Voice source effects of diaphragmatic activity in singing*

Voice source effects of diaphragmatic activity in singing*

Journal of Phonetics (1986) 14, 351-357 Session 1 Voice source effects of diaphragmatic activity in singing* Johan Sundberg Department of Speech Com...

3MB Sizes 0 Downloads 61 Views

Journal of Phonetics (1986) 14, 351-357

Session 1

Voice source effects of diaphragmatic activity in singing* Johan Sundberg Department of Speech Communication and Music Acoustics, Royal Institute of Technology, Stockholm, Sweden

Rolf Leandersont and Curt von Eulert t Phoniatric Department and tNeurophysiological Department, The Karolinska Institute, Stockholm , Sweden

The transdiaphragmatic pressure was displayed on an oscilloscope screen as a visual feedback signal for singers and non-singers who performed various phonatory tasks with and without voluntary co-activation of the diaphragm. the diaphragmatic co-activation tended to increase the amplitude of the flow glottogram, i.e., the transglottal air volume velocity waveform as determined by inverse filtering and also tended to reduce the formant frequency variability under conditions of changing fundamental frequency.

1. Introduction

In voice therapy and training, breathing technique is generally regarded as crucial to optimal voice function . However, to what extent the mode of breathing has acoustic effects on the voice has never been scientifically investigated. The purpose of the present investigation was to find out if, and how, a difference in breathing technique affects phonation acoustically. In two previous investigations, the role of the diaphragm in singing was analysed in male singers (Leanderson, Sundberg, von Euler & Lagercrantz, 1984; Sundberg, Leanderson, von Euler & Lagercrantz, 1985). Clear inter-individual differences were found. During sustained tones, some singers had a constantly passive diaphragm, while other singers showed an active diaphragm during most of the utterance. Also, in sudden changes of subglottal pressure for the purpose of pitch or intensity shifts, differing strategies in the usage of the diaphragm were observed. These differences were accompanied by voice timbre differences. In normal automatic breathing for metabolic purposes, an activation of the diaphragm in inspiration is associated with vocal fold abduction. Even though it cannot be assumed that this *This is a revised version of a paper published in Transcripts of the 13th Symposium: Care of the Professional Voice, New York , 1985. 0095-4470/86/030351

+

07 $03 .00/0

©

1986 Academic Press Inc. (London) Ltd.

352

J. Sundberg , Rolf Leanderson and Curt von Euler

association holds also during phonatory expiration (von Euler, 1982; 1983), the relationship between diaphragmatic activity and phonation seemed an interesting objective to study. 2. Experiment

In order to achieve a certain voice quality, trained singers have learned specific phonatory behaviors. Therefore, it was important to examine not only trained subjects. Two female and four male subjects, all of whom lacked singing education, were used along with two trained singers. The diaphragm separates the thorax from the abdomen . A diaphragmatic activation decreases the intrathoracic pressure and exerts a positive pressure on the abdominal content so that a transdiaphragmatic pressure is generated. If there is no transdiaphragmatic pressure, the diaphragm is passive. Thus, by measuring the pressure above and below the diaphragm, an indication of diaphragmatic activity can be obtained (see, for example, Newsome Davis, Goldman, Loh & Casson, 1970). This holds except for very low lung volumes, when the diaphragm is passively stretched to the extent that a significant tension develops. The oesophageal and gastric pressures were measured by means of a catheter ("Gaeltee" , 2 mm outer diameter) furnished with two pressure transducers, one at the tip and one about 15 em from the tip. The catheter was introduced through the nose and swallowed so that the tip was located in the gastric ventricle and the other transducer in Diaphragm passive

SPL O.Sm (dB)

0

Ss

Figure I. Example of recordings made during the experiment, when the subject was singing a triad pattern with interspersed /p/ . SPL, Sound pressure level at 0.5 m; Pd;, transdiaphragmatic pressure; P0 , , oesophagea l pressure; P£", gastric pressure; Poe oral pressure during the p-occlusion; F 0 , fundamental frequency.

