Professional Documents
Culture Documents
journal: TMH-QPSR
volume: 46
number: 1
year: 2004
pages: 013-024
http://www.speech.kth.se/qpsr
TMH-QPSR, KTH, Vol. 46, 2004
Paper presented at the Annual Symposium Care of the Professional Voice, Philadelphia, June 2003
Abstract
“Throaty” voice quality has been regarded by voice pedagogues as undesired and
even harmful. The present study attempts to identify acoustic and physiological
correlates of this quality. One male and one female subject read a text habitually
and with a throaty voice quality. Oral pressure during p-occlusion was measured
as an estimate of subglottic pressure. Long-term-average spectrum (LTAS) analysis
was used to describe the average voice quality. Sixteen syllables, perceptually
evaluated with regard to throaty quality by five experts, were selected for further
analyses. Formant frequencies and voice source characteristics were measured by
means of inverse filtering, and the vocal tract shape of the male subject’s throaty
and normal versions of the vowels [a,u,i,ae] was recorded by Magnetic Resonance
imaging. From this material area functions were derived and their resonance
frequencies were determined. To test the relevance of formant frequencies to
perceived throaty quality, experts rated degree of throatiness in synthetic vowel
samples in which the subjects’ measured formant frequency values were used.
The main acoustic correlates of throatiness seemed to be an increase of F1, a
decrease of F4 and in front vowels also a decrease of F2, presumably resulting
from a narrowing of the pharynx. In the male subject voice source parameters
suggested a more hyperfunctional voice in throaty samples.
Introduction
Voice quality is determined by formant dysphonic voices and found it to be correlated
frequencies and voice source characteristics. with hyperfunctional phonation.
Voice qualities in general may be hard to define. The term even seems to have social
The term “throaty voice” is an interesting connotations. According to Addington (1968),
example. It is frequently used; the Internet throatiness in male speakers is perceived as a
returned more than 15000 quotes of “throaty sign of a realistic and mature personality, while
voice”. According to Titze (http://www.ncvs.- in a female speaker it gives an impression of a
org/singers/longevi.pdf), throaty voice quality is less intelligent, masculine, ugly and careless
associated with high subglottal pressure and person.
glottal hypertension resulting in e.g., irregular Many voice pedagogues and therapists con-
pops and transient sounds and producing a voice sider throaty quality as undesirable or even
sounding strained, strangled, or otherwise harmful to the voice. By contrast, it is often
hypertense. According to Reid (1983) it sounds mentioned as a positive characteristic of singers’
“swallowed, dark, tight, covered, pinched”, and voice quality in music reviews, apparently
can be regarded as the opposite of forward serving as a timbral ornament in some non-
placement, which is typically considered a classical styles of singing.
desirable voice quality. Hammarberg (1986), There are some reasons to assume that
considering phonatory rather than resonatory resonatory characteristics are relevant to throaty
aspects of voice quality, included the quality voice quality. Laver (1975; 1980) presented a
throaty/guttural in her perceptual analysis of system for describing all possible variations of
voice quality. He based his system on The relation of the findings summarized
articulatory and glottal characteristics, referred above to the term “throaty voice” as used in
to as supralaryngeal and laryngeal settings, and vocal pedagogy is unclear. As a consequence, it
also documented the associated voice qualities is hardly possible to understand if/why the use
in terms of audio recordings and some acoustic of a throaty voice quality is harmful. The present
analyses. Nolan (1983) complemented his investigation is an attempt to identify main
documentation by means of long-term-average acoustic characteristics of throaty voice quality
spectra (LTAS) and spectrograms of speech and to elucidate their voice source, vocal tract,
produced with the various settings by Laver and and formant frequency correlates.
