ORIGINAL ARTICLE

Describing different styles of singing A comparison of a female singers oice source in Classical, Pop, Jazz and Blues
Margareta Thalen1 and Johan Sundberg2
SMI (University College of Music Education in Stockholm) and 2 KTH Voice Research Centre, Department of speech music hearing, KTH (Royal Inst of Technology), SE-100 44 Stockholm, Sweden
1

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Log Phon Vocol 2001; 26: 82 93 The voice is apparently used in quite different manners in different styles of singing. Some of these differences concern the voice source, which varies considerably with loudness, pitch, and mode of phonation. We attempt to describe voice source differences between Classical, Pop, Jazz and Blues styles of singing as produced in a triad melody pattern by a professional female singer in soft, middle and loud phonation. An expert panel was asked to identify these triads as examples of either Classical, Pop, Jazz or Blues. The voice source was analysed by inverse ltering. Subglottal pressure Ps , closed quotient QC lo sed , glottal compliance (ratio between the air volume contained in a voice pulse and Ps ), and the level difference between the two lowest source spectrum partials were analysed in the styles and in four modes of phonation: breathy, ow, neutral, and pressed. The same expert panel rated the degree of pressedness in the entire material. Averages across pitch were calculated for each mode and style and related to their total range of variation in the subject. The glottogram data showed a high correlation with the ratings of pressedness. Based on these correlations a pressedness factor was computed from the glottogram data. A phonation map was constructed with the axes representing mean adduction factor and mean Ps , respectively. In this map Classical was similar to ow phonation, Pop and Jazz to neutral and ow phonation, and Blues to pressed phonation. Key words: voice source, singing, subglottal pressure. Professor Johan Sundberg, KTH Voice Research Centre, Department of speech music hearing, KTH (Royal Inst of Technology), SE-100 44 Stockholm, Sweden. Tel: 468 790 7873. Fax: 468 790 7854. E-mail: pjohan@speech.kth.se

INTRODUCTION Important characteristics of vocal sound emanate from the mode of vocal fold vibration, which, in turn, is controlled by laryngeal muscle adjustment combined with the subglottal pressure Ps , i.e., the air pressure in the trachea under the glottis. The vibrations generate a pulsating air ow through the glottis, i.e., the voice source, which is an acoustic signal that determine essential voice characteristics. Therefore, the voice source is highly relevant to descriptions of different types of voice use. The voice source can be varied considerably with regard to pitch, vocal loudness and register. Also, by varying the degree of glottal adduction, a singer can change his her mode of phonation within a wide range. This possibility is used in music. Many country singers sing with a sound quality that appears
2001 Taylor & Francis. ISSN 1401-5439

quite hyperfunctiona l or pressed, especially in loud voice at high pitches (2, 7). Likewise, belting has been found to be associated with a high degree of glottal adduction (8, 1). Operatic singers, on the other hand, usually tend to stay within the so-called ow phonation, which is characterised by the lowest degree of glottal adduction that still produces glottal closure (11). Pedagogy in different vocal styles would pro t from explicit knowledge about the respective characteristics of the various phonation modes used. Research on the singing voice has mostly focused on Classical singing, while voice use in styles like, for example, Jazz, Blues and Pop singing has been largely neglected. The aim of this investigation was to describe voice source characteristics used by a performer and singing teacher in Classical operatic, Blues, Pop, and
Log Phon Vocol 26

Voice source in Classical, Pop, Jazz and Blues Jazz styles. Our description will not be limited to comparing such characteristics between styles. Rather, we will relate each voice source property to this particular propertys entire variation range that seemed available to the subject. Thus, the voice source characteristics of the different styles will be described by showing how they relate to the subjects total voice source variability.

