ORIGINAL ARTICLE

Voice source differences between falsetto and modal

registers in counter tenors, tenors and baritones
Johan Sundberg and Carl Högset
KTH Voice Research Centre, Department of Speech Music Hearing, KTH, 100 44 Stockholm, Sweden

Log Phon Vocol 2001; 26: 26 –36

Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13

Vocal registers are generally assumed to be associated with the voice source, i.e. the pulsating transglottal air ow. The
waveform of this air ow was analysed by inverse Ž ltering in professional singers, four counter tenors, Ž ve tenors, and four
baritones singing the syllable [pæ:] in soft, middle, and loud voice in modal and falsetto:counter tenor register. Subglottal
pressure, estimated from the intra-oral pressure during the occlusion for the consonant [p], closed quotient, relative glottal
leakage, and the relative level of the fundamental were analysed. The counter tenors used comparatively low subglottal
pressures and mostly showed a closed phase in their  ow glottogram waveform. For a given value of the closed quotient,
the fundamental tended to be stronger in falsetto than in modal register. The observed voice source differences between the
registers seem related to a greater vocal fold thickness in modal than in falsetto register.
Key words: register, singing, subglottal pressure, voice source.

Johan Sundberg, KTH Voice Research Centre, Department of Speech Music Hearing, KTH, 100 44 Stockholm, Sweden. Tel:
»468 790 7873. Fax: »468 790 7854.
For personal use only.

INTRODUCTION Many investigations of vocal registers have con-

A vocal register refers to perceptually distinct regions
of vocal quality that can be maintained over some
range of pitch (9, 20). Registers represent an impor-
tant aspect of phonation and much research has been
spent on describing perceptual, acoustic, and aerody-
namical correlates of various registers. Unfortu-
nately, register terminology is somewhat chaotic;
Mörner and coworkers (11) found more than 100
register terms used in the literature. In this article, the
terms modal will be used for the register typically
used by adults in neutral speech and the term falsetto
for the register typically used by adult males when
they phonate at their highest possible fundamental
frequencies. In the literature review below, it will be
assumed that modal register is identical with chest
and heavy register, and that falsetto is identical with
loft and light register.
Registers are important to singing, particularly in
counter tenors, who use a special register that is
commonly regarded as falsetto. On the other hand,
many voice pedagogues regard the register used by
counter tenors as different from the typical falsetto
register.
cerned the acoustic differences. Most of them have
found that the overall spectrum slope was steeper in
tones produced in falsetto than in modal register (see
(3, 7, 13)). Colton (3) noted that tones sung in
falsetto register tended to have a fundamental of
greater amplitude than tones sung in modal register
(16). On the other hand, Schutte and Miller (14)
measured the sound level relationship between lower
and higher spectrum partials, which they called reso-
nance balance, but failed to observe any correlation
with registers.
It is generally agreed that registers are associated
with the vibrational characteristics of the vocal folds.
The physiological correlates of the modal and falsetto
registers have been discussed over many decades in
the literature. According to van den Berg (21), the
chest register is characterized by a strong longitudinal
tension in the vocalis muscles while the falsetto regis-
ter is characterized by a strong longitudinal tension in
the vocalis ligament. As a result, the vocal folds are
thicker and vibrate with a greater mass in modal
register, which produces stronger high overtones.
These ideas were substantiated in EMG investiga-
tions by Hirano, Vennard and Ohala (8) and Hirano

