Acoustic Analysis of Voice in Habitual Phonation and Adhara Shruthi in Carnatic Classical Singers

ACOUSTIC ANALYSIS OF VOICE IN HABITUAL PHONATION AND ADHARA SHRUTHI IN CARNATIC CLASSICAL SINGERS The term singing has been defined simply as producing musical tones by means of the voice (Vennard, 1962). Although definition appears to be fairly simple, the act of singing is much more involved and requires a variety of phonatory, articulatory and resonatory adjustments. Singing differs from speaking, quantitatively and qualitatively on some parameters like rhythm and melodic qualities (Luchsinger, 1965). In singing, vowels are modified by changing source and filter to produce a wider dynamic range.

This enables to maintain a balance in loudness across phonemes and produce vocal timbers (Titze, 1989).

It also allows shifts in formant frequencies and subtle adjustments in the timbre and vowel quality. This shift in formant frequencies/ articulatory adjustments leads to appearance of Singer’s formant (Fs). Singers formant is an exceptionally high spectrum envelope peak around 3 KHz in the vowel spectra of singers (Sundberg et al, 1995). This produces exciting ringing quality and gives an arresting “edge” to the voice. Indian classical singing is mainly classified as of two types namely, Carnatic and Hindustani style.

In general, they have the same basis, being melodic and governed by rules of ‘raga’ and ‘taala’ structure. Carnatic style is art form of south India. The carnatic school claims to have maintained and developed and orthodox traditional while the Hindustani school is said to have experienced considerable changes and developments through the Moghul period up to the present time (Sambamurthi, 1982). In Indian classical music, each raga is composed of a different set of tones.

Some have more notes in the lower and middle octaves while some have more in the higher octaves.

While singing vocalist needs to be careful in selecting his ‘adharashruthi’ as the shruthi selected should enable them to comfortably cover all the notes in the raga as well as maintain a good appealing voice. Generally, our clinical observation indicates that many singers stabilize their ‘adhara shruthi’ at a slightly higher or lower frequency in relation to their natural fundamental frequency (habitual frequency). However, Chandra (2001) reported lower F0 values for ‘aadhara shruthi’ when compared to habitual phonation. Need Initiation into classical music, at least in our country, is generally done at a very young age (around 3 – 5 years).

At that age children will not have a conscious option to select any particular level of ‘adhara shruthi’, and invariably are guided by their teacher. However, as the children continue their music lessons and are introduced to more ragas, they become aware of tonal structure of the ragas and the range they have to traverse up and down the scale. This may necessitate readjustment of their ‘adhara shruthi’ level. Chandra (2001) commented that F0 of habitual phonation and ‘adhara shruthi’ were same in singers with less than six years of training, but different in singers with more than six years of training.

This indicates there is effect of years of singing experience on adjustment of adhara shruthi.

At this point of time limited information is available about adjustments made by the vocalists with regard to ‘adhara shruthi’. Also there are no reports available about the comparison of voice quality of vocalists in their habitual phonation and ‘adhara shruthi’ as wrong pitch selection not only ruins the songs and music concert, but its continued use may ruin the singers voice too.

Because inappropriate use of pitch for long time lead to changes in the physiological mechanism of the vocal folds which further may lead to development voice problems such as nodules and polyps. Further it will be interesting to see whether the number of years of training is a factor influencing their readjustment of ‘adhara shruthi’. Therefore, there is need of a study which compares voice related attributes in habitual phonation and ‘adhara shruthi’ , and to further see if there is any effect of years of training on this.

In this regard the present study was planned.

The purpose of the present study is investigate voice related attributes between habitual phonation ‘adhara shruthi’ as a function of number of years of training in vocal music. Objectives The objectives of the present study were to determine: a) The relationship between habitual phonation and ‘adhara shruthi’ in a group of female singers, b) To investigate effect of number of years of training on the relationship between habitual phonation and ‘adhara shruthi’. Method Subjects: Thirty female carnatic classical singers participated in the present study.

The participants where divided into three groups (ten in each) based on years of experience.

Group I consisted of ten children with a mean age of 10 years (range 5 -10 years) with singing experience of 5 years. Group II consisted of ten adults with a mean age of 30 years (range 20 – 40 years) with singing experience of 10 to 15 years. Group III consisted of ten geriatric singers with a mean age of 60 years (range (55 – 72 years) with singing experience of above 40 yrs. All the participants received training in carnatic style classical music.

Only those individuals without any history of vocal pathology, hearing loss or any pathology of upper respiratory tract at the time of the recording were selected.

