Perturbation of Voice Signals in Register Transitions

on Sustained Frequency in Professional Tenors

*Matthias Echternach, *,†Louisa Traser, and *Bernhard Richter, *yFreiburg, Germany

Summary: Purpose. Vocal register transitions in the passaggio region remain an unclarified field in classically
trained male singers.
Method. We examined the acoustic and electroglottographic signals of seven tenors’ transitions from voix mixte to
falsetto on a sustained pitch F4 (349 Hz) on the vowels /a, e, i, o, u, and æ/.
Results. It was found that in many of the tested subjects, register transitions between voix mixte and falsetto were
performed very continuously without clear register transition events. However, an increase of frequency and amplitude
perturbation (jitter, relative average perturbation, and shimmer) was observed during register transitions.
Conclusion. These data suggest that professional tenors are able to avoid sudden registration events frequently ob-
served in untrained voices.
Key Words: Register–Modal–Transition–Falsetto–Voix mixte–Jitter–Shimmer.

INTRODUCTION The increase of perturbation during ascending frequency

Vocal registers can be defined and characterized based on me- glides on the one side, and the lack of perturbation during de-
chanical principles and/or sound characteristics.1–7 Both scending frequency glides on the other, lead to the question;
kinds of register definitions are based on the assumption that what happens if register transitions are performed on the
register-associated vocal properties can only be continued same F0? In contrast to untrained voices professional voices
over a definite voice range and can be separated from each other are mostly able to perform register transitions on sustained
by means of differences between the mechanical principle and/ F0. Since jitter—in contrast to RAP—could be affected by
or the sound characteristics, respectively. Vocal registers can long-term frequency changes,17 it seems additionally of interest
overlap in their fundamental frequency (F0) range.8 In this to assess whether perturbation values are increased independent
overlapping range, voluntary or involuntary register transition from frequency changes in register transitions.
events might occur.9–11 For tenors in particular, singing in In 1972, Large et al10 were analyzing register transitions
and above the region where register events between modal from full head (the term full head is used quite similarly to
register and falsetto usually occur (passaggio) belongs to the the term voix mixte in the present study) to falsetto on sus-
most difficult tasks of voice training. In contrast to tained pitches. In their study on two tenor, two baritone, and
unmodulated falsetto, this register function for professional one bass subject, they were able to document differences of
singing above passaggio has previously been denoted as voix the overtone spectrum and flow. However, the examined sub-
mixte, Volltonstimme, Vollstimme der Bruststimme, or full jects produced their changes on different F0s and those sub-
head.10,12–15 In the present study, the term voix mixte is jects were not asked to blend acoustic differences between
chosen, although this term is sometimes used with slightly the registers, thus, resulting in what the authors denoted as
different definition.16 ‘‘breaks.’’ Register differences were also observed by studying
On the voice source level, it was shown that for ascending reg- electroglottographic (EGG) voice signals in register transitions
ister transitions from modal to falsetto register perturbation in glissandi and sustained pitches in trained and untrained voi-
values, such as jitter, relative average perturbation (RAP), ces by Roubeau et al18 and Henrich et al,19 respectively. The
and shimmer, were increased for both untrained male voices authors observed changes of EGG open quotient and ampli-
and professional tenors.15,17 However, while performing tude. However, also in these studies, the subjects were not ex-
descending transitions, there was a lack of such an increase for plicitly advised to smooth or blend register transition as much
frequency perturbation in the professional singer subjects.15
Interestingly, comparable increases of frequency perturbation
measures were found for transitions from modal to voix mixte
TABLE 1.
function in professional tenors.15
Subjects, Age, Profession, and Taxonomy According to
the Bunch and Chapman29 Criteria
Subject Age Profession Taxonomy
Accepted for publication February 22, 2012.
From the *Institute of Musicians’ Medicine, Freiburg University Medical Center, Frei- 1 34 Professional choir 2.15
burg, Germany; and the yDepartment of Otorhinolaryngology, Head and Neck Surgery, 2 45 Soloist, freelancer 3.17
Freiburg University Medical Center, Freiburg, Germany.
Address correspondence and reprint requests to Matthias Echternach, Institute of Musi- 3 24 Soloist, opera house 3.1a
cians’ Medicine, Freiburg University Medical Center, Breisacher Str. 60, 79106 Freiburg, 4 30 Soloist, opera house 3.1a
Germany. E-mail: matthias.echternach@uniklinik-freiburg.de
Journal of Voice, Vol. 26, No. 5, pp. 674.e9-674.e15
5 33 Soloist, opera house 3.1a
0892-1997/$36.00 6 36 Soloist, opera house 4.1a
Ó 2012 The Voice Foundation 7 30 Soloist, opera house 4.1a
doi:10.1016/j.jvoice.2012.02.003
Matthias Echternach, et al Perturbation of Voice Signals in Professional Tenors 674.e10