Voice source effects

353

the oesophagus. The pressure difference, i.e. the transdiaphragmatic pressure, was recorded directly from a differential amplifier. The oral pressure during /pi-occlusion was recorded from a pressure transducer with a thin plastic catheter, which the subject held in the mouth corner. The oral output was recorded by means of a pressure gradient microphone mounted in a mask (see Rothenberg, 1973). From this same signal, fundamental frequency as well as sound pressure level (SPL) were determined. During the experiments, all these signals, except for the sound level, were simultaneously recorded on a multichannel rectilinearly writing pen recorder. The transdiaphragmatic pressure was displayed on an oscilloscope screen as a visual feedback signal for the subject. Under these conditions no subject had any difficulty in bringing his diaphragm activity under conscious control. The non-singers sang gliding and sustained tones with an active and a passive diaphragm, and the singers also sang triads, octaves and tones with different intensity. Figure I gives examples of the records obtained. The general impression was that the subjects' attention was captured by the curve on the oscilloscope screen rather than by the timbre of the voice sounds. The singers tended to phonate in a particular way under conditions of diaphragmatic co-activation, the timbre sounding more " dark" or covered. However, one of the singers discovered that he could produce both a " bright" and a " dark" tone with as well as without diaphragmatic activity. 3. Analysis

The signal recorded from the mask microphone was inverse filtered by means of a computer program. This procedure implies that the frequencies of the formants are determined and the influence of these formants on the voice source signal is compensated for. Thus, partials which have been emphasized by the proximity of a formant are suppressed to a corresponding degree, and vice versa. As the end result, graphs are obtained which show the transglottal airflow versus time, the so-called flow glottograms. Figure 2 shows a typical example. From the flow glottogram the duration of the moment of glottal closure, or the closed phase, can be determined . Other relevant data in a flow glottogram is the peak-to-peak

3: 0

Open phase

Closed phase

Time~

Figure 2. Idealized flow glo ttogram.

354

J. Sundberg , Rolf Leanderson and Curt von Euler

amplitude and the closing time, which are parameters of particular acoustic significance; the peak amplitude determines the amplitude of the voice source fundamental; the maximum closing rate (or, more specifically, the peak amplitude of the differentiated glottogram) determines the amplitudes of the source spectrum overtones and, hence, the SPL of the vowel under normal conditions. The influence of diaphragmatic activity on these flow glottogram data was examined. 4. Results

The main question was to find out to what extent the values observed under conditions of a passive and an active diaphragm agree. Thus, comparisons were needed of pairs of values collected under identical conditions except for the activation of the diaphragm. Such comparisons are facilitated by plotting the data in correlograms. Figure 3 shows correlograms regarding pressure data for the various subjects. In this as in remaining figures, all paired values pertain to the same fundamental frequency . The oesophageal pressure data for the non-singers pertain to non-extreme lung volumes. If the diaphragmatic activity had no influence on these pressure data, all data points for the singers would be symmetrically scattered around the dashed line which represents a one-to-one relationship. This was not the case, as is shown in the figure; the pressures were higher when the diaphragm was active than when it was passive. Thus, the net subglottal pressure increased in spite of the pressure reduction resulting from the activated diaphragm, evidently since the contribution from the expiratory intercostal and abdominal muscles was greater. Figure 4 shows data pertaining to the duration of the closed phase collected from the gliding tones of the non-singers and from the triads and octaves of the singers. Among the non-singers, there was a trend that the closed phase became longer when the diaphragm was activated. This trend becomes clearer if only the high-pitched tones are considered. Thus, in this case the vocal folds remained in contact for a longer time. In the case of the two singers, on the other hand, no clear trend can be observed.

/

( b)

/

(a}

40

/

/

/

/

/

/

/

/

>

"'"'0

30

/

-

/ /

E ~

/

/ /

0. 0'

/

/

Q)

/

20

M

0

MPp

0.

0

L L E

10 r

/

N

r-

J J

E~OP

N

/

N

r-

I

10

s

1-

EL

I

I

I

20

R

/

J

E~~LO

fEo/ 00

~~J

M~

/

)'tl~ JJ

MN

R

/

/

..c

s

I

I

30

I

I

40

R / R

s

s

5

/ S /

I

/

s

/ / R R

I

0

I

10

I

I

20

I

I

30

Diaphragm active

Figure 3. Correlogram for the subglottal pressure data: pressures in em water column. Symbols refer to subjects: (a) non-singers' oesophageal pressure; (b) singers' oral pressure during p-occlusion.