by himself. Laver’s system includes some
settings potentially related to throaty voice: Material and Methods
uvularized, pharyngalized and laryngo-
pharyngalized. The LTAS documentation did Recordings
not suggest any striking resonatory deviations
from a normal voice quality (neutral setting). Two subjects (male and female) with no known
Also, the audio examples did not sound clearly voice pathologies volunteered as subjects. They
typical of a throaty quality according to the read a standard Swedish text (‘Ett svårt fall’)
authors’ judgment. However, in the LTAS of comprising 91 words, first with their habitual
uvularized, pharyngalized and laryngo- voice, and then with what they considered a
pharyngalized samples by Nolan there was a throaty voice quality. The recordings were made
larger level difference between the regions of twice. Second time the subjects held a plastic
the fundamental (F0) and that of the peak near tube in the corner of the mouth so that the oral
0.6 kHz, the F0 region being weaker than in pressure could be recorded. In this case, the text
neutral voice. This could suggest a more was modified such that the first consonants of
hyperfunctional voice production (Laukkanen et certain syllables were replaced by the consonant
al., 1981; Gauffin & Sundberg, 1978). Further- [p]. This resulted in nonsense words but allowed
more, the spectrum slope was less steep around measurement of oral pressure for estimation of
1 kHz in pharyngalized voice and between subglottic pressure (Löfqvist et al., 1982).
approximately 0.5-1.2 kHz in laryngo- The recordings were made in an ordinary
pharyngalized voice. This could reflect both a room. Both the audio signal and the oral
more hyperfunctional voice production (Gauffin pressure signal were recorded on a multi-
& Sundberg, 1978; Hammarberg, 1986; Kitzing, channel digital instrumentation recorder (TEAC
1986) and a downward shift of the frequency of RD-200T PCM). The audio signal was picked
the second formant (F2). In Laver's samples of up by a microphone (TCM110, omni-
laryngo-pharyngalized voice, the mean LTAS directional, frequency response 50-18000 Hz,
slope was less steep, especially between 0.5-1.5 sensitivity –52 dB; length 11/16’’, diameter
kHz. The spectrograms taken from individual 5/16’’) attached to a headset providing a
words uttered by Nolan revealed a tendency for constant mouth-to-microphone distance of 15
F1 to be higher and F2 and F3 to be lower in cm. Oral pressure was picked up by a soft
pharyngalized voice compared to normal plastic tube of 16 cm length and 0.45 cm inner
(neutral setting); in laryngo-pharyngalized voice diameter attached to a pressure transducer
F1 was higher and F2 lower (F3 could not be (Glottal Enterprises MSIF-2). The audio signal
measured). In Laver's samples of pharyngalized was calibrated for sound pressure level (SPL) by
voice, F2 and F3 were lower than in normal recording two sounds, the SPL of which were
voice; F1 was lower in pharyngalized open measured by means of a sound level meter held
vowels but higher in pharyngalized closed next to the recording microphone. These SPL
vowels as compared to the vowels in neutrally values were announced on the tape. Likewise,
uttered words. Referring to Fant (1970) and to the pressure transducer was calibrated by
the phonetic classification system by Jacobson recording a set of known pressures determined
et al. (1952). Nolan interprets the results to by means of a U-tube manometer. Also these
reflect the so-called flatness, i.e. a decrease of pressure values were announced on the tape.
the amplitudes or a lowering of the frequencies On a later occasion, the male subject
of the upper formants. Flatness has been sustained the vowels [a:, i:, u:, ae:] in normal
regarded typical of lip-rounding, pharyngali- and throaty quality for about 15 seconds, while
zation and retroflex articulation. MR-images were shot of 14 sections,
14
TMH-QPSR, KTH, Vol. 46, 2004
centre of gravity
Figure 1. Left panel: MR image of a section in the posterior part of the mouth for the vowel [a] as
pronounced by the male subject in throaty voice. Right panel: polygon approximation and center of
gravity of the vocal tract contours in the same section.
not evident from the MR images. Lip protrusion meters were derived: period length, duration of
and spreading caused similar problems. the closed phase, closed quotient (Qclosed , peak-
The formant frequencies of the area functions to-peak amplitude, negative peak amplitude of
obtained from the MR material were estimated the differentiated flow glottogram or maximum
using the custom-made Formflek program flow declination rate MFDR, normalized
(Johan Liljencrants). This program accepts as amplitude quotient NAQ defined as the peak-to-
input an area function and calculates the peak amplitude/MFDR x period duration (Alku
associated formant frequencies. et al., 2002), and level difference between the
two lowest source spectrum harmonics H1 -H2.