FIGURE 1. F0 and contact quotient data plotted against time. The construction of initial part, transition part, and final part window are shown. F0,
fundamental frequency.

as possible. In 1982, Kitzing20 was analyzing register transi-
tions in untrained subjects and a single tenor subject. Here,
the tenor subject was asked to perform register transitions on
musical scales and equalize register differences. However, he
was only able to do so for his transition to what the author de-
noted as operatic head register but not for falsetto register.20 In
fact, blending of such acoustic differences is of great interest
in register research. It is assumed that the lowering of intensity
with an increase of open quotient before the register shift be-
tween modal register and falsetto could provide a solution for
blending.19 Because for male western classical singing mostly
tenors have to phonate in the passaggio region, it seems of
interest if professional tenors are able to blend differences as-
sociated with registers, such as voix mixte and falsetto,
respectively.
The present study was performed to analyze register transi-
tions from voix mixte to falsetto on sustained F0 inside the pas-
saggio region. It was hypothesized that in professional tenors
perturbation values are increased during register transitions
and none of the subjects were able to blend register borders ab-
solutely by means of total continuous changes of voice signals,
both audio and EGG signals.

FIGURE 2. Two examples (A and B, respectively) of transitions from voix mixte to falsetto with clear register transition event. F0, fundamental
frequency; SPL, sound pressure level; EGG, electroglottogram.
674.e11 Journal of Voice, Vol. 26, No. 5, 2012

F0 SPL
400
100
90
375 80
70

F0 (Hz)
60

SPL (dB)
350
50
40
325 30
20
10
300 0
Voix mixte Transition Falsetto Voix mixte Transition Falsetto

Contact quotient
70
60
50
CQ (%)

40
30
20
10
0
Voix mixte Transition Falsetto

FIGURE 3. Mean values and standard deviation for F0, SPL, and CQ. Statistical P values are shown in Table 2. F0, fundamental frequency; SPL,
sound pressure level; CQ, contact quotient.

