘‘too narrow.’’ Conspicuousness of vibrato (if it drew undesired attention to itself on perception) rated on ‘‘too much,’’ ‘‘just right,’’ ‘‘too less.’’ Quality was rated on the ‘‘pleasantness’’ of vibrato on perception. Periodicity referred to the stability of vibrato when it was produced.6 Type of vibrato was judged between ‘‘free’’ or ‘‘forced,’’ where free refers to an effortless production and forced refers to a manual attempt (either throat or diaphragmatic region) to bring the effect of vibrato during singing. Finally, singer’s overall performance
A. Anita Reddy and Uma Subramanian
605
Perception of Vocal Vibrato
referred to how each singer’s performance was perceived by the singer group and nonsinger group (Appendix II). The rating scale was specifically developed for the purpose of this study. It was given to three singing teachers for their opinion on the parameters of vibrato. Once the approval of the singing teachers was taken on the parameters used in the rating scale, it was used for the present study. Parameters such as presence, quality, and periodicity were measured on a five-point rating scale were rated on always, frequently, sometimes, rarely, never, where ‘‘always’’ represented 100% of the time, ‘‘frequently’’75%, ‘‘sometimes’’–50%, ‘‘rarely’’–25%, and ‘‘never’’ represented 0% of the time. Singer’s overall performance was also rated on a five-point rating scale as very good, good, fair, poor, and very poor. Parameters such as rate, extent, and conspicuousness were rated on a three-point rating scale. Type of vibrato was rated on a two-point rating scale as free or forced (Appendix I). Administration procedures The 10 prerecorded audio samples of the chorus of a hymn (How great Thou Art) were randomly arranged and played via a speaker (JBL Professional, China) belonging to CSL Model 4500 in a quiet room. There was no orchestral accompaniment in the background, thereby allowing a clear perception of vibrato. Each singer had sung the entire chorus of the song. The judges were asked to listen to the samples and evaluate each sung sample using the rating scale provided. The group of NSJs was given an explanation of vocal vibrato. Both groups were given a rating scale consisting of seven parameters of vocal vibrato and one parameter related to the singer’s overall performance. They were given an explanation of each parameter that had to be evaluated. All participants were na€ıve to the experiment and had not evaluated vibrato (in particular) earlier. The parameters considered for evaluation were presence, rate, extent, conspicuousness, quality, periodicity, type of vibrato, and singers’ overall performance. However, the definition list was not given to the judges, but each judge was given the same explanation of the parameters that needed clarification. This was carried out to maintain uniformity in their understanding of the requirement of each parameter. Data analysis Once the SJs and NSJs had completed rating the samples, the ratings obtained on each parameter were entered on an Excel
sheet and subjected to statistical analysis. SPSS software version 15 (IBM, Armonk, NY) was used for statistical analysis.
RESULTS The present study investigated the perception of vocal vibrato by SJs and NSJs; also if the SJs and NSJs’ perception of vibrato would affect their ratings on singers’ overall performance. Cohen’s kappa statistics was computed using SPSS software (version 15) for inter-rater reliability within each group, and chi-square tests were computed for measuring agreement across groups (nonsinger vs singer).
Inter-rater reliability of NSJs The ratings of each judge were compared against each other as NS1-NS2; NS1-NS3; and NS2-NS3. All eight parameters of the rating scale showed varied ratings among all three NSJs. On the parameter, ‘‘presence’’ of vibrato, NS1 rated the 10 singers’ samples across four points (always, frequently, sometimes, and rarely) on the rating scale, NS2 rated across three points (always, frequently, and sometimes), and NS3 rated across two points (always and frequently). Because different categories on the rating scale were chosen, the agreement statistics could not be computed between any two judges. The rate of vibrato was rated on three points, such as ‘‘too slow,’’ ‘‘just right,’’ and ‘‘too fast.’’ Ratings varied between judges, yet, among the few singers’ samples that were agreed upon, most of the ratings were given for a ‘‘just right’’ vibrato rate and the other ratings were distributed between too slow and too fast. Kappa values were 0.123 (slight agreement) and 0.310 (fair agreement) and 0.206 (slight agreement) for NS1NS2, NS1-NS3, and NS2-NS3, respectively (Table 1). Therefore, the kappa values obtained for rate of vibrato showed only slight agreement between either group of judges. The extent of vibrato was judged on three parameters such as ‘‘too wide,’’ ‘‘just right’’ and ‘‘too narrow.’’ NS1-NS2 rated the extent of vibrato as ‘‘just right’’ on nine singers’ samples and one as ‘‘too narrow’’ thus having 100% agreement (kappa ¼ 1.000). Their ratings for other parameters were scattered along the length of the rating scale. However, NS1-NS3 and NS2-NS3 agreed on six singers’ samples, where five were rated as ‘‘just right’’ and one was rated as ‘‘too narrow.’’