83

METHOD The subject (co-author MT), has a long professional experience of singing and teaching in the four different styles: Classical, Pop, Jazz and Blues. Originally trained in the Classical style of singing she performed professionally as a soloist, mostly in the Jazz, Pop and Blues styles for about 10 years. In addition, she has worked as a voice teacher in these styles for more than 20 years. The subject sang, in one breath group, the syllable [p] twice on each of the pitches of a seventh chord (A3, C 4, E4, G4), a pattern chosen because of its frequent use particularly in the Pop, Jazz and Blues styles. She also sang the same pattern in four modes of phonation: pressed phonation and breathy phonation, as well as in neutral and ow phonation. Pressed and breathy phonation correspond to hyperand hypofunctional phonation, respectively. Neutral phonation is the type of phonation that is typically used in neutral speech. Flow phonation is produced with the lowest degree of glottal adduction that produces a vocal fold closure. It is typically used in Classical singing and seems similar to or identical with what has been referred to as resonant voice (13). The subject also sang this material in four styles: Classical, Pop, Jazz, and Blues. The Classical style of singing was that typically used in performances of the Lieder repertoire. The Pop and Jazz styles corre-

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

sponded to those typically used in Pop and Jazz ballads. Typical representatives of the Pop style are performers like Randy Crawford and Whitney Houston. Typical representative s of the Jazz style are performers like Billie Holiday and Sarah Vaughan. Typical representatives of the Blues style are Bessie Smith and Janis Joplin. The subject sang the pitch pattern in the following order: breathy, ow, neutral, pressed, Classical, Pop, Jazz, Blues starting with low, then medium and last high degree of vocal loudness. The equipment used in the experiment is shown in Fig. 1. The ow signal was obtained from a ow mask (Glottal Enterprises, MSIF2). The audio signal was recorded by a high delity microphone at a distance of 0.3 m from the mouth. The larynx height was registered by means of a two-channel electroglottograph (Glottal Enterprises, MS-F1). Ps , was recorded as the oral pressure during the occlusion of the consonant p in the syllable [pae:]. This pressure was measured with a pressure transducer connected to a small plastic tube, mounted in the ow mask, that the subject had in the corner of her mouth. All these signals were recorded on separate channels in a TEAC PCM recorder.

ANALYSIS Perceptual ealuation The recorded material was evaluated perceptually by a panel of 10 listeners, all experts on voice pedagogy and with acquaintance with singing in different singing styles. The task of the panel was (1) to classify the triads sung in different styles according to a forced choice design and (2) to assess the degree of phonatory pressedness in all triads sung in all styles and all modes. For the former task the subjects were asked to mark on an answering sheet either of the alternatives Classical, Pop, Jazz and Blues. For the assessment of phonatory pressedness the subjects were asked to rate on 100-mm long visual analogue scales where they thought the phonation was situated along a scale with breathy phonation and pressed phonation as its two extremes. The extremes of the lines were marked Extremely breathy and Extremely pressed on the answering sheet. These terms were familiar to the listeners. Each of the panel members received a copy of an audiotape with two parts. In the rst each of the 12 triads sung in different styles occurred two times in randomized order (4 styles 3 loudnesses 2 presentations). In the second part, each of the triads occurred two times (4 phonation modes 3 loudnesses 2 presentations plus 4 styles 3 loudnesses 2 presentations) . Two of the subjects felt
Log Phon Vocol 26

Fig. 1. Block diagram of the equipment used during the recordings.

84

M. Thalen and J. Sundberg

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Fig. 2. Expert panels mean ratings of pressedness for the various phonation modes and styles. The bars represent 91 standard deviation. that they could not perform the classi cation task while all subjects made the second part of the test. Table 1 shows the results from the classi cation test. The classi cation of replicated stimuli shows that the singers examples of Classical and Blues were most readily recognised in all loudnesses and that the distinction between Pop and Jazz seemed less clear. Also, the mid and loud examples of Jazz were often classi ed as Blues. The mean correlation for the responses to repeated stimuli in the second part of the test, where the panel rated degree of phonatory pressedness, was 0.794 (SD 0.295). All subjects but 8 produced a correlation greater than 0.56, see Table 2. The answers of subject 8 were therefore discarded. Figure 2 shows the panels mean rated pressedness for the various triads. For both modes and styles the rated pressedness was lowest for the soft triads and