MATERIALS AND METHODS The Netherlands). Like in our previous studies,15,17 in all
Seven professional opera tenors, all without voice complaints, sequences, the attempt was made to detect audible register
served as subjects (Table 1). Vocal fold pathologies were ex- transitions by three experts.
cluded by videostroboscopic or laryngoscopic high-speed im- The recorded sequences were divided into three sections,
aging in all of the subjects. each 500 milliseconds: (1) the initial voix mixte part, (2) the
The subjects were asked to first sustain a tone (F4, 349 Hz) in register transition part, and (3) the final falsetto part. Initial
their stage performance voice above the passaggio (voix mixte) part and final part windows were taken from the signal where
and then perform a transition to falsetto on the same F0 without audio and EGG contact quotient (CQ) were quasi constant. Be-
major changes of sound pressure level (SPL), avoiding vibrato cause in our material, no clear point of audible register transi-
as far as was feasible and with as much blending of register dif- tions were found by the experts, we used CQ for building the
ferences during the transition as possible. The loudness was transition part window: In previous investigations, it was shown
self-selected by each subject. The task was performed on the that CQ and open quotient (¼1  CQ) is different in different
vowels /a/, /e/, /i/, /o/, /u/, and /æ/. register conditions.4,15,17–19,21,22 Therefore, it could be
All recordings were performed in a not sound-treated room, expected that during the transition part of the experiment, CQ
which is usually used as a voice laboratory for clinical voice will fall. The midpoint of the transition part window in time
evaluation. Acoustic and EGG recordings were made with the scale was set at the point where the CQ difference between
Laryngograph (Fa. Laryngograph, London, UK) as described initial part (maximum) and final part (minimum) reaches
previously.15,17 The analysis of the signals was performed with 50%, as shown in Figure 1. The window margins were set at
Speech Studio software (Fa. Laryngograph, London, UK), ±250 milliseconds.
Soundswell Signal Workstation (Saven Hitech, T€aby, Sweden), The laryngograph signal was analyzed with respect to F0 and
and PRAAT software (University of Amsterdam, Amsterdam, CQ, applying a 30% above the base criterion, as provided by the
manufacturer. Like in previous studies,15,17 SPL was evaluated
by using the headset microphone with a distance of 3 cm at the
lateral mouth angle. Calibration was performed with a sound
TABLE 2. pressure meter using the Soundswell Software. The Speech
Statistical P Values for the Mean Values of Figure 3 Studio multidimensional voice profile, including automatic
P Values calculation of jitter, RAP, and shimmer, was established for
the sections just described. These calculations are based on
Voix Mixte– Voix Mixte– Transition–
definitions proposed in previous investigations.23
Parameters Transition Falsetto Falsetto
For the statistical evaluation, the changes from initial part (t1)
F0 <0.01 Not significant <0.01 to transition part (t2) and to final part (t3) were measured apply-
SPL <0.001 <0.001 Not significant ing parametric, multivariate variance analyses (general linear
CQ <0.001 <0.001 <0.001
model repeated measurement procedure in SPSS 15.0 [SPSS,
Abbreviations: F0, fundamental frequency; SPL, sound pressure level; CQ, Chicago, IL]) for each vowel. When significant multivariate
contact quotient.
analyses were present, the individual results of the univariate
Matthias Echternach, et al Perturbation of Voice Signals in Professional Tenors 674.e12

analyses were checked for significant progressions, and the reg-
isters were subjected to pair comparison (Bonferroni). The gen-
eral level of significance was set at a ¼ 0.05.
RESULTS
All 42 sequences (7 subjects 3 6 vowels) were analyzed with
respect to F0 in each of the three sections. There were no statis-
tically significant differences between the mean values for the
vowel conditions in each of the variables.
Although the comparisons of voix mixte and falsetto in the
stable parts showed perceptual differences, it was not possible
to identify a clear point of register shift during the transition
part. Voice breaks or major pitch jumps were not observed in
any of the sequences. In a few sequences, register transitions
were accompanied by sudden changes of F0 (Figure 2A), which
resulted in an increased mean F0 of 8 Hz which is less than
one semitone (approximately 20 Hz at 349 Hz, Figure 3).
Also, EGG amplitude or CQ was changed in some subjects
(Figure 2B). However, most of the sequences were very conti-
nuous concerning audio signal, EGG signal, and with a conti-
nuous decrease of CQ (Figure 4). All sequences showed
quasi-periodic voice signals (according to Titze24 type 1 voice
signals) without F0 discontinuites, chaotic voice signals, or sub-
harmonics (type 2 and 3 voice signals) during the transition part.
Although the subjects were advised to keep SPL as constant
as possible, SPL was statistically significantly lowered by 4 dB
from initial voix mixte part to final falsetto part (Figure 3).
Frequency perturbation values, both jitter and RAP, were
increased during the transition part window (Figure 5). The

FIGURE 4. Example of a register transition from voix mixte to falsetto without clear sudden register transition event. F0, fundamental frequency;
CQ, contact quotient; EGG, electroglottogram.
674.e13 Journal of Voice, Vol. 26, No. 5, 2012

Jitter RAP
2
2.5
1.5
2

RAP (%)
Jitter (%)
1.5 1

1
0.5
0.5

0 0
Voix mixte Transition Falsetto Voix mixte Transition Falsetto

Shimmer
7
6
5
Shimmer (%)

4
3
2
1
0
Voix mixte Transition Falsetto

FIGURE 5. Mean values and standard deviation for jitter, RAP, and shimmer. Statistical P values are shown in Table 3. RAP, relative average
perturbation.