40

Voice source effects

Q

(a)

60t--

355

/

High pitch

/

(b)

/

/

/

rQ)

-~

c

/

E

©

/

40t--

r-

0 M

30

/

N

®

©

M

/

/

rI

2020

40

30

/

I

60

50

/

/

I

'

//' / /

'

/

0

Octave

ll Triad

I

I

20

r:,/

/



ll

/

/

©

/

• •

I-

/

I-

'



I-

9@@~

Q.

c

IN

Bright ~hil

••

ll

/

([.~ N N

/

/

.<:::

0

I-

/

r-

Q.

"'~

ll ~/ k

/

50t--

V> V>

/

//0

0

r-

/

/

0

I-

/

I

I

30

CJ Sustained

I

40

50

60

Diaphragm active

Figure 4. Correlogram for the data on the relative duration of the closed phase (%). Symbols refer to subjects: (a) non-singers; (b) singers. 70

/

(a)

(b)

/

/

/ /

60

/

-

/ /

/

/

/

/ /

Q)

>

·;;; V> c

50

/

-

Q.

"'~

.s::

N

Q.

0

30

/

0

0

/

N NNN

E 40

c

0

/ /

0

/ /

r:, / A /

-

/

#../

/ / MM M

-

/

/

20

/ /

I

I

20

I

I

30

I

I

40

I

I

50

I

I

60

I

/

I

0

I

10

J"Bright"J

& • 6 0

/

/

10 / 10

{:,

' ' 'D

/

-

/

/0

/ttJ

/ /

/

/

-

/ /L /L /~ tL

/~ ark

r:,ll / /

/ /

/

I

I

Sustained Triad . Octave Subject RL Triad S b" JS Octave u 1ect

-1

20

.I

30

I

I

40

Diaphragm active

Figure 5. Correlogram for the data on the glottogram peak amplitude (1/s, arbitrary scales). Symbols refer to subjects: (a) non-singers; (b) singers.

Figure 5 shows the peak amplitude of the glottograms. All non-singers except subject N produced a higher glottogram amplitude when the diaphragm was activated. The singers showed smaller differences but the general trend was the same as for the nonsingers. In the case of the sustained tones of one of the singers, an active diaphragm yielded a higher amplitude no matter if the voice color was "dark" or "bright". This means that, regardless of this coloring, the peak trans glottal airflow was higher when the diaphragm was active. The above parameters all concern the voice source. In addition, the influence of the diaphragmatic activity on articulation was studied. Figure 6 shows the formant frequency variability during the non-singers' gliding tones and during the singers' triads and

356

J. Sundberg , Rolf Leanderson and Curt von Euler 30

/

/

0

/

Singers /

/

/

/ / /

/

.~

/

20f-

/

"'0"' a.

/ / //M

E

E

0> 0

/ /

/

.<:a.

6

/

L

N

10

// /

~

ffil /

/

/

/

{B]

/ /

0

/

o~/ 0

/

I

10

20

I

30

Diaphragm active

Figure 6. Correlogram for the data on the formant frequency variability under conditions of changing pitch. Symbols refer to subjects.

octaves, respectively. Except for subject 0 , the variability was greater under conditions of diaphragmatic passivity for both singers and non-singers. 5. Discussion

Some previous investigations have reported on a phonatory dimension which seems relevant to the voice source effects observed, and which is apparently associated with the degree of vocal fold adduction (Rothenberg, 1972; Sundberg & Gauffin, 1979; Gauffin & Sundberg, 1980). The types of phonation representing the extremes of this dimension have been called "pressed", where both the adduction activity and the subglottal pressure are high, and "breathy", where the subglottal pressure is low and the vocal folds are so weakly compressed medially that they fail to close the glottis. " Flow" phonation, in between these two extremes, is characterized by a high amplitude of the flow glottogram combined with a marked closed phase. These different types of phonation are clearly manifested not only in flow glottograms but also in the voice timbre. For instance, pressed phonation is typically associated with a long closed phase and a low peak amplitude of the glottogram, and , in the spectrum, the amplitude of the fundamental is low. It seems that a change in the diaphragmatic activity affects this phonatory dimension , at least in some subjects. Under conditions of diaphragmatic co-activation, the glottogram amplitude rose while the closed/open ratio increased and, probably, the subglottal pressure remained essentially unchanged. This implies that in most subjects a diaphragmatic co-activation discernably affected the mode of phonation, presumably toward flow phonation, as suggested by the increase of the glottogram amplitude. Subject N formed an exception; when he phonated with a flaccid diaphragm, he exhibited a higher glottogram amplitude, probably in combination with a lower subglottal pressure. Thus, he seemed to approach flow phonation when he phonated with a flaccid diaphragm. It might be relevant that this subject is a clarinet player; this may have caused him to develop special aerodynamic patterns. We conclude that co-activating the diaphragm does not necessarily lead to identical phonatory effects in all subjects.