© 2001 Taylor & Francis. ISSN 1401-5439 Log Phon Vocol 26


(6) who found that modal register was characterized
by an activation of the vocalis muscles, which would
lead to a thicker vibrating mass. Hirano (7) describes
the role of the body, transition, and cover layers of
the vocal folds in voice register control. In modal
register, the body is stiffer than the cover so that the
cover creates a muscosal wave with a vertical phase
difference. In falsetto register, the vocalis activity is
small and the various layers are stretched, thus reduc-
ing the muscosal wave.
The idea that the vocal folds are thicker in modal
than in falsetto register has found some support from
electro-glottographi c (EGG) data. The amplitude of
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
an EGG signal re ects the vocal fold contact area
and this amplitude has been found to suddenly de-
crease, as the register shifted from modal to falsetto
(1, 13).
On the basis of the above Ž ndings it seems fair to
assume that the modal and falsetto registers differ
with respect to vocal fold vibration. Differences in
vocal fold vibration characteristics should affect the
waveform of the pulsating transglottal air ow, i.e. the
voice source. This air ow can be derived from inverse
Ž ltering, implying that the in uence on the  ow
For personal use only.
waveform of the vocal tract sound transfer character-
istics is eliminated. Thus, the result, the  ow glot-
togram, shows the waveform of the voice source
itself. The  ow signal is generally derived from the
pressure difference across an acoustic resistance. The
method has the advantage being reasonably non-in-
vasive, only slightly disturbing the subject’s auditory
feedback. The representation of the voice source ob-
tained is informative, as it represents the sound pro-
duced by the pulsating transglottal air ow.
Morevover, this representation is physiologically real-
istic, re ecting, e.g. the duration of the closed phase
of the vocal fold vibration cycle. The waveform also
re ects important aspects of phonation. Thus,
breathy (hypofunctional ) phonation typically pro-
duces high peak-to-peak amplitude of the  ow pulses
and the lack of a closed phase. Conversely, pressed
(hyperfunctional ) phonation produces low peak-to-
peak pulse ampitude and a clearly marked, long
closed phase. In addition, spectrum analysis of this
waveform reveals the amplitudes of the source spec-
trum partials. In particular, the lowest spectrum par-
tial, i.e. the fundamental, is relevant as its amplitude
is dependent on the shape and amplitude of the  ow
pulse. Thus, pressed phonation produces a weak am-
plitude of this partial, while breathy phonation pro-
duces a high amplitude (4). The level of the
fundamental is generally measured in terms of the
level difference between the Ž rst and the second
source spectrum partial (2, 15).
Voice source in modal and falsetto register 27
In a previous investigation (18), this method was
applied to derive the voice source in the modal and
falsetto registers in four counter tenor singers and an
amateur baritone singer. The results showed that
lower subglottal pressures were used in the falsetto
register and that the subglottal pressure range be-
tween soft and loud singing was wider in the modal
register. It was also observed that under equal pitch
and loudness conditions the peak-to-peak amplitude
was higher and the closed phase shorter in falsetto.
Signs of a higher degree of glottal adduction were
observed in the modal register. Glottal leakage was
found to be greater in the baritone singer’s falsetto
than in the falsetto of the counter tenors.
The pitch range of a voice depends on the dimen-
sions of the vocal folds (20). Thus, tenors have
shorter vocal folds than baritones. Also, tenors some-
times use the falsetto register for soft tones sung at
high pitches. Therefore, it is interesting to compare
the falsetto characteristics in different voice
classiŽ cations.
The purpose of the present investigation was to
complement the previous study mentioned by com-
paring the voice source in falsetto and modal registers
in a larger number of subjects of different voice
classiŽ cations.
METHOD
The  ow signal was recorded by means of a Rothen-
berg mask (12). Subglottal pressure was estimated
from the intra-oral pressure during the occlusion for
the consonant [p] (5). This pressure was captured
from a thin plastic tube which the subject held in the
corner of his mouth and which was connected to a
pressure transducer mounted in the mask. As this
pressure is typically equal or nearly equal to subglot-
tal pressure, it will henceforth be referred to as the
subglottal pressure. The pressure and  ow signals
were recorded on two tracks of a multi-channel
TEAC PCM recorder TEAC Corporation, Tokyo,
RD-200T. Calibration of pressure was made by
recording on the same tape pressures determined
from a manometer. Similarly,  ow calibration signals,
determined by means of a Rotameter  ow meter
(VEB PruÈ fgeraÈ te Werk, Dresden TG300) were
recorded. An audio signal was also recorded for the
purpose of protocol.
Four counter tenor singers (age range 40 – 53), Ž ve
tenors (age range 32 –53), and four baritones (age
range 24 – 35) were recruited as subjects. All were
professional singers in the sense that they are regu-
larly engaged as soloists in public concerts. Some of