TABLE 1. Inter-Rater Agreement Between Nonsinger Judges (NS1, NS2, and NS3) on the Number of Singers for Each Parameter Parameter Presence Rate Extent Conspicuousness Quality Periodicity Type Singer’s overall performance
NS1-NS2
NS1-NS3
NS2-NS3
5 5 (K ¼ 0.123) 10 (K ¼ 1.000) 7 2 2 5 4
2 6 (K ¼ 0.310) 6 3 4 3 5 (K ¼ 0.194) 6
3 5 (K ¼ 0.206) 6 2 1 4 7 4
606 Because the categories selected for rating varied, there was no significant agreement among them. Conspicuousness of vibrato was rated on ‘‘too much,’’ ‘‘just right,’’ and ‘‘too less.’’ NS1 and NS2 agreed on seven singers’ samples, where two were perceived as ‘‘too less’’ and five were perceived as ‘‘just right,’’ whereas NS1 and NS3 agreed only on three singers’ samples as being ‘‘just right.’’ NS2 and NS3 agreed only on two samples, where one sample was perceived as having a vibrato that was ‘‘too much’’ vibrato and another sample of vibrato was perceived as ‘‘just right.’’ Rankings differed among them; hence, agreement statistics could not be computed. Quality of vibrato. This parameter was rated on a five-point rating scale on how pleasant the vibrato sounded for each singer (always [100% of the time], frequently [75%], sometimes [50%], rarely [25%], and never [0% of the time]). With reference to quality of vibrato, NS1 and NS2 agreed only on two singers’ samples, and hence, there was a poor agreement among them. NS1 and NS3 agreed on four samples of which one singer’s sample was rated as having a pleasant vibrato ‘‘always’’ (100% of the time), another as having a pleasant vibrato ‘‘frequently’’ (75% of the time) and the rest as having a pleasant vibrato only ‘‘sometimes’’ (50% of the time). NS2 and NS3 agreed on only one case as having a pleasant vibrato ‘‘sometimes’’ (50% of the time). Because judges varied in their rankings, agreement statistics could not be computed. Periodicity (stability) of vibrato. The periodicity of vibrato was rated on whether vibrato was stable always (100%), frequently (75%), sometimes (50%), rarely (25%), or never (0% of the time), while singing. NS1 and NS2 agreed on only two singers’ samples. NS1 and NS3 agreed on the rating of periodicity on three samples, where two of the singers were rated as having a periodic vibrato ‘‘always’’ (100% of the time) and one was ‘‘rarely’’ periodic or stable (25% of the time). NS2 and NS3 agreed on four samples where three of them were rated as having a stable vibrato only ‘‘sometimes’’ (50% of the time) while singing. Because the rankings largely varied among the judges, agreement statistics could not be computed. Type of vibrato. The type of vibrato was rated across two points between ‘‘free’’ and ‘‘forced.’’ NS1 and NS2 agreed on five singers’ samples, of which three were rated as ‘‘forced’’ and two were rated as ‘‘free.’’ NS2 rated one sample as ‘‘cannot say.’’ Because their rankings varied, no significant agreement was found among them. NS1 and NS3 agreed on five samples, of which three were rated as ‘‘forced’’ and two were rated as ‘‘free.’’ Because the categories selected for rating were similar for the five samples, they had a slight agreement (kappa ¼ 0.194). Although NS2 and NS3 agreed on seven samples (70%), of which two were rated as ‘‘forced’’ and five were rated as ‘‘free,’’ (as mentioned earlier, NS2 had rated one sample as ‘‘cannot say’’) no significant agreement was found among them. Overall performance. Each singer’s overall performance was rated on five points, such as very good, good, fair, poor, and very poor. NS1 and NS2 agreed only on four singers (40%) as having a good performance. NS1 and NS3 agreed on the ratings of overall performance on six singers (60%), of
Journal of Voice, Vol. 29, No. 5, 2015
whom two singers were rated as fair, three as good, and one as very good; however, there was no significant agreement among them. NS2 and NS3 agreed on a good performance for four singers (40%). Because they varied in their rankings, agreement statistics could not be computed. Parameters such as ‘‘presence,’’ ‘‘conspicuousness,’’ ‘‘quality,’’ ‘‘periodicity,’’ ‘‘type,’’ and ‘‘singer’s overall performance’’ ratings showed poor agreement between all three NSJs (kappa values that were obtained wherever possible are shown in Table 1). Inter-rater reliability of SJs Three experienced singers (S1, S2, and S3) participated as judges. They also had to rate singers’ vibrato on the same parameters. The singer judges were grouped as S1-S2, S1-S3, and S2-S3. Although all the three singer judges agreed on vibrato being present in all 10 singers yet, they varied on the five categories (always, frequently, sometimes, rarely, and never) selected for rating. Hence, agreement statistics on presence of vibrato could not be computed (between any two groups of SJs). Rate of vibrato showed only slight agreement (kappa value ¼ 0.231) between S1 and S3. These judges (S1 and S3) agreed that of 10 samples, two samples had a vibrato that was ‘‘too slow’’ and three that were ‘‘just right.’’ However, only S3 perceived one sample to have a fast vibrato between the two SJs. Although the other two groups of SJs (S1-S2 and S2-S3) agreed on seven samples (ie, S1-S2) and three samples (S2-S3), respectively, the categories (too slow, just right, and too fast) selected for rating varied; hence, agreement statistics could not be computed (Table 2). For the extent of vibrato, S1 and S3 perceived two samples as ‘‘too wide,’’ three as ‘‘just right’’ and two as ‘‘too narrow,’’ leading to a significant agreement between them on a total of seven samples (kappa value ¼ 0.545/moderate agreement; approximate significance ¼ 0.010). The other two groups of SJs (S1-S2 and S2-S3) did not show adequate agreement between their respective groups. Hence, kappa scores could not be obtained (Table 2). Only S1 and S3 showed moderate agreement (seven samples; kappa value ¼ 0.455) on conspicuousness of vibrato (Table 2). Here they perceived one singer to have a very conspicuous (too much) vibrato, five singers whose vibrato was ‘‘just right’’ and one singer whose vibrato was ‘‘too less.’’ The other two groups of judges varied in their rankings, and hence, agreement statistics could not be computed. With reference to the quality of vibrato, there was moderate agreement (kappa value ¼ 0.474) on seven samples between S2 and S3. They perceived two samples to have a pleasant vibrato 75% (frequently pleasant) of the time, four samples to have a pleasant vibrato 50% (sometimes pleasant) of the time and one sample with a pleasant vibrato 25% (rarely pleasant) of the time. The other two groups of SJs (S1-S3 and S1-S2) showed poor agreement; hence, agreement statistics could not be computed (Table 2). There was substantial agreement (kappa ¼ 0.655) between S2 and S3 over periodicity of vibrato on eight samples. They agreed that one sample had a stable vibrato 100% of the time, three samples had a stable vibrato 75% of the time, and four samples had a
A. Anita Reddy and Uma Subramanian
607
Perception of Vocal Vibrato
TABLE 2. Inter-Rater Agreement Between Singer Judges (S1, S2, and S3) on the Number of Singers Parameter Presence Rate Extent Conspicuousness Quality Periodicity Type Singer’s overall performance
S1 and S2
S1 and S3
S2 and S3
7 7 5 6 3 3 7 6 (K ¼ 0.385)
6 5 (K ¼ 0.231) 7 (K ¼ 0.545) 7 (K ¼ 0.455) 6 3 6 3
6 3 8 9 7 (K ¼ 0.474) 8 (K ¼ 0.655) 7 4
stable vibrato 50% of the time (Table 2). Again, the other two groups of judges (S1-S3 and S1-S2) varied in the categories selected for rating, and hence, kappa scores could not be obtained. Type of vibrato showed ratings on equal number of samples (seven singers) between S1-S2 and S2-S3 (Table 2) nevertheless, S1 mentioned that she could not tell if one of the singers’ vibrato was free or forced and S3 mentioned that one singer’s vibrato was ‘‘in between’’ free and forced (S1 rated sample ‘‘H’’ as ‘‘cant say’’ and S2 rated sample ‘‘G’’ as ‘‘in between’’). Because they did not agree on similar categories, agreement statistics could not be computed. On singer’s overall performance, there was fair agreement (kappa ¼ 0.385) between S1 and S2, where they perceived two singers to have a ‘‘fair’’ performance, three singers as ‘‘good,’’ performance and one singer as ‘‘very good’’ performance. The other groups of SJs did not agree on similar categories, and hence, agreement statistics could not be computed (Table 2). In summary, judges S1 and S3 had higher agreement than the other groups of SJs. Although S1 and S3 showed agreement on lesser number of singers for parameters such as rate, extent, and conspicuousness, yet they seemed to agree on similar categories on each rating scale. The kappa agreement varied between fair (K ¼ 0.231) to moderate agreement (K ¼ 0.545) between them (Table 2). Although S2-S3 showed agreement only on two parameters (quality and periodicity), it ranged from moderate to substantial agreement. Comparison of ratings between NSJs and SJs A statistical analysis of singers and nonsingers’ ratings on all eight parameters was done using chi-square tests to examine if there were differences in the rankings between singers and nonsingers. Singers and nonsingers’ ratings on the presence of vibrato were compared. Both SJ and NSJ had similar ratings of vibrato occurring only 50% of the time. NSJs agreed on five samples and SJs agreed on six samples. Only one singer was perceived to have a vibrato 100% of the time (always) during singing by two SJs and one NSJ. For this reason, there was no significant difference in their ratings (Table 3). Rate of vibrato was perceived as ‘‘just right’’ by NSJs who showed a slightly higher percentage (50%) of rating than SJs (44.4%). The rest of the ratings were distributed between ‘‘too fast’’ and ‘‘too slow.’’ Therefore, both NSJs and SJs did not show a significant difference in their ratings. Extent of