Voice source effects

357

The "dark" and "bright" timbre tones sung by one of the singers with and without diaphragmatic co-activation offer additional support for the hypothesis that diaphragmatic co-activation has a phonatory effect. These tones were sung with practically identical oesophageal pressure. Still, the glottogram amplitude was considerably higher in the case of an active diaphragm. Thus, in this case an activated diaphragm was associated with a change of phonation mode toward flow phonation. The results also showed a formant frequency effect; under conditions of diaphragmatic co-activation, vocal tract parameters, such as vertical larynx position, were more stable according to the reduced formant frequency variability. Thus, it is possible that a co-activation of the diaphragm helps the singer to stabilize his voice organ. Finally, it might be mentioned that the bare ability of subjects to voluntarily control their diaphragmatic activation during phonation has been doubted (Wade, 1954). It is interesting that the visual feedback provided to the subjects was sufficient to enable all subjects to develop a voluntary control of diaphragmatic activation. It is also interesting that the singer subjects had developed such a voluntary control during the course of their training, without visual feedback (cf. Sundberg et al., 1985). This supports the previous conclusion that diaphragmatic co-activation is associated with a perceptible phonatory effect. 6. Conclusions Different subjects show different phonatory behaviors when they activate the diaphragm, although the subglottal pressure does not necessarily change. In most subjects a diaphragmatic activation tends to increase the peak amplitude of the glottogram, suggesting a change of phonation mode toward flow phonation. In addition, the coactivation of the diaphragm tended to stabilize the formant frequencies and, hence, the vocal tract during pitch changes. The subjects are gratefully acknowledged for their participation in the experiment.

References von Euler, C. (1982) Some aspects of speech breathing physiology. In Speech motor control (S. Grillner, editor), pp. 93- 103, Oxford: Pergamon Press. von Euler, C. (1983) On the central pattern generator for the basic breathing rhythmicity , Journal of Applied Physiology: Respiration. Environment, Exercise Physiology, 55, 1647- 1659. Gauffin, J. & Sundberg, J. (1980) Data on the glottal voice source behavior in vowel production, STLQPSR 2-3/1980, 61 - 70. Stockholm: Royal Institute of Technology. Leanderson, R., Sundberg, J. , von Euler, C. & Lagercrantz, H. (1984) Diaphragmatic control of the subglottic pressure during singing. Transcripts of the XII symposium: care of the professional voice, 216--220. Newsome Davis, J., Goldman, M. , Loh, L. & Casson, M. (1970) Diaphragm function and alveolar hypoventilation, Quarterly Journal of Medicine, New Series, XLV, 177, 87-100. Rothenberg, M. (1972) The glottal volume velocity waveform during loose and tight voiced glottal adjustment. Proceedings of the seventh international congress of Phonetic Sciences in Montreal, pp. 380 - 388, Mouton, The Hague. Rothenberg, M. (1973) A new inverse-filtering technique for deriving the glottal air flow waveform during voicing, Journal of the Acoustical Society of America, 53, 1632- 1645. Sundberg, J. & Gauffin, J. (1979) Waveform and spectrum of the glottal voice source. In Frontiers of speech communication research, (B. Lindblom & S. Ohman, editors), pp. 301 - 320. London: Academic Press. Sundberg, J. , Leanderson, R., von Euler, C. & Lagercrantz, H. (1985) Activation of the diaphragm during singing. In Proceedings of the Stockholm music acoustics conference, (SMAC) (A. Askenfelt, S. Felicetti, E. Jansson & J. Sundberg, editors), pp. 279- 290. Stockholm: Publications issued by the Royal Swedish Academy of Music, No. 46. Wade, 0. L. (1954) Movements of the thoratic cage and diaphragm in respiration, Journal of Physiology, 124, 193- 212.