them had also sung in performances of contemporary
or baroque operas.
Log Phon Vocol 26
 28 J. Sundberg and C. Högset
The singers were asked to sing a sequence of the
syllable [pæ:] in what they perceived as middle, loud,
and soft degrees of vocal loudness in both falsetto
and modal registers at the pitches shown in Table 1.
The duration of the four Ž rst syllables was about 700
msec. Some singers preferred not to sing the highest
pitches. The pitches were sung to the subjects by the
experimenter.
The  ow signal was fed from the tape into a
transient recorder (Glottal Enterprises, Syracuse,
New York, BT-1) and the second [pæ:] in each se-
quence was captured. The output from the transient
recorder was fed to an inverse Ž lter (Glottal Enter-
prises, Syracuse, New York, MSIF-2). The frequen-
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
cies and bandwidths of the two lowest formants were
set such that the ripple during the closed phase was
minimized. In some cases of soft singing where no
closed phase was visible, the same Ž lter setting was
used as for the adjacent louder productions of the
same singer at the same pitch. Then, the entire se-
quence of [pæ:] syllables was run through the inverse
Ž lter and the output plus the intra-oral pressure
signal were recorded on two channels in sound Ž les,
using the Sound Swell signal workstation (HighTech,
Stockholm). For each pitch, the second [pæ:] syllable
For personal use only.
was selected for analysis. Measurements were carried
out on a  ow glottogram selected from near the
middle of the duration of the vowel. Here, phonation
had stabilized after the initial variations associated
with the pronunciation of the consonant [p]. Data
were secured from one waveform, as the waveform
characteristics showed little variation from period to
period.
The following waveform characteristics were deter-
mined from the sound Ž les: intra-oral pressure, pe-
riod and closed phase duration, peak-to-peak
amplitude of the  ow pulse, and the glottal leakage,
deŽ ned as the mean air ow during the closed phase.
In addition, the spectrum of the selected waveform
was analysed by means of a subroutine in the Sound
Swell signal workstation, and the level difference
between the two lowest partials (H1-H2) was mea-
sured. The closed quotient was computed from these
values, and also the relative glottal leakage, deŽ ned
Table 1. Pitches sung by the three ×oice classiŽ cations
in falsetto and modal registers. Pitches in bold italics
were sung in both registers
ClassiŽ cation Falsetto tones Modal tones
Counter tenors B3, C4, F4, A4, D3, F3, B3, C4
Tenors C4, E4, F 4, C4, E4, F 4, G 4,
G 4, A 4 A 4
Baritones C4, Eb4, F 4, F3, B3, C4, Eb4,
G 4 F 4, G 4
Log Phon Vocol 26
as the ratio between the glottal leakage and the peak
amplitude of the  ow pulse.
RESULTS
Subglottal pressure is a major control parameter in
phonation. Fig. 1 shows this pressure as function of
fundamental frequency for the different singers. By
and large, the curves show expected trends; the
highest pressures were used for the loudest produc-
tions and within a given degree of vocal loudness the
pressures tended to increase with fundamental fre-
quency in both registers.
A comparison between singer categorizations re-
vealed that the baritones and, in particular, the tenors
used comparatively high pressures in modal register,
some values amounting to 50 cm H2 O or even higher.
The counter tenors, on the other hand, used low
pressures, particularly subject L whose maximum
pressure amounted only to 15 cm H2 O. These obser-
vations agree with those reported by Södersten and
coworkers (18). Also, the pressure ranges for the
counter tenors for a given pitch was narrow. Indeed
for singers C and L, the different degrees of vocal
loudness were produced with almost constant
pressures.
Fig. 2 shows typical examples of  ow glottograms
for modal and falsetto register as produced by
baritone singer L and counter tenor B. Disregarding
the ripple during the closed phase, which is because
of incomplete cancellation of formants, it can be seen
that in falsetto the closed phase is shorter, the pulses
are more rounded, and the air ow during the closed
phase is greater.
Subglottal pressure has been found to affect the
closed quotient in professional baritone singers (17).
In most cases, the closed quotient reached about 0.55
for the highest pressures in modal register and lower
values in falsetto. For high subglottal pressures, the
closed quotient was greater at higher pitches than it
was for lower pitches. For the baritones, the closed
quotient generally increased to a value near 0.55 at
the highest pressures (cf. (17)). Relative glottal leak-
age showed a weak tendency to decrease with increas-
ing subglottal pressure in many subjects. For all tones
sung in both registers tenor J’s leakage values de-
creased almost linearly with subglottal pressure,
reaching zero for pressures greater than 25 cm H 2 O.
Figs 3 – 6 compare subglottal pressure and  ow
glottogram characteristics for falsetto and modal reg-
ister tones under conditions of identical pitch and
degree of vocal loudness. For all three categoriza-
tions, the subglottal pressures used in falsetto showed
a linear correlation with those used in modal register,
the correlations ranging between 0.715 and 0.948