Material: Following tasks was recorded from each subject. * Task I: Subjects are asked to take deep inhalation and phonate vowels /a/ at a comfortable habitual pitch. * Task II: Subjects are asked to take deep inhalation and phonate vowels /a/, at their ‘adhara shruthi’ pitch. Recording: Samples were recorded in a quiet set-up with minimal noise. Subjects were instructed to phonate into the microphone kept at a distance of 5 cm away from their mouth.

Readings were directly recorded on to a digital recorder.

Both tasks were recorded thrice. Acoustical analysis: The recorded samples were directly line fed to the computer memory using speech interface unit of Vaghmi software (Voice and speech systems, Bangalore). The Vaghmi software was used for analysis. The following parameters were analyzed: 1. Mean fundamental frequency 2. Fundamental frequency range: difference between the highest and the lowest fundamental frequency.

3. Mean intensity 4. Intensity range: difference between the highest and the lowest intensity. . Jitter: it is the cycle to cycle variation in the fundamental frequency. 6.

Shimmer: it is the cycle to cycle variation in the intensity. 7. Alpha ratio: Energy (0 to 1kHz)/Energy (1 to 5kHz) 8. Beta ratio: Energy (0 to 2kHz)/Energy (2 to 8kHz) 9. Gamma ratio: Energy (0 to 1kHz)/Energy (5 to 8kHz) 10.

Harmonic to noise ratio (HNR) Statistical analysis: The analyzed data tabulated for each subject was subjected to statistical analysis. SPSS (Version 11) was used for the statistical analysis. Means and standard deviations were calculated.

Paired sampled ‘t’test was done to find the significant difference between the habitual phonation and ‘adhara shruthi’ phonation. One way ANOVA was used to compare across the groups.

Results 1. Mean fundamental frequency: The mean fundamental frequency values were lower in adhara shruthi for group II and group III. The results of paired sample‘t’ test revealed significant difference between habitual phonation and adhara shruthi only in group III [Group I: t=0. 139, df=9, p>0. 05; Group II: t=1. 532, df=9, p>0.

05; Group III: t=4. 777, df=9, p <0. 5]The results revealed higher mean fundamental frequency values in habitual phonation for group II and in ‘adhara shruthi’ phonation for group I. The results of one way ANOVA indicated statistically significant difference between the three age groups only for adhara shruthi [Habitual phonation- F (2,27)=1. 088, p;gt;0. 05; Adhara shruthi- F(2,27)=3.

391, p ;lt; 0. 05]. The results of the Duncan’s post hoc analysis indicated statistically significant difference (p< 0. 05) between group I and group III. Figure 1 shows error bars for mean F0 in all the three groups in habitual phonation and adhara shruthi.

Figure 1: Error bars for mean F0 in three groups for habitual phonation and adhara shruthi.

2. Fundamental frequency range: The mean frequency range was lower in adhara shruthi for all the three groups. The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi in all the three groups [Group I: t=0. 347, df=9, p;gt;0. 05; Group II: t=1. 195, df=9, p;gt;0.

05; Group III: t=1. 222, df=9, p;gt;0. 05]. The results revealed higher frequency range values in habitual phonation and adhara shruthi for group III.

The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=1. 333, p ;gt;0.

05; Adhara shruthi – F(2,27)=0. 359, p;gt;0. 05]. Figure 2 shows error bars for F0 range in all the three groups in habitual phonation and adhara shruthi. Figure 2: Error bars for F0 range in three groups for habitual phonation and adhara shruthi.

3. Mean intensity: The mean intensity values were higher in adhara shruthi for group I and lower in group II and group III.

The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi. [Group I: t=1. 312, df=9, p>0.

05; Group II: t=. 837, df=9, p>0. 05; Group III: t=. 366, df=9, p >0. 05]. The results revealed higher mean intensity values in habitual phonation for group II and in adhara shruthi for group I.

The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=1. 177, p >0. 05; Adhara shruthi- F(2,27)=1. 032, p>0. 05].

Figure 3 shows error bars for mean intensity in all the three groups in habitual phonation and adhara shruthi.

Figure 3: Error bars for mean intensity in three groups for habitual phonation and adhara shruthi. 4. Intensity range: The mean intensity range values were lower in adhara shruthi for group I and group II and higher in group III. The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi for all the three groups [Group I: t=0. 274, df=9, p;gt;0.

05; group II: t=0. 393, df=9, p;gt;0. 05; group III: t=1. 087, df=9, p ;gt;0. 05. ].

The results revealed higher intensity range values in habitual phonation for group I and in adhara shruthi for group III. The results of one way ANOVA indicated no statistically significant difference between the three age groups. [Habitual phonation- F (2, 27) =. 804, p ;gt;0. 05; Adhara shruthi- F (2, 27) =0.

