ORIGINAL ARTICLE
Mirroring the voice from Garcia to the present day: Some insights
into singing voice registers
NATHALIE HENRICH
Laboratoire d’Acoustique Musicale, Paris, France
Abstract
Starting from Garcia’s definition, the historical evolution of the notion of vocal registers from then until now is considered.
Even though much research has been carried out on vocal registers since then, the notion of registers is still confused in the
singing voice community, and defined in many different ways. While some authors consider a vocal register as a totally
laryngeal event, others define it in terms of overall voice quality similarities. This confusion is reflected in the multiplicity of
labellings, and it lies in the difficulty of identifying and specifying the mechanisms distinguished by these terms. The concept
of laryngeal mechanism is then introduced, on the basis of laryngeal transition phenomena detected by means of
electroglottography. It helps to specify at least the laryngeal nature of a given singing voice register. On this basis, the main
physiological, acoustic, and perceptual characteristics of the most common singing voice registers are surveyed.
Key words: Laryngeal mechanism, singing voice, vocal register
Introduction
Before the nineteenth century, knowledge about
human phonation was limited to what could be
perceived by auditory perception and proprioceptive
sensations, or observed from outside during phona-
tion and inside on cadavers. The singing voice
registers are a known fact, related to perception of
different voice qualities. In this context, the singing
voice teacher Manuel Garcia II (1805
/1906) shows
not only an interest in the voice sound quality but
also in the underlying vocal physiology mechanisms.
On November 16th, 1840, he presented the results
of his observations as a voice teacher and his
experiments on the human voice to the French
Acade
´
mie des Sciences (1), prior to the publication
of his famous singing voice method Ecole de Garcia:
Traite
´
complet de l’art du chant (2). In this paper, he
claims that the human voice is composed of different
registers: poitrine (chest), fausset-te
ˆ
te (falsetto-head),
and contre-basse (counter bass). He defines the
registers as follows: ‘By the word register we mean
a series of consecutive and homogeneous tones going
from low to high, produced by the same mechanical
principle, and whose nature differs essentially from
another series of tones equally consecutive and
homogeneous produced by another mechanical
principle. All the tones belonging to the same
register are consequently of the same nature, what-
ever may be the modifications of timbre or of the
force to which one subjects them’.
1
Therefore,
Garcia pioneers the definition of vocal registers in
relation to a given mechanical principle, indepen-
dently of any timbre variation. In his definition of
vocal registers, the concept of a mechanical principle
prevails over the perceptual dimension. Basing his
rationale on physiological observations of the larynx
position, he demonstrated that the falsetto and head
registers result from the same laryngeal mechanism,
their difference being a timbre effect.
2
Nowadays, we have experimental techniques to
which Garcia did not have access at that time. How
has the knowledge about singing voice registers
evolved since Garcia’s time? How are the vocal
registers defined and identified? What are their
physiological, acoustic and perceptual attributes?
By reviewing the historical evolution of the notion
of vocal register, this paper aims to provide a possible
Correspondence: Dr. Nathalie Henrich, Institut de la Communication Parle´e, 46 avenue Fe´lix Viallet, 38031 Grenoble Cedex 01, France. E-mail:
henrich@icp.inpg.fr
Logopedics Phoniatrics Vocology. 2006; 31: 3 /14
ISSN 1401-5439 print/ISSN 1651-2022 online # 2006 Taylor & Francis
DOI: 10.1080/14015430500344844
guide for a better understanding of this notion. A
brief overview of the evolution of definition, num-
bers and labellings will be given. The different means
of identifying a vocal register will be discussed, and
the concept of laryngeal mechanism will be intro-
duced, on the basis of laryngeal transition phenom-
ena. Finally, the most common singing voice
registers will be reviewed, and their physiological,
acoustic and perceptual properties will be given.
Historical evolution of the notion of vocal
register since Garcia
From Garcia to the present time, the notion of vocal
registers has evolved in many ways. In particular, this
evolution concerns the definition of a vocal register,
the number of registers identified and their labelling.
From Garcia to the 1960s
As mentioned in the introduction, the singing
teacher Manuel Garcia in 1847 is considering three
main different registers, chest , falsetto-head and
contre-basse , the other possible registers, such as
voix mixte, sharing similar mechanical principle
with these main ones (for a detailed discussion about
Garcia’s falsetto-head register, we refer the reader to
(3)). The frequency ranges of these registers are
shown in Figure 1. Interested in understanding the
physiology of these registers, Garcia explores his own
larynx and those of a few of his students by using a
laryngoscope, a clever observation tool which would
give rise to our present endoscopic camera. It
consists of a small mirror fixed to the end of a long
handle and placed at the back of the throat (see
Figure 2). When the throat is lighted in the proper
way, the glottis may be observed from above. The
laryngoscope was already available in the early
nineteenth century *
/Babington used it in 1827
(4)*
/so Garcia did not invent it, but he was the
first to use it successfully, being able not only to see
the vocal folds at rest, but also in the act of singing.