vibrato was rated on ‘‘too wide,’’ ‘‘just right,’’ and ‘‘too narrow.’’ SJs agreed on four samples (40%) whose vibrato extent was perceived as ‘‘just right’’ and four other samples (40%) as ‘‘too narrow.’’ Nonsingers on the other hand, agreed on nine samples (90%) whose vibrato extent was perceived as ‘‘just right’’ and on one sample as ‘‘too narrow.’’ Thus, there was a significantly higher percentage of rating among NSJs compared with SJs. Although the ratings did show a higher percentage among NSJs, it was not statistically significant (Table 3). Conspicuousness of vibrato was rated on ‘‘too much,’’ ‘‘just right,’’ and ‘‘too less.’’ SJs had higher ratings on ‘‘just right’’ (eight samples; 80%), ‘‘too much’’ (one sample; 10%), and ‘‘too less’’ (one sample; 10%) than NSJs, whose ratings were distributed between ‘‘just right’’ (four samples; 40%) and ‘‘too less’’ (five samples; 50%) and ‘‘too much’’ vibrato (one sample; 10%). Thus, there was a slightly higher percentage of rating among SJs on the conspicuousness of vibrato being just right. Yet, there was no significant difference between both groups (Table 3). Quality of vibrato had equal ratings by both SJs and NSJs, and they seemed to agree that vibrato was pleasant only 50% (sometimes) of the time in five samples and 75% of the time in four other samples. There was a difference in their rating only by one sample, where NSJs rated one sample as ‘‘always’’ pleasant (100% of the time) and SJs rated one sample as ‘‘rarely’’ pleasant. Thus, similar ratings were found between SJs and NSJs. Hence the chi-square analysis did not show a significant difference (Table 3). With reference to periodicity of vibrato, SJs had rated six samples as sometimes stable, three as frequently stable, and only one as always stable. But NSJs rated only three as TABLE 3. Comparison of ratings between NSJs and SJs Parameter Presence Rate Extent Conspicuousness Quality (pleasantness) Periodicity (stability) Type Singer’s overall performance
Singer (%) Nonsinger (%) 60 44.4 40 80 50 60 90 44.5
50 50 90 50 50 30 55.6 70
608 ‘‘sometimes’’ stable, two as ‘‘frequently’’ stable, four as ‘‘always’’ stable, and one as ‘‘rarely’’ stable. There was no significant difference among the two groups of judges (Table 3). SJs rated nine samples (90%), as having a ‘‘free’’ vibrato and only one sample as ‘‘forced’’ vibrato. So, there was significantly higher percentage of rating (90%) among SJs on a ‘‘free’’ vibrato than NSJs, whose ratings were distributed between ‘‘free vibrato’’ (five samples; 55.6%) and ‘‘forced’’ vibrato (44.4%). One sample was judged by NS2 as ‘‘can’t say.’’ Although SJs had a higher percentage of rating, there was no significant difference between both groups (Table 3). Overall performance (of singing) of each singer was rated on a five-point rating scale (very good, good, fair, poor, and very poor). NSJs rated 70% of singer’s overall performance as being ‘‘good,’’ 20% as ‘‘fair’’ and 10% as ‘‘very good.’’ On the other hand, SJs agreed on 44.5% as ‘‘good’’ performance and 44.4% as ‘‘fair’’ performance and 11.1% as ‘‘very good’’ performance. Thus, NSJs had higher ratings than SJs but showed no statistical significance (Table 3). DISCUSSION This study aimed at investigating how singers’ and nonsingers’ perceived vibrato and if perception of vocal vibrato would affect their ratings of singer’s overall performance. Rating vocal vibrato was challenging for both singers and nonsingers. Because the judges had to focus only on vibrato while listening to the entire music excerpt, they found it intricate to rate vibrato on each of the eight parameters. It was interesting to note that NSJs perceived most singers to have an appropriate rate and extent for their vibrato, but SJs did not perceive it the same way, the ratings seemed to be divided between appropriate and either too slow or fast. It seemed that SJs perceived vibrato in a more detailed manner for almost all parameters. Rate and extent of vibrato did not seem to bother the NSJs as much as it did for SJs, and