015, p;gt;0. 05]. Figure 4 shows error bars for intensity range in all the three groups in habitual phonation and adhara shruthi. Figure 4: Error bars for intensity range in three groups for habitual phonation and adhara shruthi. 5.

Jitter: The mean jitter values were higher in adhara shruthi for group II and group III.

The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi in all groups [Group I: t=1. 53, df=9, p>0. 05; Group II: t=1. 366, df=9, p>0. 05; Group III: t=1.

635, df=9, p >0. 05]. The results revealed high jitter values in habitual phonation for group I and in adhara shruthi phonation for group III. The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=1. 349, p >0.

05; adhara shruthi- F(2,27)=0. 388, p>0. 05].

Figure 5 shows error bars for jitter in all the three groups in habitual phonation and adhara shruthi. Figure 5: Error bars for jitter in three groups for habitual phonation and adhara shruthi.

6. Shimmer: The mean shimmer values were lower in adhara shruthi for group II and higher for group III. The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi [Group I: t=0. 216, df=9, p;gt;0. 05; Group II: t=0. 525, df=9, p;gt;0.

05; Group III: t=1. 642, df=9, p ;gt;0. 05]. The results revealed high shimmer values in habitual phonation for group I and in adhara shruthi for group III.

The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=0.

861, p ;gt;0. 05; Adhara shruthi- F(2,27)=0. 226, p;gt;0. 05]. Figure 6 shows error bars for shimmer in all the three groups in habitual phonation and adhara shruthi. Figure 6: Error bars for shimmer in three groups for habitual phonation and adhara shruthi.

7. Alpha ratio: The mean alpha values were lower in adhara shruthi for group II and group III, but higher in group I.

The results of paired sample‘t’ test revealed significant difference between habitual phonation and adhara shruthi in adult group [Group I: t=1. 044, df=9, p>0. 05; Group II: t=2.

753, df=9, p<0. 05; Group III: t=1. 000, df=9, p >0. 05]. The results revealed high alpha values in habitual phonation and adhara shruthi phonation for group III. The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F (2, 27) =0.

675, p>0. 05; Adhara shruthi- F (2, 27) =2. 31 p>0. 05]. Figure 7 shows error bars for alpha ratio in all the three groups in habitual phonation and adhara shruthi.

Figure 7: Error bars for alpha ratio in three groups for habitual phonation and adhara shruthi. 8. Beta ratio: The mean beta values were lower in adhara shruthi for group I and group II, but higher for group III. The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi in any groups [Group I: t=0. 138, df=9, p;gt;0.

05; Group II: t=0. 387, df=9, p;gt;0. 05; Group III: t=0. 896, df=9, p ;gt;0. 5].

The results revealed high beta values in habitual phonation for group II and in adhara shruthi phonation for group III. The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=0. 401, p ;gt;0. 05; Adhara shruthi – F(2,27)=0. 834, p;gt;0.

05]. Figure 8 shows error bars for beta ratio in all the three groups in habitual phonation and adhara shruthi. Figure 8: Error bars for beta ratio in three groups for habitual phonation and adhara shruthi. 9.

Gamma ratio: The mean gamma values were higher in base pitch for adults and geriatrics, but lower for children. The results of paired sample‘t’ test revealed no significant difference between phonation and basepitch in any groups [Group I: t=1.

406, df=9, p>0. 05; Group II: t=1. 464, df=9, p>0. 05; Group III: t=1. 867, df=9, p >0. 05].

The results revealed high gamma values for both habitual phonation and adhara shruthi in groups III. The results of one way ANOVA indicated statistically significant difference between the three age groups in adhara shruthi [Habitual phonation- F (2, 27) =0. 19, p >0. 05; Adhara shruthi F (2, 27) =4. 136, p<0.

05]. The results of the Duncan’s post hoc analysis indicated statistically significant difference (p;lt; 0. 05) between group I and group III. Figure 9 shows error bars for gamma ratio in all the three groups in habitual phonation and adhara shruthi. Figure 9: Error bars for gamma ratio in three groups for habitual phonation and adhara shruthi.

10. Harmonic to noise ratio: The mean harmonic to noise ratio values were lower in adhara shruthi for groups III, but higher for group I and group II.

The results of paired sample‘t’ test revealed no significant difference between habitual phonation and adhara shruthi. [Group I: t=0. 888, df=9, p>0. 05; Group II: t=0.