Garcia’s observations were communicated to the
Royal Society by Dr Sharpey on May 24th, 1855
(4,5).
In 1880, the physiologist and voice production
teacher Emil Behnke and the throat surgeon Lennox
Browne also made use of the laryngoscope to obtain
in vivo images of the glottis. Behnke defines the vocal
registers as follows: ‘a register consists of a series of
tones which are produced by the same mechanism.
[...] There are, broadly speaking, three registers in
the human voice, and the mechanisms are plainly
visible, as follows: 1) During the lowest series of
tones the vocal ligaments vibrate in their entire
thickness. 2) During the next series of tones the
vocal ligaments vibrate only with their thin inner
edges. 3) During the highest series of tones a portion
of the vocal chink is firmly closed, and only a small
part of the vocal ligaments vibrates.’ ((4), p. 86). In
this definition, similarly to Garcia’s, the concept of
laryngeal mechanisms is underlying. On the basis of
these physiological observations, he adopts Curwen’s
thick, thin and small labelling, and he describes three
registers for the male voice (lower thick, upper thick
and upper thin) and five registers for the female voice
Figure 1. Frequency range of human voice and vocal registers, as defined by different authors: (A) Garcia (1), (B) Behnke (4), (C) Hollien
(11), (D) Miller (17).
4 N. Henrich
(lower thick, upper thick , lower thin , upper thin and
small). Their corresponding frequency ranges are
shown in Figure 1. The transition between thick and
thin registers is characterized by a break which
occurs around F4 or F4# in both male and female
voices, individual differences apart.
In the early twentieth century, whereas few singing
teachers maintain that a ‘natural’ voice has only one
single register (6,7), the general tendency is to accept
the existence of at least two and at maximum five
registers. The term ‘register’ is even sometimes
replaced by ‘mechanism’, for example by Wilcox
(6,7) who suggests the terms ‘heavy mechanism’ and
‘light mechanism’.
The terms used to label the vocal registers are
abundant and author dependent, and most of the
time their usage is ambiguous. In 1963, a literature
survey summary concerning the pitch range and
labelling of voice registers (8) mentions that ‘the only
secure common denominator for defining a register
is by means of its range on the musical scale.’
Reasonable agreement is found on ‘the average pitch
of the boundaries between registers, i.e. the breaks or
voice transitions.’ It is also mentioned that these
average boundaries do not vary much with the type
of voice, and that the nature of the breaks is ‘still a
matter of controversy’.
The two most common registers are labelled as
chest register and head or falsetto register. The
distinction between head and falsetto is very much
author dependent. In his time, Garcia placed the
falsetto register between chest and head register, and
it may thus correspond to a middle register, as
suggested by Vennard (7). Yet, when the head
register is recognized as a distinct register, it is very
often placed between chest and falsetto. Otherwise,
either it is put aside, or it replaces the falsetto
register, in the case of female voice for instance.
The two main registers overlap in a frequency region
where a third register is sometimes mentioned, and
labelled as medium , mid , middle ,ormixed register. It
may correspond to the French voix mixte or the
Italian voce mista . Two additional registers are
mentioned at the extreme low and high ends of the
frequency range: the Strohbass register, and the bell,
flute or whistle register.
From the 1960s to the 1980s
From the early 1960s to the 1980s, a good deal of
research effort was dedicated to the understanding of
the acoustic, mechanical, and physiological proper-
ties of these registers. Working on excised larynges,
the physicist Janwillem Van den Berg 1963 explored
the vibratory pattern of the vocal folds and the
influence of the subglottal resonance. He came to
the conclusion that the mid register may not be
considered as a distinct register (9). The physician
Minoru Hirano, together with two voice teachers
William Vennard and John Ohala, investigated the
role of intrinsic laryngeal muscles using electromyo-
graphy (10). They found that the vocalis muscle is
essential to register regulation, its activity being
greater in heavy register. In particular, the vocalis
muscular activity decreases during a register shift
from heavy to light and increases from light to heavy.
The lateral cricoarytenoid, interarytenoid and cri-
cothyroid muscles also contribute to register regula-
tion, but to a smaller extent.