Table 1. Result from the eight subjects who answered the classi cation test. The column marked Diergent shows the number of times that a triad was classi ed differently the second time it occurred in the test tape
Number of responses Style Classical Pop Jazz Blues Soft Middle Loud Soft Middle Loud Soft Middle Loud Soft Middle Loud Classical 13 16 16 0 0 0 0 0 0 0 0 0 Pop 0 0 0 7 11 10 4 7 6 3 2 5 Jazz 3 0 0 7 4 2 10 2 1 0 0 0 Blues 0 0 0 2 1 4 2 7 9 13 14 11 Divergent 1 0 0 1 4 4 2 5 4 1 0 1

Table 2. Intra -rater consistency for ratings of phonatory press according to the Pearson correlation coef cient r. The lines marked Slope and Itcpt show the slope and intercept of the best linear t to the data
Subject Slope Itcpt R 1 0.98 2.3 0.974 2 1.13 11.9 0.969 3 1.00 4.2 0.932 4 0.62 22.0 0.669 5 0.51 22.8 0.562 6 0.91 2.6 0.877 7 1.00 4.3 0.961 8 0.26 40.6 0.216 9 0.98 4.4 0.855 10 0.92 2.68 0.926

Log Phon Vocol 26

Voice source in Classical, Pop, Jazz and Blues

85

Fig. 3. Block diagram of the equipment used during the analysis. highest for loud triads except for the pressed mode and the Blues style, where the differences were small and showed a weak opposite trend. For the modes the results show the lowest degree of rated pressedness for breathy and the highest for pressed, as expected. For the styles, Classical showed the lowest and Blues the highest degree of pressedness. Acoustic analysis A block-scheme of the analysis equipment is shown in Fig. 3. The voice source waveform was derived by inverse ltering the ow signal (Glottal Enterprises, MSIF2). The lowest pitch in each seventh chord was captured on a transient recorder (Glottal Enterprises, BT-1) and the frequencies and bandwidths of the two lowest formants were adjusted so as to minimise the ripple during the closed phase. Then, all pitches of the seventh chord were run through the inverse lter and stored in a le using the Swell sound signal

workstation (12). Ps was simultaneously recorded on another channel of the same le. For each note in the triad a representative waveform was selected from the middle part of the second syllable, Fig. 4. The glottogram characteristics as well as Ps and the larynx height were then measured using the Swell sound signal workstation (12). As illustrated in Fig. 5a, the following ow glottogram parameters were measured: the period time between two adjacent discontinuities corresponding to the closing of the glottis, the closed phase between the end of a ow pulse to the onset of the next ow pulse, the peak-to-peak pulse amplitude, de ned as the peak value of the pulse, and the air volume contained in the pulse, estimated from a triangle approximation of the pulse. In addition, Ps was determined from the [p] preceding the syllable analysed; as the subject kept vocal loudness constant during the syllable analysed this value of Ps was assumed to be representative of the entire subsequent syllable. Larynx height was measured as the mean during the syllable. The level difference between the rst and second voice source spectrum partials, henceforth H1 H2, was measured by means of a narrow bandwidth FFT analysis of the ow glottogram analysed, Fig. 5b. From these data were calculated the closed quotient and the glottal compliance, de ned as the ratio between the air volume contained in the glottal pulse and Ps . This ratio should be related to the degree of glottal adduction; the greater the adduction, the lower the compliance. Thus, two physiological measures were studied, Ps and larynx height, and three measures related to glottal adduction, closed quotient (QC lo sed ), glottal compliance, and H1 H2. Average air ow during the closed phase, re ecting glottal leakage, could not be measured due to low-frequency disturbances in the ow signal. The larynx height data showed the highest larynx position for Classical style. This highly unexpected result was interpreted as an artifact; according to observations reported by Laukkanen and co-workers (5), this artifact can be expected if the larynx moves in the anterior-posterior direction. An anterior movement of the larynx is likely in Classical singing, as it can be assumed to be associated with a widening of the pharynx. As it seemed risky to assume that none of the other styles involved a variation of pharyngeal width, the larynx height data for all modes and styles were discarded. RESULTS

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Fig. 4. Example of sound le. The upper and lower traces show oral pressure and ow, respectively.