614, df=9, p>0. 05; Group III: t=1. 335, df=9, p >0. 05]. The results revealed high harmonic to noise ratio values in habitual phonation for group III and in adhara shruthi phonation for group II. The results of one way ANOVA indicated no statistically significant difference between the three age groups in both habitual phonation and adhara shruthi [Habitual phonation- F(2,27)=0.

34, p >0. 05; Adhara shruthi- F(2,27)=0. 566, p>0. 05]. Figure 10 shows error bars for HNR in all the three groups in habitual phonation and adhara shruthi.

Figure 10: Error bars for HNR in three groups for habitual phonation and adhara shruthi. Discussion The purpose of the present investigation is to compare the voice parameters between the habitual phonation and adhara shruthi in trained carnatic classical singers, and also the effect of years of training was investigated. The results revealed several points of interest.

First, when compared to habitual phonation, the mean fundamental frequency for adhara shruthi was different in all the three groups. Both adults and geriatrics had lower mean fundamental frequency, (although significant only in geriatrics) where as children had higher mean fundamental frequency. The result of the present study supports Chandra (2001) study who reported similar findings.

The lower fundamental frequency in adhara shruthi in adults and geriatrics could be because in most of the ragas vocalists need to sing more in the higher range than in the lower range.

Even if they have to sing the lower notes it would be for a lesser extent compared to frequency and duration of higher notes. Children in contrast adopted slightly higher adhara shruthi. This indicates that there is an influence of years of singing experience on the pitch selection for the adhara shruthi. Second, for all the three groups in all the other voice parameters except alpha ratio i. e.

frequency range, mean intensity, intensity range, jitter (%), shimmer, beta ratio, gamma ratio and HNR, the adhara shruthi values did not vary significantly from habitual phonation.

For alpha ratio adults had significantly lower values in adhara shruthi when compared to habitual phonation. Alpha ratio indicates the relative energy concentration between 0 to 1 kHz and 1 to 5 kHz. Lower alpha values indicate higher energy level in the 1 to 5 kHz region which is generally seen in trained singers. Third when the voice parameters were compared between the three groups separately for habitual phonation and adhara shruthi there was no clear effect of years of singing experience.

Only the mean fundamental frequency values and the gamma ratio values were significantly different between children and geriatrics. The mean fundamental frequency values in adhara shruthi were significantly lower in geriatrics when compared to children. This can be attributed to the effect of years of singing experience. As the singers become aware of the tonal structure of the ragas and the range they have to transverse up and down on the scale they may do necessary readjustment of the adhara shruthi level. By keeping lower level adhara shruthi there may be facilitation for singers to intone higher notes.

Another reason why a female vocalist may select a lower adhara shruthi could be that classical music demands a well projected loud voice. It may be the perception of the female vocalist that she may achieve this by selecting a lower adhara shruthi than her habitual fundamental frequency. The gamma ratio values were significantly different between children and geriatrics Gamma ratio indicates the relative energy concentration between 0 to 1 kHz and 5 to 8 kHz. Lower gamma values indicate higher energy level in the 5 to 8 kHz region which is generally seen in trained singers.

However these explanations are highly hypothetical in nature and there is no direct evidence for the reasoning put forward.

Conclusion The present study attempts at providing the empirical data on the carnatic classical music, aimed at studying the relationship between the habitual phonation and the adhara shruthi in singers who are practitioners of carnatic classical music. Also the effect of years of training or the chronological age was investigated. The results revealed that the adults and geriatrics adopt slightly lower pitch for adhara shruthi and they had better energy concentration in the high frequency region.

The results of the present study have clinical relevance as they provide a basis for the proper vocal education for the singers. As the results are preliminary in nature future studies needs to be done to substantiate the present findings.

Reference Arushi, C. (2001). Relationship between the fundamental frequency of voice and ‘adhara shruthi’ in carnatic vocal music. Unpublished Master’s Dissertation, University of Mysore. Luchsinger, R.

& Arnold, G. E (1965). Voice speech language. California: Wadworth Sambamurthy, P. (1982). History of Indian music”.

Madras: Indian music publisher Sundberg, J. , Prame, E. , & Iwarsson, J. (1995). Replicability and accuracy of pitch patterns in professional singers.

STL- Quarterly Progress Status Report, 2-3, 51-62. Titze. I. (1989). Regulation of vocal power and efficiency by subglottal pressure and glottal width.

In O. Fujimura (Ed. ) Vocal Physiology: Voice Production, Mechamisms and Functions. Pp 227 – 237. New York: Raven Press.

Vennard, W. (1962). Singing mechanism and technique. Edward Brothers Inc. , Ann Arbor Michigraz.