Analyses Oral pressure during [p] was measured for an
Analyses were carried out using the various estimate of subglottic pressure (Psub).
modules in the Soundswell Signal Workstation A second listening test was carried out to test
(Ternström, 1992) complemented by a custom the relevance of formant frequencies to
made program for inverse filtering (DeCap, perception of throatiness. In this test, synthetic
Svante Granqvist). The two subjects’ readings vowel stimuli were made using the KTH
were digitized and stored as sound files. From MUSSE synthesizer (Sundberg, 1989) with the
these sound files, sixteen mono- or bisyllabic formant frequencies measured in the vowels that
words or morphems (‘bond’, ‘bror’, ‘dag’, ‘den’, had shown the greatest difference in perceived
‘din’, ‘där’, ‘från’, ‘först’, ‘i’, ‘sa’, ‘spade’, throatiness between the habitual and throaty
‘står’, ‘sva’, ‘svar’, ‘till’, ‘ut’), uttered both versions in the first listening test. These vowels
habitually and in throaty voice, i.e., 64 words in (20 in total) were [y:] and two different samples
total, were selected for further analysis. The of [a:] from the female subject and [i:, i, e, æ:]
selection was based on the authors’ impression and three samples of [a:] from the male. The
of great difference in throatiness. Each of these following settings were used in the synthesis: F0
samples was copied separately into a sound file. 110 and 220 Hz for male and female voice
Five voice and speech specialists (2 females, 3 synthesis examples, vibrato amplitude 6 cent,
males) evaluated the files with respect to flutter amplitude 55 cent, flutter rate 6 Hz,
throatiness in a listening test run with the aid of flutter filter bandwidth 3 Hz. This provided
the custom-made computer program (Judge, reasonably natural sounding voice samples. The
Svante Granqvist). This program presents the samples, which were of 2 second duration and
stimulus files in random order and allows each had a smoothed onset and offset (linear
evaluator to listen to a given stimulus as many amplitude change during 100 ms) were
times as he/she wishes. The listeners give their presented in Listening test 2 to seven voice and
ratings by adjusting the position of a slider speech specialists (5 females, 2 males), who
controllable by the mouse. The ratings are were asked to rate the degree of throatiness,
automatically saved in files as numbers from 0 following the same procedure as in Listening
to 1000. The scale was labeled “Throatiness” test 1. As the number of the stimuli was
and its extremes were marked “Nil” and sufficiently small, each stimulus was included
“Extreme”. The stimuli were presented via twice at random intervals in the set, thus
headphones (Sennheiser HD 435 Manhattan). allowing assessment of intra-rater reliability.
One evaluator listened to all the samples twice
on different days to estimate intrarater Results
reliability. Average spectrum characteristics
were analyzed in terms of LTAS of the whole Listening test 1: Normal stimuli
reading samples using the spectrum analysis Intra- and inter-rater reliability were rather high
program of the Swell Workstation with 8 kHz (intrarater reliability: alpha = 0.99; inter-rater
frequency range, Hanning window and 150 Hz reliability: alpha = 0.77). The samples repre-
bandwidth; voiceless sounds were excluded senting throaty voice were perceived as signi-
from the analysis. ficantly (p = 0.000) more throaty than the
Formant frequencies were measured in samples produced in the habitual way, although
vowels carrying the main stress in the 16 words some habitual voice samples were evaluated
listed above. These measurements were made by somewhat throaty, too. The samples produced
means of the DeCap program which also by the male subject in habitual voice were
provided the flow glottograms of the vowels. perceived to have a lower average degree of
From these glottograms, the following para-
16
TMH-QPSR, KTH, Vol. 46, 2004
0 0
-10 -10
Level [dB]
-20
Level [dB]
-20
EL Normal
-30 -30
EL Throaty JS Normal
JS Throaty
-40
-40
-50
-50
100 1000 10000
100 1000 10000
Frequency [Hz] Frequency [Hz]
Figure 3. LTAS of the female (left panel) and male (right panel) subjects’ readings in normal and
throaty voice. The level scale has been normalized with respect to the highest spectrum level.