values for the final falsetto part remain on a higher level
compared with the initial voix mixte part. Also, shimmer
was increased during the transition part window (Figure 5).
However, the differences in values for voix mixte were not sta-
tistically significant.
As can be seen in Figure 6, jitter and RAP were increased
with lower SPL. In contrast, shimmer showed no systematic be-
havior in relation to SPL.
DISCUSSION
This study presents the perturbation and EGG data of profes-
sional tenors’ register transitions on sustained F0 in the passag-
gio region. Here, it is shown that transitions between voix mixte
and falsetto are associated with no major irregularities, lacking
voice breaks or major F0 jumps. In contrast, most of the sub-
jects were able to perform register shifts without clear detect-
able register transition events, such as peaks in CQ course,
sudden SPL shifts, or pitch jumps. However, even here, fre-
quency perturbation values, such as jitter and RAP, were signif-
icantly increased during register transition.
TABLE 3.
Statistical P Values for the Mean Values of Figure 5
P Values
Voix Mixte– Voix Mixte–
Parameters Transition Falsetto
Jitter <0.001 <0.01
RAP <0.001 <0.01
Shimmer Not significant Not significant
Abbreviation: RAP, relative average perturbation.
Singing in the passaggio region and above is one of the most
challenging skills for professional male singers, especially for
tenors. In general, when singing in this frequency range, three
options seem to be possible: unmodulated falsetto, modulated
male altos’ falsetto, and tenors’ voix mixte.10,15 The voix
mixte function—defined as stage voice above passaggio in
this study—remains unclear because potentially this vocal
technique could be developed from two different directions,
modal and falsetto. It was shown before that voix mixte
(defined by phonation in the shared zone of laryngeal
mechanisms) could be performed from both laryngeal
mechanisms, 1 (quite comparable with modal register) and 2
(quite comparable with falsetto register).16 However, the au-
thors found spectral overtone characteristics to allocate the sig-
nals to either one or the other mechanism.16 Such differences
could not be ascertained in the presented tenors’ voix mixte
functions. Although the authors believe that, for most subjects,
voix mixte was built outgoing from modal register, it seemed
not possible to clearly allocate the origin of voix mixte to
a modal or a falsetto register.
Henrich et al4,18,19 observed clear register transition events in
all of their untrained and trained subjects. Earlier, Large et al10
found clear register transition events in an experiment analyz-
ing register transitions from full head (quite comparable to
voix mixte in the presented study) to falsetto in sustained
pitches. However, the subjects were not especially advised to
equalize register differences as much as possible. In the present
Transition– study, only a few sequences were associated with clear register
Falsetto transition events, such as sudden changes of SPL or F0. But
even those changes were not clearly perceived by the experts.
<0.05
<0.05
The increase of mean values for F0 was much lower than one
<0.05 semitone and therefore in the usual range of professional clas-
sical singers’ vibrato25 (although subjects were instructed to
avoid vibrato as much as possible). In contrast, in the present
Matthias Echternach, et al Perturbation of Voice Signals in Professional Tenors 674.e14

Jitter/SPL RAP/SPL
4.5 Initial VM part
Initial VM part 4.5
4 Transition part Transition part
4
3.5 Final falsetto part Final falsetto part
Potential (transition part) 3.5
3 Potential (final falsetto part)
Jitter (%)

RAP (%)
Potential (initial VM part)
2.5 2.5
2 2
1.5 1.5
1 1
0.5 0.5
0 0
50 60 70 80 90 100 110 120 50 60 70 80 90 100 110 120
SPL (dB) SPL (dB)

Shimmer/SPL
Initial VM part
25 Transition part
Final falsetto part
20
Shimmer (%)

15

10

0
50 70 90 110 130
SPL (dB)
FIGURE 6. Jitter, RAP, and shimmer plotted against SPL for all the three analysis windows. RAP, relative average perturbation; SPL, sound pres-
sure level.