(Fig. 3, Table 2). The slope of the best linear Ž t to the
data points varied considerably between subjects. For
three of the counter tenors it was close to 2, and for
the tenors and baritones, it varied between 0.71 and 4.29
and between 0.95 and 2.77, respectively. On average
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
For personal use only.
Fig. 1.
Voice source in modal and falsetto register 29
across all subjects, the pressures tended to be almost
twice as high in modal than in falsetto register for given
pitch and loudness. The counter tenors used markedly
low pressures, the maximum in falsetto register
amounting to 15 cm H2 O and in modal 24 cm H 2 O.
Log Phon Vocol 26
 30 J. Sundberg and C. Högset
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
For personal use only.

Fig. 1. Subglottal pressure as function of fundamental frequency for the different singers. Filled and open
symbols refer to modal and falsetto register, respectively.

The closed quotient was mostly greater in modal
than in falsetto for all three singer classiŽ cations (Fig.
4). For the counter tenors, it varied between zero and
0.33 in falsetto and between zero and 0.55 in modal
register. In falsetto, counter tenors M and C had a
closed quotient near 0.3 and 0.2, respectively, regard-
less of pitch and loudness, while in modal register
their closed quotients varied considerably. For the
tenors and baritones, the values of closed quotient
varied between 0 and 0.45 and between 0 and 0.57 in
falsetto. The corresponding values for modal register
were between 0.1 and 0.5 and between 0.15 and 0.66,
respectively. Thus, their closed phase was always
greater than zero in modal register but not in falsetto.

Fig. 2. Typical examples of  ow glottograms produced at the same pitches by barytone singer A and counter
tenor singer B in the modal and falsetto registers.

Log Phon Vocol 26

 Voice source in modal and falsetto register 31
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13

Fig. 3. Comparison of subglottal pressure for tones sung in falsetto and modal register at identical pitches and
degrees of vocal loudness. Symbols refer to subjects.
For personal use only.

Fig. 4. Comparison of the closed quotient for tones sung in falsetto and modal register at identical pitches and
degrees of vocal loudness. Symbols refer to subjects.

Fig. 5. Comparison of relative glottal leakage, deŽ ned as the ratio between the air ow during the quasi-closed
quotient and the peak amplitude of the  ow pulse, for tones sung in falsetto and modal register at identical
pitches and degrees of vocal loudness. Symbols refer to subjects.

In most cases, the relative leakage values were relative leakage varied between 0.22 and 0.48 for the
greater in falsetto than in modal register (Fig. 5). This counter tenors. The corresponding ranges for the
was true for all of the counter tenors, for four of the tenors and baritones were between 0 and 0.55. In
tenors, and for three of the baritones. In falsetto, the modal register, the range was between 0 and 0.5,

Log Phon Vocol 26

 32 J. Sundberg and C. Högset
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13

Fig. 6. Comparison of H1-H2, i.e. the level difference between the fundamental and the second partial of the
voice source spectrum, for tones sung in falsetto and modal register at identical pitches and degrees of vocal
loudness. Symbols refer to subjects.

approximately. For a given pitch and degree of vocal
loudness the leakage was mostly greater in falsetto
than in modal register for the counter tenors.
H1-H2 was generally greater in falsetto than in
modal register (Fig. 6). In falsetto, the values ranged
between 10 dB and 25 dB for the counter tenors and
between approximately 7 dB and 34 dB for the tenors
For personal use only.
same pitch and degree of vocal loudness subglottal
pressure was lower, the closed phase was mostly
shorter, while the relative glottal leakage and the
H1-H2 were mostly greater in falsetto than in modal.
These differences applied to all classiŽ cations. The
counter tenors used narrower pressure ranges than
the tenors and baritones, particularly in the modal

and baritones. In modal, the range was between 0 dB register and their closed phase, relative glottal leak-
and 20 dB for the counter tenors, 3 dB and 22 dB for age and H1-H2 varied within narrower ranges than in
the tenors, and 3 dB and 28 dB for the baritones.
the tenors and baritones.
Particularly for counter tenors A, B, and C, the
values remained rather constant regardless of loud-
ness and pitch in falsetto, while it varied considerably
Table 2. Trendline slopes and, intercepts (Icpt) and
more for the tenors and for the baritones. Con-
the correlation coefŽ cients (r) for the relationship be -
versely, for the tenors and the baritones the variation
tween the subglottal pressures used by the singers in
tended to be less in modal than in falsetto register.
falsetto and modal register at identical pitches and
The level difference between the two lowest partials
degrees of ×ocal loudness
of the source spectrum, H1-H2, tends to be linearly
correlated with the closed quotient (17). Fig. 7 shows ClassiŽ cation Singer Slope Icpt r
this relation for the present singers. A trend toward a
linear relation can be found. However, for a given Counter tenor A 1.89 ¼5.6 0.948
closed quotient value, many data points pertaining to Counter tenor B 2.23 ¼3.7 0.752
Counter tenor C 0.44 6.2 0.768
falsetto register are markedly higher than those sung Counter tenor L 1.84 ¼6.2 0.862
in modal register. This implies that the same closed Counter tenors Mean 1.60 ¼2.3 0.833
quotient value was often associated with a smaller SD 0.80 5.8 0.091
H1-H2 value in modal than in falsetto. This is be- Tenor S 0.78 ¼18.2 0.936
cause of the waveform difference illustrated in Fig. 2; Tenor B 0.71 11.9 0.747
the  ow glottogram pulse in modal showed a triangu- Tenor U 1.92 ¼0.5 0.813
lar shape while in falsetto the pulse had a more Tenor M 1.33 9.3 0.893
Tenor J 4.29 4.0 0.821
rounded shape, somewhat like a Roman arch.
Tenors Mean 2.06 6.2 0.819
SD 1.57 5.5 0.060
Baritone B 1.82 4.5 0.873
DISCUSSION Baritone M 0.95 16.9 0.786
Our results have demonstrate d several voice source Baritone J 2.77 1.7 0.829
Baritone A 2.50 2.3 0.715
differences between modal and falsetto register, and Baritones Mean 2.01 6.3 0.801
the differences varied somewhat between the three SD 0.81 7.1 0.067
singer classiŽ cations. Thus, for tones sung at the