F1 F2
2000
800
Male
700
Female
1500
600
Throaty
Throaty
500
1000
400
300
500
200 500 1000 1500 2000
200 400 600 800
Normal
Normal
F3 F4
3000 4000
3500
2500
Throaty
Throaty
3000
2000
2500
1500
2000
1500 2000 2500 3000
2000 2500 3000 3500 4000
Normal Normal
Figure 4. Scatterplots comparing formant frequencies in habitual and throaty voice. For each vowel,
the values for habitual voice are plotted along the x-axis and those for throaty voice along the y-axis.
Circles and triangles refer to the female and the male subject’s values, respectively. The graphs only
include data for those samples that differed by more than 600 units (out of 1000) in mean rating of
throatiness.
by a higher laryngeal position in the throaty in the left and right panels in Figure 6 show the
version (Reid, 1982). As the orientation of F1 and F2 ratios, respectively, between the
section 1 was fixed relative to the body, a raised subject’s throaty and the normal versions of the
larynx caused the glottis to approach this vowels. The white columns represent the results
section. As can be observed in area function obtained for the area functions. There are
published by Engwall (2002), the area 2 cm considerable discrepancies. Part of these
above the glottis is considerably wider than that differences may be related to the difference in
1 cm above the glottis. larynx height suggested by the vocal tract
Given the lack of detail regarding vocal tract contour in section 1. An attempt to model this
length, only rough estimations of formant vocal tract length difference was made by
frequencies could be made. The black columns adding 1 cm at the glottal end to the normal
18
TMH-QPSR, KTH, Vol. 46, 2004
10 8
Normal
8
6
Throaty
Area [cm2]
6
Area [cm2]
4
2
2
0
0 0 2 4 6 8 10 12 14 16 18
0 2 4 6 8 10 12 14 16 18
8 12
10
6
8
Area [cm2]
Area [cm2]
4 6
4
2
2
0 0
0 2 4 6 8 10 12 14 16 18
0 2 4 6 8 10 12 14 16 18
Figure 5. Vocal tract cross-sectional areas as function of distance to glottis for the male subject’s
productions of the indicated vowels in habitual and throaty voice.
versions, using the cross sectional area measured habitual and throaty voice quality (Figure 7 and
in section 1. The formant frequency ratios Table 1). Thus, in throaty as compared with
between the throaty and normal versions habitual voice, Psub was higher, pulse peak-to-
obtained after this lengthening are represented peak amplitude was lower, Qclosed was higher,
by the gray columns in the same Figure 6. The NAQ was lower and H1 -H2 was lower. These
added vocal tract length increased the similarity differences suggest a more hyperfunctional-
or decreased the dissimilarity with the subject’s /pressed type of phonation in the throaty
data. samples. In the samples of the female subject,
there was no significant difference in the flow
Flow glottograms and subglottal glottogram parameters between the voice
pressures qualities studied.
For the male subject, all parameters except
MFDR showed significant differences between
F1, Throaty / Normal F2, Throaty / Normal
0.2 0.3
0.1
0.2
0.0
0.1
-0.1 a ae i u
-0.2 0.0
-0.3 a ae i u
-0.1
-0.4
-0.5 -0.2
Figure 6. F1 and F2 ratios (left and right panels, respectively) between the male subject’s habitual
and throaty productions of the indicated vowels. Black, white and gray columns show data obtained
from audio recordings, from the area functions in Figure 5, and from the same area functions after
lengthening the glottal section of those pertaining to habitual voice by 1 cm, so as to reflect longer
vocal tracts.