The inverse lters were reasonably simple to tune, particularly for the lower F0. For the pitch of G4,
Log Phon Vocol 26

86

M. Thalen and J. Sundberg Fig. 5. Flow glottogram parameters analysed: 1) period time, 2) duration of closed phase, 3) peak ow amplitude of the pulse (AC ow), 4) mean ow during the closed phase (DC ow), and 5) air volume contained in the pulse, estimated from a triangle approximation of the ow pulse. The right panel shows the spectrum of the ow glottogram from which the level difference between the two lowest partials H1 and H2 of the source spectrum was determined.

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

however, it was dif cult to obtain a ripple-free closed phase in many cases, the ow glottogram often assuming shapes similar to the one illustrated in Fig. 6. It may be relevant that at this F0 the rst formant was close to the second partial. Also, this pitch is close to the register transition in this voice. Consequently, all glottograms pertaining to this pitch were discarded. Figure 7a d shows how Ps , QC lo sed , compliance and H1 H2 varied with pitch in soft, middle and loud in breathy, ow, neutral and pressed. Ps tended to increase with loudness and also with pitch, as expected. Pressed phonation showed clearly higher values than the remaining modes. QC lo sed was high in the pressed mode and low or zero in breathy phonation. Unexpectedly, soft, middle and loud did not differ systematically. The compliance decreased with pitch. Pressed showed low values and breathy high, as expected. The H1 H2 showed low values for pressed and higher values for the remaining modes. To obtain a gross idea of the characteristics of the four vocal styles, the various glottogram measures were related to the subjects variation range of the respective measure. This was realized in the following way. For each measure, the highest and lowest values were identi ed at each pitch and for each degree of vocal loudness. Figure 8 shows these extreme values for Ps , QC lo s ed , compliance and H1 H2. Then, observed values of this measure for the various styles and phonation types were expressed as percentages of this variation range. For example, the lowest and highest Ps values observed when the subject sang the pitch of E4 in middle loudness were 7.9 and 13.1 cm H 2 O. At that pitch and loudness she used a Ps of 9.2
Log Phon Vocol 26

cm H2 O in Classical. This value corresponds to a percentage of 100 (9.2 7.9) (13.1 7.9) 25% of her pressure variation range at that pitch and degree of vocal loudness. After converting all values of the glottogram measures in this way, means across pitch and loudness were calculated for each style and phonation mode. These means re ect how the subject used her voice in the various styles and modes. Figure 9 shows these means. Each mode and each style show a particular characteristic combination of glottogram measures. Ps was less than 20% of the subjects total range in breathy, ow, and neutral and approaches 90% in pressed. QC lo sed of pressed and breathy assumed extreme values close to 100% and 0%, respectively. Compliance shows a mirror image of QC lo sed . H1 H2 was low for pressed and gradually increased from neutral to ow to breathy. For the styles, P s was more than 90% for Blues, and gradually decreased for Pop, Jazz and Classical. QC lo sed was more than 70% for Blues and varied between about 30% and 60%

Fig. 6. Typical example of inverse lter output waveform obtained from the pitch of G4.

Voice source in Classical, Pop, Jazz and Blues

87

Fig. 7. a d. Ps , QC lo sed , compliance, and H1 H2 as function of pitch in soft, middle and loud (left, middle and right panels) in pressed, neutral, ow and breathy phonation. QC lo sed in soft, breathy phonation was zero.

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

between the other styles. Compliance and H1 H2 were low in Blues. Compliance was highest in Classical. H1 H2 was highest in Pop and lowest in Blues and higher in Classical than in Jazz. As the recordings were analysed both perceptually and acoustically, the relationship between glottogram characteristics and perceptual data can be examined. Figure 10a, b, and c shows the relationships between mean rated pressedness and the normalised values of QC lo s ed , H1 H2, and compliance, respectively. The correlation amounted to 0.838, 0.813, and 0.869 for QC lo sed , H1 H2, and compliance, respectively. Thus, a good deal of the variability in perceived pressedness could be explained by these ow glottogram parameters. The data shown in Fig. 10 were used to construct a compound measure of perceived degree of pressedness. Using the solver function in the Microsoft Excel

program package the constants for QC lo sed , H1 H2, and compliance were varied such that the correlation with mean pressedness ratings was maximised. Thereby, the values of H1 H2 and the compliance were subtracted from 1.0, so that all three terms increased with increasing Ps . The relationship between mean rated pressedness and the glottogram measures can be described as MVP ressN 1.65 PsN 0.19 QC lo sed N 2.16 (1 Compl N ) .0001 [1 (H1 H2) N ] where MVP ressN is the normalised mean rated degree of pressedness, and PsN , QC lo sed N , ComplN , and (H1 H2) N are the normalised values of P, closed quotient, compliance and the dominance of the fundamental, respectively. The equation shows that the
Log Phon Vocol 26