study, we observed, in general, many transitions from one to an-
other register without any clear register transition event. For ex-
ample, as shown in Figure 4, only CQ was changing. Especially,
SPL and F0 remained nearly constant. If we assume that the
register transition occurs in this figure during the time of CQ
change (approximately 300 milliseconds)—we would expect
approximately 100 glottal cycles for F4 (349 Hz). During this
relatively long time interval, it was not possible to allocate
the transition to one of the mechanical principles. In other
words, if we assume that the initial voix mixte part is the color
black and final falsetto part is the color white, we found a large
gray scale in between where we were not able to detect where
black ends or white begins. These cases therefore question
whether there are clearly separable mechanical principles dur-
ing the transition part.
Because for many untrained voices register transitions were
associated with major irregularities during the register transi-
tion from modal to falsetto,17 it seems striking that the blending
of register differences in the recent material is a consequence
of vocal training. It was assumed that smoothing of register
transitions could be performed by lowering SPL in front of
a register transition from laryngeal mechanism 1 to laryngeal
mechanism 2.19 The consequence would be an increase of
the open quotient in the first mechanism to diminish the open
quotient difference between these laryngeal mechanisms. Al-
though some SPL courses during the transitions showed such
a tendency, our material failed to confirm this hypothesis for
all cases of smooth transitions. In general, SPL levels showed
more unsystematic courses. As shown in Figure 4, SPL was
sometimes even increased slightly before the CQ decrease
started. Furthermore, SPL differences between register func-
tions were quite small.
Besides the question of register transition events, we found
an increase of perturbation values during the register transition.
Therefore, even for quite smooth courses of voice signals, the
register transition could not be regarded as totally stable. Al-
though the subjects in the present study are different to a previ-
ous study, the frequency perturbation values for the sustained
pitch task are located in the middle, between the values of as-
cending (from modal register to falsetto), and descending reg-
ister transitions (from falsetto to modal register) in other
professional tenors.15 In the previous study, jitter and RAP in-
crease was found only in ascending but not descending register
transitions from modal to falsetto register.15
In the present study, subjects were advised to keep SPL as
constant as possible. However, falsetto was associated with
lower SPL than voix mixte. Such difference could have an influ-
ence on frequency perturbation values. Brockmann et al26 ob-
served a jitter increase with low SPL. The same tendency was
found in our data. Thus, in part, the higher jitter and RAP values
in the falsetto might be explained by this phenomenon. No such
tendency was found for the shimmer in our data and the differ-
ences of the values between voix mixte and falsetto failed to
reach statistical significance.
674.e15
Although there were no clear points of register shifts during
the transition part, we found clear perceptual differences in the
stable parts between voix mixte and falsetto. Here, it should be
kept in mind, however, that these differences might be not only
related to the laryngeal oscillation patterns. It has been shown
earlier that professional tenors use different vocal tract config-
urations for their voix mixte compared with their falsetto12,27,28
and that formant frequencies also differ between these register
functions.28 Therefore, in part, perceptual differences between
these registers might be also caused by differences in the vocal
tract configurations and formants.
CONCLUSION
Professional tenors are able to perform very smooth register
transitions from voix mixte to falsetto on a sustained pitch
lacking clear sudden registration events in the voice signals,
however, accompanied with a small increase of perturbation
values. During these smooth register transitions, it seems diffi-
cult to identify clear register borders in the voice signals.
Acknowledgments
The authors would like to thank Mark F. Zander, PhD, for his
help in statistical analysis and Johan Sundberg, PhD, for discus-
sion about the material.
REFERENCES
1. Garcia M. Memoire sur la voix humaine presente a l’Academie des Sciences
en 1840. [Memory of the Human Voice Presented at the Academy of