Throaty
Throaty
Throaty
-200
11
9 -300
0.2
7 -400
5 -500 0.0
5 7 9 11 13 15 -500 -400 -300 -200 -100 0 0.0 0.2 0.4 0.6
Normal Normal Normal
Throaty
Throaty
0.4 0,3
15
0.3
0,2
0.2 10
0.1 0,1 5
0.0
0,0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0,0 0,1 0,2 0,3 0,4 0,5 0 5 10 15 20 25 30
Normal Normal Normal
Figure 7. Scatterplots comparing subglottal pressure and flow glottogram parameter values observed
in the subjects’ habitual and throaty voices. For each vowel, the values for habitual voice are plotted
along the x-axis and those for throaty voice along the y-axis. Circles and triangles refer to the female
and the male subject’s values, respectively. The graphs only include data for those samples that
differed by more than 600 units (out of 1000) in mean rating of throatiness.
Table 1. Means of subglottic pressure and flow glottogram parameters in the subjects’
habitual and throaty versions of the vowels analyzed. Psub = subglottic pressure, Qclosed=
closed quotient, Aptp = peak -to-peak amplitude, H1 -H2 = dB level difference between the two
lowest source spectrum harmonics, MFDR = maximum flow declination rate, NAQ =
normalized amplitude quotient.
Listening test 2: Synthetic stimuli 0.67, respectively). The mean ratings for the
Also, in the second listening test with stimuli synthesized with formants from throaty
synthesized stimuli, the intra- and inter-rater and normal vowels differed in the expected way
reliabilities were rather high (alpha 0.84 and in most cases (Figure 8). Even though [i:] and
20
TMH-QPSR, KTH, Vol. 46, 2004
[u:] differed negligibly in mean rating and the In the male voice, it was produced with a narrow
normal version of [i] was rated as more throaty pharynx. He used a more hyperfunctional/-
than the throaty version, the results supported pressed mode of phonation in throaty voice as
the assumption that formant frequencies are shown by the higher Psub , the higher CQclosed, the
relevant to perceived throatiness. Thus, on lower pulse amplitude, the lower NAQ and the
average the stimuli synthesized with formant smaller H1-H2 difference. (Jacobson et al, 1952;
frequencies from throaty samples were per- Hammarberg, 1986; Verdolini et al, 1998).
ceived as significantly more throaty (p = 0.01) F1 is affected by jaw position, or, more
than the samples with the formant frequencies specifically by the cross sectional area in the
taken from habitual samples. pharynx as compared to that in the mouth (Fant,
The difference in mean rating between the 1970), a narrowing of the pharynx and a
throaty and habitual versions were substantially widening of the mouth raising F1. Hence the
smaller for the synthesized than for the natural narrow pharynx found in the male subject’s MR
stimuli (Figure 8). For samples synthesized after images of throaty samples is in agreement with
habitually spoken vowels, the mean rated the higher F1 values observed. The lower F2 in
throatiness was 282 (SD 138) and for samples front vowels may be caused by the narrowing of
synthesized after throaty vowels the corre- the pharynx observed in the MR material, even
sponding value was 421 (SD 175). This shows though a raised larynx would counteract this
that the perception of throatiness was less clear effect. Thus, many of the formant frequency
for the synthetic than for the natural stimuli. shifts observed may be caused by the narrowing
Interestingly, some evaluators commented that of the pharynx.
some of the samples sounded more pressed but An LTAS reflects both formant frequency
not necessarily throaty. and voice source characteristics and should
hence agree with measurements of these para-
meters. The LTAS of the throaty voice showed a
lower spectrum level below 0.5 kHz, and higher
Mean ratings
levels in the range of 1-3 kHz, particularly in the
1000 male voice. When the frequency distance
between two formants is decreased, spectrum
800 level at and between these formants will rise
(Fant, 1970). Therefore, an increase of F1 will
raise the spectrum level above its frequency,
Throaty
600
other things being equal. At least part of the rise
400 Female synthesized
between 1 and 3 kHz may be caused by the
Male synthesized
higher F1. The lower LTAS level in the F0
200 Female natural
range could be caused by the more hyper-
Male natural
functional type of phonation observed in the
0 throaty samples of the male voice. Also, both
0 200 400 600 800 1000 subjects raised F0 when speaking with a throaty
Normal voice, and this may contribute to lowering the
Figure 8. Result of listening test 2 with LTAS level in the low frequency range.