88

M. Thalen and J. Sundberg

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Fig. 8. Extreme values (Max and Min) for Ps , QC lo s ed , compliance and H1 H2 observed at the three degrees of vocal loudness ff (loud), mf (middle), and pp (soft).

Fig. 9. Ps and glottogram characteristics of the phonatory modes and vocal styles (left and right panel). Each data point is an average across pitch of the parameter values expressed as percentages of the total variation range at the respective pitch and degree of vocal loudness.
Log Phon Vocol 26

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Voice source in Classical, Pop, Jazz and Blues

Fig. 10. a, b and c. Relationships between mean rated pressedness and QC lo sed , compliance, and H1 H2 (left, middle and right panel, respectively). Each data point represents the normalised mean across pitch of the parameter values expressed as percentages of the total variation range at the respective pitch and degree of vocal loudness. The equations and the lines show the best linear t of the data points. R2 shows the squared correlation.

Log Phon Vocol 26

89

90

M. Thalen and J. Sundberg words, the degree of perceived pressedness could be estimated from the glottogram data. As pressedness can be assumed to be closely related to the degree of glottal adduction, the physiological characteristics of the various styles and phonation modes can be described as positions in a plane of phonation map, where the axes represent Ps and the press factor. Thus, the normalised values of Ps , averaged across pitch for each style and phonation mode were plotted against the normalised press factor, as calculated from the glottogram data by means of the equation given above. The result is shown in Fig. 12. The data points are distributed over the phonation map in an expected manner. Pressed and breathy assume opposite extreme positions along the press factor and ow phonation is located between neutral and breathy. Blues is close to pressed, though slightly less pressed, and Classical is close to ow and neutral phonation. Jazz and Pop assume mid positions in both press factor and Ps . DISCUSSION The results concerning the characteristics for the different phonation modes seem convincing on the whole. In breathy phonation Ps and QC lo sed were low while compliance and H1 H2 were high. In ow phonation, Ps was low, QC lo sed , intermediate, and compliance and H1 H2 high. In neutral phonation, Ps was low, QC lo sed moderate, and compliance and H1 H2 rather high. In pressed phonation, Ps and QC lo sed were high compliance and H1 H2 low. All these data are in good agreement with those observed in previous investigations. Along with the results of the perceptual evaluation these ndings show that the subject produced representative samples of the different modes of phonation. According to previous investigations, the Classical singing style is similar to ow phonation (11). This similarity was illustrated by the phonation map in Fig. 12 where Classical was located between ow and neutral. Likewise, the perceptual evaluation indicated a high degree of similarity between Classical and ow phonation. This indicates that the subject produced typical examples of Classical singing. Inverse ltering becomes unreliable when the fundamental frequency is higher than at least half of the rst formant frequency. For this reason the vowel ae: was chosen and pitches only up to E4 (330 Hz) were analysed. It is possible that the voice source differences between the styles are greater at higher pitches; for pitches above E4 some styles of singing would use chest register while in the Classical style middle register is used.

contribution of the H1 H2 was negligible. Henceforth, this compound will be referred to as the normalised press factor. The compound was normalised such that it varied between 0 and 1. Figure 11 shows the relationship between normalised press factor and the mean rated degree of pressedness. The very high correlation of 0.936 implies that for this material as sung by this voice, 88% of the variation in perceived pressedness could be explained by the variation of glottogram characteristics. In other

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

Fig. 11. Relationship between mean rated pressedness and the normalised press factor as de ned in the equation. The equation and the line show the best linear t of the data points. R2 is the squared correlation.