Log Phon Vocol 26

 Voice source in modal and falsetto register 33
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13

Fig. 7. Relationship between H1-H2 and the closed quotient in falsetto and modal register (open and Ž lled
symbols, respectively). Symbols refer to subjects.

What could be the physiological origin of these
differences? Subglottal pressure is a parameter of
decisive importance to the voice source (17). Thus, it
has been found that an increase of the closed phase
and a decrease of the H1-H2 accompany an increase
of subglottal pressure. Therefore, the counter tenors’
For personal use only.
more dominant fundamental than a triangular pulse
shape; as mentioned in the introduction the falsetto
register has been found to have a stronger fundamen-
tal than the modal register (3). Thus, our observa-
tions seem compatible with previous Ž ndings
regarding vocal fold thickness in falsetto and modal

narrower pressure range may be related to their registers and highlight the relationship between vocal
narrower range of the closed phase and the H1-H2. fold characteristics and acoustic properties of the
The reason why they use a narrower pressure range is voice source.
less clear. One possibility would be that they sing
with less contraction of the thyroarytenoi d muscles;
this would increase the mobility of the vocal folds,
such that lower pressures can drive them. Indeed,
Hirano and coworkers (8) found less contraction of
these muscles in falsetto than in modal register. More
information on the vocal fold mechanism in falsetto
could shed some light on this issue.
As previously mentioned, the vocal folds have been
found to be thicker in modal than in falsetto register
(21). A rounded pulse shape can be expected from
thin vocal folds where the phase lag between the
upper and lower layers of the glottis would be minor.
A triangular shape can be expected from thick vocal
folds, producing a greater phase difference between
the upper and lower layers (Fig. 8). Thus, the pulse
shape difference is in accordance with the observation
that the folds tend to be thinner in falsetto than in Fig. 8. Schematical illustration of the effect of vocal
modal register. The closed phase tended to be clearly fold thickness on the glottal area. Thick vocal folds
longer in modal than in falsetto register for all three (upper panel) imply a great phase lag between the
voice classiŽ cations. Also this seems compatible with upper and lower layer of the folds, such that the
the idea that the vocal folds are thicker and, hence, closing of the lower layer interrupts the opening of
the phase lag between the upper and lower layers of the upper layer. As a result the area waveform be-
the folds is greater; a great phase lag will tend to comes triangular in shape. For thin vocal folds (lower
prolong the closed phase, as illustrated in Fig. 8. panel) the phase lag is small, and the area waveform
Note also, that a rounded pulse shape will generate a is more rounded.