synthesized vowel stimuli. Open symbols The panel in both listening tests showed
represent mean ratings of stimuli with formant reasonably high intra- and inter-subject relia -
frequencies observed in female and male bilities, suggesting that they had similar ideas of
subjects’ habitual and throaty versions of the the term throaty. On the other hand, it is also
possible that they simply rated the degree of
vowels. Filled symbols refer to the mean
voice abnormality. To test this, the term throaty
ratings of stimuli produced by the subjects.
should be included as one out of many
perceptual parameters in a future test. On the
other hand, there are reasons to assume that the
Discussion term was familiar to the subjects and had a
Our findings have shown that throaty voice similar meaning. It may be relevant that arti-
quality was associated with a higher F1, a lower culation of throaty voice was actually throaty in
F2 in front vowels, and a lower F4 in all vowels. the sense that the lower pharynx was more
constricted in throaty than in habitual production associated with a more hyperfunctional type of
of the vowels [u, i, a, ae]. phonation. Thus, a narrowing of the pharynx
It is interesting that the term “throaty” is would easily (if not necessarily) lead to a firmer
commonly used for this type of voice. glottal adduction. This, in turn, may explain why
Generally, our intuitive knowledge of articu- throaty/guttural quality has been regarded even
lation is quite limited and vague. Yet, the term as harmful to voice. Hyperadduction increases
throaty rightfully suggests that articulation is collision force of the vocal folds during
associated with a particular pharyngeal con- phonation (Jiang & Titze, 1994), which means
figuration. This reflects an intuitive realism, higher mechanical loading on the vocal fold
possibly based on proprioceptive feedback. tissue and, hence, a higher risk for tissue
Incidentally, such intuitive realism on articu- damage.
lation would constitute an essential part of voice
teachers’ skills. Conclusions
The attempts to synthesize a throaty quality
by formant frequency combination only were Throaty voice quality seems to result from a
partly successful, and the perceptual voice narrowing of the pharynx, probably combined
quality difference between habitual and throaty with a somewhat hyperfunctional type of
was much smaller for the synthesized than for phonation. Acoustically the narrow pharynx
the natural stimuli. This indicates that the appears to induce an increase of F1, a decrease
formant frequencies alone do not exhaustively of F2 in front vowels and a decrease of F4. The
define throaty quality. Also voice source voice source characteristics lead to an attenu-
characteristics seem relevant. Interestingly, the ation of the fundamental and an increase of the
male subject’s examples, which differed also spectrum level between 1 and 3 kHz. The
with respect to phonatory hyperfunction, were formant characteristics and also hyperfunctional
perceived as throatier than the female subject’s phonation seem perceptually important to
examples, which were produced with similar throaty voice quality.
voice source properties in habitual and throaty
voice. According to some listeners’ comments Acknowledgements
voice quality sounded more pressed in the male The kind cooperation of the subjects and the
subject’s throaty samples. listeners is gratefully acknowledged. The MR
Our results support the assumption that
recordings were made possible by Didier
throaty can be regarded as a setting in the sense
Demolin, Laboratoire de Phonologie Université
of the term proposed by Laver (1980). It is then
Libre de Bruxelles and were made at the Hôpital
interesting to find out if this setting is identical Erasme, Brussels, in cooperation with
with or similar to any setting described already radiologist Thierry Metens and phonetician
by Laver (1980) or Nolan (1983). The LTAS Alain Soquet, Laboratoire de Phonologie
differences we observed between normal and Université Libre de Bruxelles. The investigation
throaty samples resembled those obtained by was presented at the Annual Symposium Care of
Nolan (1983) between neutral and pharyngalized
the Professional Voice, Philadelphia, June 2003.
or laryngo-pharyngalized samples. In Nolan’s
Eva Björkner is financially supported by the
and Laver’s samples, F1 either rose (open
European Community’s Human Potential Pro-
vowels) or decreased (closed vowels) and F2 gramme under contract HPRN-CT-2002-00276
and F3 decreased. In the present study, F1 rose (HOARSE-network).
for the male both in front and back vowels and
in open and closed vowels, F2 decreased in front
vowels and F4 decreased in all vowels. These References
findings suggest that apart from certain spectral Addington DW (1968). The relationship of selected
similarities ‘throaty’ is not exactly the same as vocal characteristics to personality perception.