Fig. 12. Phonation map showing mean normalised Ps versus mean normalised press factor for the various styles and modes. The locations of the style and phonation mode terms represent the Ps and press factor averages, computed across pitches and degree of vocal loudness.
Log Phon Vocol 26

Voice source in Classical, Pop, Jazz and Blues QC lo sed has been found to increase with increasing Ps (6). No such variation was found in the present data. The reason for this may be related to the fact that the pitch range studied was close to the subjects transition from chest to middle register; in this transition QC lo sed tends to decrease (10). The main idea of this investigation was to relate different singing styles, as produced by a professional singer, to the same subjects phonation modes. Thus, all comparisons were made within the same voice. This strategy is advantageous , since the voice source is dependent on the individual characteristics of the vocal folds. The method, however, requires that the subject can master different phonation modes. An alternative approach would be to study voice source characteristics in different groups of singers, each representing a certain singing style, and then to compare the results between groups. The results from such an approach can be regarded as statistical, in the sense that it would reveal how a majority of representatives use their voices when singing in a particular style. In such a study the variations across individuals with respect to phonatory potentials would contribute to the scatter of the results. The method tested in the present study is more analytic, as it relates the phonatory characteristics of various singing styles to the particular expert subjects phonatory range. Our approach seems relevant to vocal pedagogy. In the singing teachers studio, the task is to help the student nd the way to healthy voice use for a particular style of singing within his her phonatory range. The description of the similarities of the different styles with phonatory modes should be helpful in that situation. For example, it may be relevant that a typical Blues style is not as hyperfunctional as pressed phonation; pressed phonation is generally considered harmful to the voice. The normalised press factor for Pop and Jazz was similar to that for neutral. It can be argued that the lyrics in the Pop and Jazz ballad repertoires often seem to play a more prominent role than in the classical repertoire, at least in opera and oratorio arias from the Baroque era, where the text segments are generally repeated several times. Furthermore, in Pop and Jazz music female soloists tend to select a key that allows them to use the lower part of their pitch range, i.e., a more speech-like range than that used in operatic singing. It may be appropriate that the voice used in the Pop and Jazz repertoire is similar to that used in speech. The phonatory characteristics in Blues singing, by contrast, were found to be similar to that of pressed phonation, characterised by high Ps , low compliance, high QC lo s ed and a weak fundamental, which is somewhat similar to what has been found in the country singing style (7). Also in

91

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

this case the typical musical style may be relevant; agitated emotional moods occur frequently in these repertoires. Thus, in some respects our study lends support to the assumption that vocal aesthetics is related to musical functionality. In computing the press factor, the H1 H2 measure added negligibly to the correlation with the mean ratings for pressedness. A greater signi cance of the H1 H2 may occur for a material comprising also phonation in different vocal registers. H1 H2 has been found to be strongly correlated with QC lo s ed in the male modal register but not in falsetto register (9). The phonation map was based on the high correlation found between glottogram characteristics and mean rated degree of pressedness. It would be interesting to nd out if a similar correlation can be found also for other voices. Anyhow, the phonation map seems a useful tool for describing phonatory habits in a given pitch range. In cases where pitch is relevant, it may be worthwhile to complement the phonetogram with a third dimension representing the press factor (4). CONCLUSIONS The basic idea of this investigation was to describe the voice source characteristics in different styles of singing by relating them to the phonatory potentials of a professional singer. These potentials were determined in terms of four phonation modes related to different degrees of glottal adduction: breathy, ow, neutral, and pressed. The subjects examples of these phonation modes were mostly identi ed correctly by the expert panel and were thus representative. Also, her examples of the various singing styles were found to be similar to different modes. This suggests that the strategy of the investigation was successful. The results revealed that her voice source in Classical style was close to ow and neutral phonation, while in the Pop and Jazz styles her voice source was more similar to her neutral phonation, although Ps was higher. Her Blues style showed the greatest similarity with pressed phonation.