this may be attributed to the nonsingers having exposure to not just different genres of music besides classical and pop music but also to singers who may not have had formal vocal training and yet have begun their singing career. This is where they may have heard vibratos ranging from a narrow to a wide extent. It also may be possible that trained singers get accustomed to hearing a certain rate and extent of vibrato and anything above or below a certain range would be considered for correction. If the rate and extent of vibrato was at either extreme (too fast or too slow and too narrow or too wide, respectively) it would still sound unpleasant (which is mostly seen during the early vocal training period). Vocal pedagogs who participated in this study mentioned that vibrato is never taught specifically but the student may be instructed to correct it if it drew attention to itself. By and large, vocal pedagogs allow the fluctuations of pitch in the student’s voice and allow vibrato to gain its own speed (Van Christy, 1967). With reference to conspicuousness of vibrato, NSJs could not conclude at a particular category of rating but SJs were more consistent, where vibrato was perceived to be ‘‘just right’’ for most of the samples (Table 2). Singers who are vocal pedagogs
Journal of Voice, Vol. 29, No. 5, 2015
have practical knowledge of vibrato, whereas nonsingers have a more psychoacoustical outlook toward vibrato. The group of NSJs did mention that it was a difficult task as they had not perceived vibrato in such detail in the past as it is generally overridden by the performer’s voice and the orchestra. In general, no one focuses on listening to a vibrato rather than the song (unless it is done intentionally). The manner by which SJs perceived vibrato seemed to reflect how each of these parameters, such as presence of vibrato, conspicuousness, periodicity, and type of vibrato, were connected. The influence of other parameters on the quality of vibrato was evident. This was perceived by both SJs and NSJs. Perhaps, it may be assumed that ‘‘presence’’ is the foremost feature for any musical sound and ‘‘quality’’ is its final feature. Even if one of these parameters, such as rate, extent, conspicuousness, or periodicity, embedded in between these two features (presence and quality) was not appropriate, the final output (quality) would be affected. For example, if extent was too wide, it would still sound unpleasant, or if vibrato was not periodic/stable, it would still affect the quality. Diaz and Rothman5 studied the acoustic comparison between good and poor vibratos in singers and found that the good vibratos were the most periodic ones. Vibrato has also been reported to have an influence on overall quality of singing voice8 and its presence is what determines a ‘‘trained sound.’’22 With reference to type of vibrato, whether free or forced, a free vibrato would be the result of a free vocal tone which is again the result of ‘‘freedom and good technique.’’7 But a forced vibrato would often be produced by manipulating the throat (in an up and down manner) or diaphragmatic region (by pulsating it voluntarily) or through constricted vocal cords.1 One SJ expressed that she could not tell if vibrato was free or forced, for one of the samples. Another SJ also did mention that one sample of vibrato was ‘‘in between’’ free and forced (it was not the same sample). Nevertheless, SJs had a higher percentage of rating on a free vibrato rather than forced vibrato unlike NSJs. SJs did remark on the overall performance of singers and also on the vibrato. Remarks such as good vocal tone, controlled/ uncontrolled vibrato, too much pitch variation, inadequate breath support, good vocal agility, and weak/good vocal resonance had been written at the end of the rating scale, in the space below. SJs had distributed ratings between good and fair performance. NSJs also made similar remarks with regard to singers’ pitches, vibrato rate decreasing along with lowering of pitch, control of voice. Despite remarks of such kind, the singer’s overall performance ratings ranged from fair to very good performance by the NSJs. But when SJs’ ratings were compared with each other within their respective group, they had a wider range, ranging from poor to very good. Perhaps, it can be attributed to their (SJs) experience in teaching and training students on their performance that their expectations of overall performance would have been higher. Finally, it was observed that third NSJ (NS3) had similar ratings as SJs (S1 and S2). The varied ratings among NSJs may be justified for this reason.