Log Phon Vocol 26

 34 J. Sundberg and C. Högset
The closed phase showed much more variation
with subglottal pressure than was previously observed
for professional baritone singers (17). The greater
variability would be a result of the experimental
conditions. In the baritone experiment, the subjects’
task was to sing a tone with continuously decreasing
loudness, repeatedly inserting the consonant [p].
Thus, vocal loudness was the only parameter varied.
In the present experiment, the singers produced the
three degrees of loudnesses in different takes.
Our data were all assembled from the syllable
[pæ:]. The singers’ consistency with respect to the two
lowest formant frequencies was quite high. While the
task of singing the same syllable would be particu-
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
larly revealing for voice source characteristics in a
neutral singing mode, it is quite likely that the singers
vary these characteristics in songs, depending on the
musical context. Yet, our data would be representa-
tive of the voice source differences between the two
registers analysed.
In comparing the  ow glottogram properties in the
two registers, pitch and degree of vocal loudness were
kept identical. This implies that tones that were high
within the modal register range were compared with
tones that were low within the pitch range of the
For personal use only.
falsetto. This may account for some of the observed
subglottal pressure differences between the registers.
In this study, we measured glottal leakage in rela-
tion to the amplitude of the modulated transglottal
air ow while, normally, this leakage is measured in
absolute terms. For a given degree of glottal adduc-
tion, both absolute leakage and peak amplitude are
in uenced by subglottal pressure and vocal fold
length: the higher the pressure and the longer the
folds, the greater the leakage and the peak air ow
amplitude. The exact relation between pressure on
the one hand and leakage and pulse amplitude on the
other may not be the same. Still, by dividing leakage
by peak air ow, the in uence of these factors should
be reduced such that the emerging ratio should more
efŽ ciently re ect the magnitude of the leakage.
An incomplete glottal closure is regarded as typical
of the falsetto register in untrained voices. It is inter-
esting, however, that many of the counter tenors’ and
tenors’ falsetto productions showed a clear closed
phase, sometimes even in soft singing. In laryngo-
Ž breoptic recordings, Södersten and Lindestad found
complete glottal closures in counter tenors (10, 19).
Thus, it seems fair to conclude that the falsetto
register used by counter tenors is often associated
with a partial or even a complete glottal closure. Such
a closure should be advantageou s from the point of
view of vocal loudness, as it reduces the sound energy
losses within the vocal tract. The reason is that a
completely closed glottis more efŽ ciently re ects the
Log Phon Vocol 26
sound waves; this narrows the formant bandwidths
and so contributes to producing a loud tone at a
given subglottal pressure. Falsetto register tones com-
prising a clear closed phase should, therefore, be
advantageou s in singing. It is interesting that also
three of the tenors produced falsetto tones with a
marked closed phase. It may be relevant that tenors
often resort to a falsetto-like register when singing
softly at high pitches.
By and large, our results agree with those found in
an earlier preliminary investigation (18). Thus, in that
study the subglottal pressures were found to be lower,
the range of subglottal pressures narrower, the closed
phase shorter, and the glottal DC  ow, i.e. the leak-
age, to be greater in counter tenor falsetto as com-
pared to modal register. This suggests that our data
are reasonably representative.
Certain differences were found between the falsetto
characteristics of the three classiŽ cations. The coun-
ter tenors used subglottal pressures lower than about
15 cm H2 O for all tones in both registers, while the
tenors and baritones used pressures up to 30 and 50
cm H2 O or higher for their falsetto and modal tones,
respectively. The use of low pressures may be a
signiŽ cant part of the singing technique of counter
tenors. Moreover, unlike the tenors and the
baritones, three of the counter tenors showed a com-
paratively narrow range of variation with regard to
the H1-H2 measure. This implies that the level of the
fundamental in the source spectrum was kept rather
constant across loudness and pitch conditions. This
may be relevant to the timbral similarity of tones that
differ in pitch and loudness. In other words, the small
variability of H1-H2 may re ect a striving for equali-
sation of voice timbre in these singers.
CONCLUSIONS
The voice source differed in various ways between
falsetto and modal register. Some differences were
similar for counter tenors, tenors, and baritones.
Thus, in falsetto, as compared with modal register,
subglottal pressure was lower and closed phase gener-
ally shorter, while relative glottal leakage and H1-H2
were mostly greater. The counter tenors used nar-
rower pressure ranges than the tenors and baritones,
particularly in the modal register and their closed
phase, relative glottal leakage and H1-H2 varied
within narrower ranges than in the tenors and
baritones. These Ž ndings are compatible with the
earlier observation that the vocal folds are thinner in
falsetto than in modal register, probably re ecting a
lesser activation of the thyroarytenoi d muscles. Thin-
ner vocal folds should require lower subglottal pres-