ACKNOWLEDGEMENTS This investigation was carried out as the rst authors Research and Development in the Arts at the SMI (University College of Music Education in Stockholm). The inverse ltering and the glottogram measurements were made by Jenny Iwarsson. The authors pro ted from advice from Ninni Elliot in elaborating the experimental design. The project was supported by SMI.
Log Phon Vocol 26

92

M. Thalen and J. Sundberg Subglottala trycket uppskattades fran oraltrycket un der tillslutningen for konsonanten [p]. En expertpanel klassi cerade fyrklangerna som endera klassisk, pop-, jazz- eller bluestil och angav ocksa graden av fona torisk press pa visuella analogskalor. Exemplen pa klassisk och bluesstil klassi cerades mestadels som sangerskan avsett, medan en del pop- och jazzexempel forvaxlades. Flodessignalen invers ltrerades och slutenkvot, grundtonshal t och glottal komplians (kvoten mellan luftvolymen i en rostpuls och det underliggande subglottala trycket) bestamdes. Dessa matt relaterades till det totala variationsomradet for respektive tonhojd och styrkenyans, och medelvarden beraknades av vart matt for varje styrkenyans, stil och fonationssatt. Resultatet visade att sangerskan an vande de hogsta subglottala trycken i blues och pressad fonation och de lagsta i lackande fonation. Medelvar dena korrelerade med genomsnittliga graden av uppfattad fonatorisk press. Pa basis av detta resultat beraknades for var stil och vart fonationssatt en pressfaktor de nierad som en vagd summering av genomsnittlig normerad slutenkvot, komplians och grundtonshalt . Denna pressfaktor gav korrelationen 0,936 med medelvardena av skattningen av fonatorisk press. En fonationskarta konstruerades dar axlarna representerade genomsnittligt subglottalt tryck och genomsnittlig pressfaktor. Fonationskartan visade att bluesstilen lag na pressad fonation men anvande ra lagre subglottalt tryck, att odig fonation lag mellan lackande och neutral, och att pop och jazz hade hogre pressfaktor men ungefar samma subglottala tryck som neutral fonation. YHTEENVETO Eri laulutyylien kuailu. Naislaulajan aanen ertailua klassisessa -, pop -, jazz- ja blueslaulussa ytetaan eri tavalla eri laulutyyleissa ja on Aanta ka selvaa, etta aanen sointi silloin vaihtelee. Jotkut naista eroavaisuuksist a aiheutuvat varmasti aanilah teen eroavaisuuksista , koska aanilahteen muoto vaihtelee huomattavasti johtuen joko aanen voimakkuudesta , -korkeudesta ja fonaatiotavasta . Ammattilaulajata r ja laulupedagogi, jolla on laaja kokemus klassisen, jazz-, pop- ja bluestyylin laulamisesta ja opettamisesta, lauloi tavusta [pae] edella mainituilla tyyleilla nousevaa melodiaa save lesta A3 heikolla, normaalilla ja voimakkaall a aanella. Niiden lisaksi ha lauloi saman materiaalin n neljalla eri fonaatiotavalla : vuotavalla, runsaalla, neutraalilla ja puristeisella aanella. Virtaussignaali rekisteroitiin Rothenbergin maskilla. Subglottaalista painetta arvioitiin oraalipaineesta konsonantin [p] sulkuvaiheesta. Asiantuntijaraat i luokitteli eri soin-

REFERENCES
1. Bestebreurtje M, Schutte H. Resonance strategies for the belting style: Results of a single female subject study, J Voice 2000; 14: 194 204. 2. Cleveland T, Stone RE, Sundberg J, Iwarsson J. Estimated subglottal pressure in six professional country singers. J Voice 1997; 11: 403 9. 3. Gauf n J, Sundberg J. Spectral correlates of glottal voice source waveform characteristics. J Speech Hear Res 1989; 32: 556 65. 4. Gramming P, Sundberg J, Ternstro S, Leanderson R, m Perkins W. Relationships between changes in voice pitch and loudness. J Voice 1988; 2: 118 26. 5. Laukkanen AM, Takalao R, Vilkman E, Nummenranta J, Lipponen T. Simultaneous video uorographic and dual-channel electroglottographic registration of the vertical laryngeal position in various phonatory tasks. J Voice 1999; 13: 60 71. 6. Sundberg J, Andersson M, Hultqvist C. Effects of subglottal pressure variation on professional baritone singers voice sources. J Acoust Soc Am 1999; 105: 1965 71. 7. Sundberg J, Cleveland T, Stone RE, Iwarsson J. Voice source characteristics in six premier country singers. J Voice 1999; 13: 168 83. 8. Sundberg J, Gramming P, Lovetri J. Voice source, formant frequency and subglottal pressure characteristics in a professional female musical theatre singer. J Voice 1993; 7: 301 10. 9. Sundberg J, Hogset C. Voice source differences between falsetto and modal registers in counter tenor, tenor, and baritone singers. Logopedics Phoniatrics Vocology 2001; 26: 26 36. 10. Sundberg J, Kullberg A. Voice source studies of register differences in untrained female singing. Logopedics Phoniatrics Vocology 1999; 24: 76 83. 11. Sundberg J. The Science of the Singing Voice. DeKalb: N. Ill. Univ. Press, 1987. 12. Ternstrom S. Soundswell signal workstation software manual. 1992. 13. Verdolini K, Druker DG, Palmer PM, Samawi H. Laryngeal adduction in resonant voice. J Voice 1998; 12: 315 27.