A. Anita Reddy and Uma Subramanian
Perception of Vocal Vibrato
The present study brought forth three aspects: (1) SJs perceived vibrato on practical aspects such as presence, conspicuousness, periodicity, and type of vibrato, which is not surprising as they are more mindful of what kind and how much of vibrato is required. On the other hand, NSJs did not seem to have too much concern with regard to rate and extent of vibrato. (2) The perception of the quality of vibrato seemed to be similar for both SJs and NSJs. (3) Perceptual ratings on singer’s overall performance seemed to be affected for the SJs but it did not seem to affect the ratings of NSJs. However, it cannot be said that the ratings of singer’s overall performance were affected because of vibrato. It could be attributed to the tone, breath support, quality of voice, and resonance, based on how both groups of judges had mentioned as remarks. Furthermore, it can also be a possibility that the location of the parameter singer’s overall performance at the very end of the score sheet may have been a possible reason for the ratings on singer’s overall performance being affected (for SJs). CONCLUSION Limitations of the study Intra-rater reliability was not measured for both groups of judges as it would have given a better representation of each individual’s perception of vibrato. Perception differs from person to person; nevertheless, singers rely on their listeners’ feedback. Furthermore, this rating scale provides detailed features of vocal vibrato, which may be used by singing teachers to explain to their students where their vibrato may need correction. Because production of vibrato depends on good singing technique, this rating scale may suggest on improvement of singing techniques for the young student. REFERENCES 1. Bennet G. From Rock to Opera: The Basic Technique of Voice. 2nd ed. Milwaukee, WI: Hal Leonard Corporation; 2001:65.
609
2. de Almeida Bezerra A, Cukier-Blaj S, Duprat A, Camargo Z, Granato L. The characterization of the vibrato in Lyric and Sertanejo singing styles: acoustic and perceptual auditory aspects. J Voice. 2009;23:666–670. 3. Brown WS Jr, Rothman HB, Sapienza CM. Perceptual and acoustic study of professionally trained versus untrained voices. J Voice. 2000;14:301–309. 4. Bunch MD. Dynamics of Singing Voice. 5th ed. New York, NY: Springer Wien; 2009:108–109. 5. Diaz JA, Rothman HB. Acoustical comparison between samples of good and poor vibrato in singers. J Voice. 2003;17:179–184. 6. Dromey C, Carter N, Hopkin A. Vibrato rate adjustment. J Voice. 2003;17: 168–178. 7. Howes P, Callaghan J, Davis P, Kenny D, Thorpe W. The relationship between measured vibrato characteristics and perception in western operatic singing. J Voice. 2004;18:216–230. 8. Kenny DT, Mitchell HF. Acoustical and perceptual appraisal of vocal gestures in the female classical voice. J Voice. 2006;20:55–70. 9. McKinney CJ. The Diagnosis and Correction of Vocal Faults. Revised ed. Naashville, TN: Genevox Music Group; 1994. 10. Mitchell HF, MacDonald RAR. Non-singers as spectators? Audio visual integration improves music performer identification. Psychol Music. 2012;42:112–127. 11. Murbe D, Zahnert T, Kuhlisch E, Sundberg J. Effects of professional singing education on vocal vibrato- a longitudinal study. J Voice. 2007; 21:683–688. 12. Prame E. Measurement of vibrato rate of ten singers. J Acoust Soc Am. 1994;96:1979–1984. 13. Sanders AD. Auditory Perception of Speech: An Introduction to Principles and Problems. Englewood, NJ: Prentice Hall Inc; 1977:14. 