sures. Also, they would be associated with a smaller
phase lag between the lower and upper margins. This
generates a shorter closed phase and a more rounded
pulse form, which, in turn, produces a stronger
fundamental.
ACKNOWLEDGEMENTS
The kind cooperation of the singers is gratefully
acknowledged. Co-author CH’s participation in the
investigations was made feasible partly by a grant
from the Almenvidenskaplige ForskningsraÊ d,
Norway.
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
REFERENCES
1. Askenfelt A, GaufŽ n J, Sundberg J, Kitzing P. A
comparison of contact microphone and electroglot-
tograph for the measurement of vocal fundamental
frequency. J Speech Hearing Res 1980; 23: 258 –73.
2. Bickley CA. Acoustic Analysis and Perception of
Breathy Vowels. Speech Communication Group Work-
ing Papers. Cambridge, MA: Research Laboratory of
Electronics, MIT2, 1982. pp. 71 –81.
3. Colton RH. Spectral characteristics of the modal and
falsetto registers. Folia Phoniatrica 1972; 24: 337 –44.
For personal use only.
4. GaufŽ n J, Sundberg J. Spectral correlates of glottal
voice source waveform characteristics. J Speech Hear-
ing Res 1989; 32: 556 –65.
5. HertegaÊ rd S, GaufŽ n J, Lindestad P-AÍ . A comparison
of subglottal and intraoral pressure measurements dur-
ing phonation. J Voice 1995; 9: 149 – 55.
6. Hirano M. The function of the intrinsic laryngeal mus-
cles in singing. In: Stevens KN, Hirano M (editors).
Vocal Fold Physiology. Tokyo: University of Tokyo
Press, 1981. pp. 155 – 67.
7. Hirano M. The role of the layer structure of the vocal
folds in register control. In: Hurme P (editor). Vox
Humana. Jyväskylä, Finland: University of Jyväskylä,
1982. pp. 49 –62.
8. Hirano M, Vennard W, Ohala J. Regulation of regis-
ter, pitch and intensity of voice. Folia Phoniatrica
1970; 22: 1– 20.
9. Large J. An acoustical study of isoparametric tones in
the female chest and middle registers in singing. Natl
Assoc Teachers Singing Bull Dec. 1968; 24: 12 – 15.
10. Lindestad P-AÍ , Södersten M. Laryngeal and pharyn-
geal behavior in countertenor and baritone singing-a
videoŽ berscopic study. J Voice 1988; 2: 132 –9.
11. Mörner M, Fransson F, Fant G. Voice register termi-
nology and standard pitch. Speech Transmission Labo-
ratory Quarterly Progress and Status Report 4. UTH,
Stockholm, 1963: 17 – 23.
12. Rothenberg M. A new inverse Ž ltering technique for
deriving the glottal air ow waveform during voicing. J
Acoust Soc Am 1973; 53: 1632 – 43.
13. Roubeau B, Chevrie-Muller C, Arabia-Guidet C. Elec-
troglottographic study of the changes of voice registers.
Folia Phoniatrica 1987; 39: 280– 9.
14. Schutte HK, Miller R. Resonance balance in register
categories of the singing voice: a spectral analysis
study. Folia Phoniatrica 1984; 36: 289–95.
Voice source in modal and falsetto register 35
15. Stevens K, Hanson H. ClassiŽ cation of glottal vibra-
tion from acoustic measurements. In: Fujimura O,
Hirano M (editors), Vocal Fold Physiology: Voice
Quality Control. San Diego, CA: Singular Publishing
Group; 1995:147 –70.
16. Sundberg J. The source spectrum in professional
singing. Folia Phoniatrica 1973; 25: 71 –90.
17. Sundberg J, Andersson M, Hultqvist C. Effects of
subglottal pressure variation on professional baritone
singers’ voice sources. J Acoust Soc Amer 1999; 105:
1965– 71.
18. Södersten M, Högset C, Sundberg J. Glottal air ow
and subglottal pressure in the falsetto and modal regis-
ters of counter tenor singers. In: Friberg A, Iwarsson J,
Jansson E, Sundberg J (editors). Proceedings of SMAC
93 (Stockhom Music Acoustic Conference 1993. Stock-
holm: Royal Swedish Academy of Music, Publication
No. 79, 1994. pp. 249–53.
19. Södersten M, Lindestad P-AÍ . Geographic and cultural
differences in singing. Acta Phoniatrica Latina 1987; 9:
19 –22.
20. Titze I. Priciples of Voice Production. Englewood
Cliffs, NJ: Prentice Hall, 1994.
21. van den Berg J-W. Vocal ligaments versus registers.
Natl Assoc Teachers Singing Bull Dec 1963; 31: 16 –21.