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

SAMMANFATTNING Att beskria olika sangstilar. En jamforelse a en sangerskas rostkalla i klassisk , pop , jazz och blues Rosten anvands pa olika satt i olika repertoarer, vilket leder till olika rostklang. Nagra av dessa olikheter kan ha ras till skillnader i rostkallan, som normalt nfo varierar avsevart beroende pa roststyrka, tonhojd och fonationssatt. En professionell sangerska och sangpedgagog med stor erfarenhet av att sjunga och undervisa i klassisk, pop-, jazz- och bluesstil sjong pa stavelsen [p] en stigande fyrklangsmelodi fran tonen A3 i dessa stilar med svag, mellanstark och stark rost. Dartill sjong hon ocksa samma material med fyra fonationssatt: lackande, odig, neutral och pressad. Flodessignalen registrerades med en Rothenbergmask .
Log Phon Vocol 26

Voice source in Classical, Pop, Jazz and Blues nit joko klassiseen-, pop-, jazz- tai bluestyyliin kuuluvaksi ja antoi myos arvion fonaation puristeisuudesta analogisella asteikolla. Esimerkit klassisesta ja bluestyylista luokiteltiin enimmakseen kuten laulaja oli tarkoittanutkin , mutta osa pop- ja jazztulkinnoista sekoittui. Virtaussignaali kaanteis suodatettiin ja sulkuosamaara perustaajuus ja glot , taalinen joustavuus (ilman volyymin ja subglottaalisen paineen suhde) paateltiin. Na lukuja ita verrattiin kyseessa olleisiin aanen korkeuden ja aanen voimakkuuden vaihtelualueisiin. Keskiarvo laskettiin jokaisesta mitasta voimakkuudelle, tyylille ja fonaatiotavalle . Tulokset osoittivat, etta laula jatar kaytti korkeimpia subglottaalisi a paineita blueslaulannassa . Vuotoiseen tuottotapaa n puolestaan osoittautui liittyvan puristeinen fonaatio ja matalimmat sa velkorkeudet. Keskiarvot korreloivat

93

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by University of Athens on 09/19/11 For personal use only.

fonatoriseksi paineeksi tulkitun ma ran kanssa. a Naiden tulosten pohjalta laskettiin jokaiselle laulu tyylille ja fonaatiotavall e oma painefaktori , joka maariteltiin summaamalla keskimaaraisen loppuosa maara joustavuus ja perustaajuus. Tama painefak , tori antoi korrelaation 0,936 keskiarvoksi fonatorisen paineen arvioinnille. Tama n jalkeen muodostettiin fonaatiokartta , jossa akselit edustavat keskimaaraista subglottaalista painetta ja keski maara painefaktoria. Fonaatiokartta na ista ytti, etta bluestyyli sijoittuu lahelle puristeista fonaatiota, mutta ka ytti alhaisempaa subglottaalista painetta. Kartasta ilmeni myos, etta runsas fonaatio sijoittui vuotavan ja neutraalin fonaation va limaastoon ja etta pop- ja jazzlaululla oli korkeampi painefaktori mutta lahes sama subglottaaline n paine kuin neu traalilla fonaatiolla.

Log Phon Vocol 26