14. Seashore CE. Psychology of Music. New York, NY: Dover Publications Inc; 1967:33. 15. Sell K. The Disciplines of Vocal Pedagogy: Towards an Holistic Approach. Aldershot, UK: Ashgate Publishing Co; 2005:117–118. 16. Stark AJ. Bel Canto: A History of Vocal Pedagogy. Toronto, Canada: University of Toronto Press; 1999:146–147. 17. Sundberg J. The Science of the Singing Voice. Dekalb, IL: Northern Illinois University Press; 1987:163–164. 18. Sundberg J. Acoustic and psychoacoustic aspects of vocal vibrato. In: Dejoncjeres PH, Hirano M, Sundberg J, eds. Vibrato. San Diego, CA: Singular publishing; 1995:39. 19. Sundberg J. The perception of singing. In: Deutsch D, ed. The Psychology of Music. 2nd ed. San Diego, CA: Academic Press Series in Cognition and Perception; 1999:171–214. 20. Van Christy A. Expressive Singing. Dubuque, IA: Wm.C. Brown Company Publishers; 1972:42–43. 21. Zemlin RW. Speech and Hearing Science: Anatomy and Physiology. 4th ed. Englewood Cliffs, NJ: Allyn and Bacon; 1998:170–171. 22. Sataloff RT. Professional Voice: The Art and Science of Clinical Care. 3rd Edn. Vol. 2. San Diego, CA: Plural Publishing; 2005.
610
Journal of Voice, Vol. 29, No. 5, 2015
APPENDIX I Score sheet Instructions to judges: parameters such as presence, quality, periodicity, and singer’s overall performance are rated on a
Parameter
Always
five-point rating scale, where always ¼ 100%; almost always ¼ 75%; sometimes ¼ 50%; rarely ¼ 25%; and never ¼ 0% of the time. Parameters such as rate, extent, and conspicuousness are rated on a three-point rating scale, and parameters such as type are rated on just two points. Kindly tick wherever appropriate.
Presence of vibrato Frequently Sometimes
Rarely
Never
Presence of vibrato Too slow
Rate of vibrato Just right
Too fast
Too wide
Extent of vibrato Just right
Too narrow
Too much
Conspicuousness (obvious) of vibrato Just right
Too less
Parameter
Always
Quality of vibrato Frequently Sometimes
Rarely
Never
Always
Periodicity (stability) of vibrato Frequently Sometimes
Rarely
Never
Pleasant Parameter Stable Type of vibrato Free/forced Singer’s overall performance: very good, good, fair, poor, and very poor. Specify as to why it was one of the five.
APPENDIX II Definitions Presence of vibrato is evaluated between present or absent (never) on a five-point rating, where presence can range from ‘‘always,’’ frequently, sometimes, and rarely. Rate of vibrato refers to the speed of oscillations that are perceived in the singing voice and is evaluated on ‘‘too slow,’’; ‘‘just right,’’ and ‘‘too fast.’’ Extent of vibrato: how large or small the oscillations are perceived in the singing voice and is evaluated on ‘‘too wide,’’ ‘‘just right,’’ and ‘‘too narrow.’’
Conspicuousness of vibrato: if it draws undesired attention to itself on perception and is rated on ‘‘too much,’’ ‘‘just right’’, and ‘‘too less.’’ Quality refers to the ‘‘pleasantness’’ of vibrato on perception. It can range from ‘‘always,’’ ‘‘frequently,’’ ‘‘sometimes,’’ ‘‘rarely’’ and ‘‘never.’’ Periodicity refers to the stability of vibrato oscillations on a sustained musical note. Type of vibrato is judged between ‘‘free’’ or ‘‘forced,’’ where free refers to an effortless production and forced refers to an attempt to bring the effect of vibrato during singing. Singer’s overall performance refers to how each singer’s performance is perceived based on parameters such as ‘‘very good,’’ ‘‘good,’’ ‘‘fair,’’ ‘‘poor,’’ and ‘‘very poor.’’