Sciences in 1840]. 2nd ed. Paris, France: Imprimerie d’E Duverger; 1847.
2. Titze IR. Principles of Voice Production. Upper Saddle River, NJ: Prentice
Hall; 1994.
3. Nadoleczny M. Untersuchungen €uber den Kunstgesang. [Examinations on
Artistic Singing]. Berlin, Germany: Springer Verlag; 1923.
4. Henrich N. Mirroring the voice from Garcia to the present day: some
insights into singing voice registers. Logoped Phoniatr Vocol. 2006;31:
3–14.
5. M€uller J. Handbuch der Physiologie des Menschen f€ur Vorlesungen [Hand-
book of Human Physiology for Lectures]. Coblenz, Germany: Verlag von J.
H€olscher; 1840.
6. Van den Berg JW. Vocal ligaments versus registers. NATS Bull. 1963;20:
16–21.
7. Large J. Towards an integrated physiologic acoustic theory of vocal regis-
ters. NATS Bull. 1972;28:18–36.
8. Roubeau B, Castellengo M, Bodin P, Ragot M. Phonetogramme par registre
larynge [Laryngeal registers as shown in the voice range profile]. Folia Pho-
niatr Logop. 2004;56:321–333.
9. Svec JG, Schutte HK, Miller DG. On pitch jumps between chest and
falsetto registers in voice: data from living and excised human larynges.
J Acoust Soc Am. 1999;106:1523–1531.
10. Large J, Iwata S, von Leden H. The male operatic head register versus fal-
setto. Folia Phoniatr (Basel). 1972;24:19–29.
Journal of Voice, Vol. 26, No. 5, 2012
11. Za~nartu M, Mehta DD, Ho JC, Wodicka GR, Hillman RE. Observation
and analysis of in vivo vocal fold tissue instabilities produced by non-
linear source-filter coupling: a case study. J Acoust Soc Am. 2011;129:
326–339.
12. Echternach M, Sundberg J, Markl M, Richter B. Professional opera tenor’s
vocal tract configurations in registers. Folia Phoniatr Logop. 2010;62:
278–287.
13. Miller DG. Registers in Singing: Empirical and Systematic Studies in the
Theory of the Singing Voice. Groningen, The Netherlands: University of
Groningen; 2000.
14. Luchsinger R, Arnold G. Voice-Speech-Language. Belmont, CA: Wads-
worth; 1965.
15. Echternach M, Richter B. Professional tenors’ passaggio—evaluation of
perturbation values. J Voice. 2011;. Epub ahead of print.
16. Castellengo M, Chuberre B, Henrich N. Is voix mixte, the vocal technique
used to smooth the transition across the two main laryngeal mechanisms, an
independent mechanism? Proceedings of the International Symposium on
Musical Acoustics; March/April 2005, Nara, Japan.
17. Echternach M, Sundberg J, Zander MF, Richter B. Perturbation measure-
ments in untrained male voices’—transitions from modal to falsetto regis-
ter. J Voice. 2011;25:663–669.
18. Roubeau B, Henrich N, Castellengo M. Laryngeal vibratory mechanisms:
the notion of vocal register revisited. J Voice. 2009;23:425–438.
19. Henrich N, D’Alessandro C, Doval B, Castellengo M. Glottal open quotient
in singing: measurements and correlation with laryngeal mechanisms,
vocal intensity, and fundamental frequency. J Acoust Soc Am. 2005;117:
1417–1430.
20. Kitzing P. Photo- and electroglottographical recording of the laryngeal
vibratory pattern during different registers. Folia Phoniatr (Basel). 1982;
34:234–241.
21. Echternach M, Dippold S, Sundberg J, Arndt S, Zander M, Richter B.
High-speed imaging and electroglottography measurements of the open
quotient in untrained male voices’ register transitions. J Voice. 2010;24:
644–650.
22. Roubeau B, Chevrie-Muller C, Arabia-Guidet C. Electroglottographic
study of the changes of voice registers. Folia Phoniatr (Basel). 1987;39:
280–289.
23. Baken RJ, Orlikoff RF. Clinical Measurement of Speech and Voice. 2nd ed.
San Diego, CA: Singular Publishing Group; 2000.
24. Titze IR. Workshop on Acoustic Voice Analysis: Summary Statement. Iowa
City, IA: National Center for Voice and Speech, Iowa University; 1995.
1–36.
25. M€urbe D, Zahnert T, Kuhlisch E, Sundberg J. Effects of professional sing-
ing education on vocal vibrato—a longitudinal study. J Voice. 2007;21:
683–688.
26. Brockmann M, Storck C, Carding PN, Drinnan MJ. Voice loudness and
gender effects on jitter and shimmer in healthy adults. J Speech Lang
Hear Res. 2008;51:1152–1160.
27. Echternach M, Sundberg J, Arndt S, Breyer T, Markl M, Schumacher M,
Richter B. Vocal tract and register changes analysed by real time MRI in
male professional singers—a pilot study. Logoped Phoniatr Vocol. 2008;
33:67–73.
28. Echternach M, Sundberg J, Baumann T, Markl M, Richter B. Vocal tract
area functions and formant frequencies in opera tenors’ modal and falsetto
registers. J Acoust Soc Am. 2011;129:3955–3963.
29. Bunch M, Chapman J. Taxonomy of singers used as subjects in scientific
research. J Voice. 2000;14:363–369.