‘pharyngalized’ or ‘laryngo-pharyngalized’. Speech Monographs: 35: 492-503.
Our results strongly suggest that throaty Alku P, Bäckström T & Vilkman E (2002).
Normalized amplitude quotient for parametrization
voice quality is associated with a narrowing of of the glottal flow. J Acoust Soc Am 112: 707-710.
the pharynx. It is tempting to speculate that this Bele I (2002). Professional Speaking Voice. A
narrowing is caused by a contraction of the perceptual and acoustic study of actors’ and
middle constrictor muscle, which also tends to teachers’ voices, Doctoral thesis, Department of
raise the larynx. A raised larynx is often Special Needs Education, University of Oslo.
22
TMH-QPSR, KTH, Vol. 46, 2004
Engwall O (2002). Tongue Talking - Studies in Friberg A, Iwarsson J, Jansson E & Sundberg J
Intraoral Speech Synthesis. Doctoral thesis. (eds). SMAC93, Proc of Stockholm Music Acoust
Department of Speech, Music and Hearing, KTH, Conf 1993, Stockholm: The Royal Swedish
Stockholm. Academy of Music, No 79: 206-210.
Fant G (1970). Acoustic Theory of Speech Pro- Löfqvist A, Carlborg B & Kitzing P (1982). Initial
duction. The Hague: Mouton. validation of an indirect measure of subglottal
Gauffin J & Sundberg J (1978). Clinical applications pressure during vowels. J Acoust Soc Am 72: 633-
of acoustic voice analysis - acoustical analysis, 635.
results and discussion. In: Buch NH, ed. Proc of Int Nawka T, Anders LC, Cebulla M & Zurakowski D
Assoc Log Phon Congress, Aug 5-18, 1977, (1997). The speaker's formant in male voices. J
Copenhagen, Denmark. Herning: Organizing Voice 11: 422-428.
Committee of the Congress, 489-502. Nolan F (1983). The Phonetic Bases of Speaker
Hammarberg B (1986). Perceptual and acoustic Recognition. Cambridge: Cambridge University
analysis of dysphonia. Doctoral dissertation. Press.
Karolinska Institutet, Stockholm. OSIRIS (http://www.expasy.org/www/UIN/html1/-
Jakobson R, Fant G & Halle M (1952). Preliminaries projects/osiris/osiris.html)
to Speech Analysis. Cambridge, Mass: MIT Press. Reid C (1983). A dictionary of vocal terminology. An
Jiang JJ & Titze IR (1994). Measurement of vocal analysis. New York: Music House.
fold intraglottal pressure and impact stress. J Voice Sundberg J (1989). Synthesis of singing by rule. In:
2: 132-144. Mathews M & Pierce J (eds.): Current Directions
Kitzing P (1986). LTAS criteria pertinent to the in Computer Music Research, System
measurement of voice quality. J Phonetics 14: 477- Development Foundation Benchmark Series,
482. Cambridge, MA: The MIT Press, , 45-55 and 401-
Laukkanen A-M, Vintturi J, Vilkman E, Sala E, 403.
Siikki I, Lukkarila P & Sihvo M (2001). Ternström S (1992). Soundswell-signal workstation
Perceptual, acoustic and self-reported correlates of software: manual version 3.06. Stockholm,
vocal loading. Proc of 25th World Congress of the Soundswell Acoustics.
Int Assoc Log Phon, Montreal 5.-9.8. Titze I. A few thoughts about longevity in singing,
Laver J (1975). Individual features in voice quality. http://www.ncvs.org/singers/longevi.pdf
Ph.D. thesis. University of Edinburgh. Verdolini K, Druker DG, Palmer PM & Samawi H
Laver J (1980). The Phonetic Description of Voice (1998). Laryngeal adduction in resonant voice. J
Quality. Cambridge: Cambridge University Press. Voice 12: 50-67.
Leino T (1994). Long-term average spectrum study
on speaking voice quality in male actors. In:
24