SAMMANFATTNING
Skillnader i röstkällan mellan falsett - och
modalregister hos counter tenor, tenor, och
barytonsaÊ ngare
Röstens registerfunktion antas härröra fraÊ n röst-
källan, d v s det pulserande transglottala luft ödet.
VaÊ gformen av detta luft öde, floÈ desglottogrammet,
analyserades hos professionella saÊ ngare genom inver-
sŽ ltering av  ödessignalen registerad med hjälp av en
 ödesmask. Försökspersonerna, som tillhörde röst-
kategorierna counter tenor, tenor, och baryton, sjöng
en serie av stavelsen [pæ:] i laÊ g, neutral och hög
röststyrka paÊ samma tonhöjder i modal- och falsett:
counter-tenor-register . Subglottala trycket uppmättes
som oraltrycket under occlusionen för konsonanten
[p]. FraÊ n  ödesglottogramme n bestämdes sluten-
kvoten, det relativa glottala läckaget, och
nivaÊ skillnaden mellan röstkällans första och andra
delton. Jämförelser av toner sjungna paÊ samma ton-
höjd i de baÊ da registren visade att subglottala trycket
var lägre i falsettregister. Pulsformen var dessutom
mera rundad, men vaÊ gformen visade i de  esta fall en
klart markerad slutenfas. Slutenfasen var vidare
kortare, relativa läckaget vanligen större och grund-
tonshalten högre än i modalregister. Jämfört med
tenorerna och barytonsaÊ ngarna använde counter-
tenor-saÊ ngarna lägre subglottalt tryck i baÊ da reg-
istren, och i deras falsett var slutenfasen kortare, och
grundtonshalte n varierade mindre med tonhöjd och
röststyrka. För ett givet värde paÊ slutenkvoten ten-
derade grundtonen att vara starkare i falsett än i
Log Phon Vocol 26
 36 J. Sundberg and C. Högset
modalregister. De observerade olikheterna i röst-
källan mellan registren skulle kunna förklaras av att
stämbanden i modalregister är tjockare; detta borde
leda till ökad fördröjning av slutningen av det övre
jämfört med det undre skiktet av glottis och paÎ saÊ vis
förlänga slutenfasen och göra pulsen mindre rundad.
YHTEENVETO
Äänilähteen erot falsetti- ja modaalirekisterissä kon-
tratenoreilla, tenoreilla ja baritoneilla
AÈ änen rekisteriominaisuuden oletetaan saavan
alkunsa äänilähteestä eli värähtelevästä transglot-
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Laurentian University on 04/22/13
taalisesta ilmavirrasta. Tämän ilmavirran aal-
tomuoto, virtausglottogra Ž a, analysoitiin ammat-
tilaulajien virtaussignaalist a käänteissuodatus-
menetelmällä. Apuna käytettiin virtausmaskia.
Tutkimushenkilöt, jotka kuuluivat äänialoihin kon-
tratenori, tenori ja baritoni, lauloivat sarjan tavua
[pæ:] hiljaisella, normaalilla ja voimakkaalla äänellä
samoilla äänenkorkeuksilla modaali – ja falsetto:
kontratenori – rekistereissä. Subglottaalinen paine
mitattiin kuten oraalipaine konsonantti [p]:n sulun
aikana. Virtausglottogrammista pääteltiin loppuo-
For personal use only.
Log Phon Vocol 26
suus, relatiivinen glottaalinen vuotoisuus ja tasoero
äänilähteen ensimmäisen ja toisen osasävelen välillä.
Samalla äänenkorkeudella laulettujen sävelten ver-
tailu molemmissa rekistereissä osoitti, että subglot-
taalinen paine oli matalampi falsettorekisterissä.
Impulssimuoto oli sen sijaan loivempi, mutta aal-
tomuoto näytti useimmissa tapauksissa selvästi erot-
tuvaa loppuvaihetta. Loppuvaihe oli myös lyhyempi,
suhteellinen vuotoisuus yleensä suurempi ja perus-
sävel korkeampi kuin modaalirekisterissä. Kun ver-
taillaan tenori – ja baritonilaulajia, niin
kontratenorit käyttivät pienempää subglottaalista
painetta molemmissa rekistereissä ja heidän falset-
tossaan loppuvaihe oli lyhyempi sekä vuotoisuus ja
perussävelarvo vaihtelivat vähemmän sävelkorkeu-
den ja äänen voimakkuude n mukana. Myös perus-
sävel vaikutti olevan voimakkaampi falsetossa kuin
modaalirekisterissä. Nämä havaitut eroavaisuude t
äänilähteessä eri rekistereissä voisi selittää sillä, että
äänihuulet ovat modaalirekisterissä paksummat, jol-
loin sulun viivästys kasvaa ylemmässä glottiksen ker-
rostumassa verrattuna alempaan kerrostumaan ja
pidentää siten loppuosuutta ja tekee impulssin vä-
hemmän loivaksi.