In 1974, the speech scientist Harry Hollien
reconsiders the definition, numbering and terminol-
ogy of registers (11). He defines a vocal register ‘as a
series or range of consecutively phonated frequencies
which can be produced with nearly identical vocal
quality and that ordinarily there should be little or no
overlap in fundamental frequency between adjacent
registers. Furthermore, [he maintains] that a voice
register is a totally laryngeal event and, before the
existence of a particular register can be established,
it must be operationally defined: 1) perceptually,
2) acoustically, 3) physiologically and 4) aerodyna-
mically.’ With regard to this definition based on a
laryngeal entity, he considers three major registers,
which have been experimentally validated, and to
which he gives unfamiliar but also ‘uncontaminated’
labels: pulse , modal and loft registers. The pulse
register includes vocal fry, creak and strohbass
registers; the modal register includes chest , head,
low, mid and high; the loft register is the equivalent of
the phoneticians’ and speech pathologists’ falsetto
register. Their corresponding frequency ranges are
shown in Figure 1. Hollien mentions a fourth
register, the flute , whistle or pipe , which he puts
aside because of a lack of empirical information and
experimental exploration.
Figure 2. The laryngoscope. From (2).
Mirroring the voice from Garcia to the present day 5
Hollien’s significant paper has pointed out the
confusion around the concept of vocal registers
among the voice community, and it may have
contributed to the emergence of an international
discussion. In the late 1970s, spurred on by the
Collegium Medicorum Theatri (CoMeT), an inter-
national organization composed of physicians, voice
scientists, voice coaches and voice pathologists, a
committee on vocal registers was formed in an
attempt to clarify the notion of vocal registers and
to find a consensual position among the interna-
tional voice community. The committee agreed on
five points, which are reported by Hollien (12):
1. Registers exist: voice registers exist in the
speaking and singing voice, and they ‘must be
recognized as an entity’. Their perceived acous-
tic effects may be compensated for by appro-
priate training, if this ‘is considered desirable’.
2. Singing/speaking registers: ‘vocal registers in
singing and voice registers in speaking (or in the
untrained voice) are different and separate
entities and must be treated as such’. However,
they ‘may overlap in function and probably
have common physiological roots in the larynx’.
3. To ‘remove’ registers: two postulates are stated
by the committee. On the one hand, ‘eliminat-
ing or concealing register effects is desirable for
the classical/western, concert/opera mode of
singing’. On the other hand, ‘register effects
can be fundamental to certain types of singing’.
It is also important to keep in mind that ‘voice
registers cannot be removed as they are phy-
siologically given.’
4. The source of registers: First, the committee
has ‘accepted the notion that there probably are
two sources for registers *
/the larynx and the
vocal tract’. However, this point seems to have
raised a great debate among the committee
members. ‘A substantial minority of the com-
mittee argued in favor of the source (of a voice
register) being only laryngeal and that the other
so-called register-like phenomena actually are
some sort of quality/timbre events.’
This latter position is in line with Garcia’s
definition. Unfortunately, no mention of what
defines and identifies a register in the human
voice is given in the report. This may explain
the controversy about the question of the
source of registers. As a consequence, a great
debate was also raised by the labelling question.
5. Labels: The committee has rejected the use of
so called ‘old terms’, and in particular chest and
head which are based upon singers’ sensations.
Two main proposals are made. The first sug-
gestion is to number them: ‘#1: for the very
lowest of registers, probably used only in speak-
ing (old terms: pulse , vocal fry, creak), #2: that
(low) register, which is used for most speaking
and singing (old terms: modal, chest , normal,
heavy), #3: a high register used primarily in
singing (old terms: falsetto, light , head), #4: a
very high register usually found only in some
women and children and not particularly re-
levant to singing (old terms: flute , whistle).’ An
additional register is referred to as ‘#2A’ and
defined as ‘that ‘‘register’’ which is described by
many voice teachers as in the middle of the
frequency range [...] (old terms: head , mid,
middle, upper).’ The second suggestion is to
favour ‘new terms, generic terms, those that are
clear and easy to understand’, such as the pair
of terms heavy, light,orlower, upper .
From the 1980s to the present time
Despite this great effort at clarification, the voice
community remains divided on the question of vocal
registers. The existence of at least two main laryn-
geal vibratory mechanisms has been experimentally
validated. Among the speech community, an implicit
agreement is found on three registers (pulse or vocal
fry, modal or chest, and falsetto registers), in relation
to specific laryngeal adjustments. In the singing
voice community, the definition, numbers and
labelling of registers are still a matter of debate,
and they continue to vary among authors. Indeed,
vocal registers have an acoustical and perceptual
reality for singers, which cannot be ignored. On the
one hand, a vocal register is defined by its laryngeal
mechanical properties (13
/15), following in this
sense Garcia, Benhke and Hollien. As an example,
Sakakibara (15) proposes the following definition:
‘the vocal register is a set or range of serial sounds
that are similar in perception and produced by
similar vocal fold vibratory patterns.’ On the other
hand, a vocal register is defined by its characteristic
voice quality, following in this sense Large ((16),
cited in (17)). As an example, Titze mentions that
‘the term ‘‘register’’ has been used to describe
perceptually distinct regions of vocal quality that
can be maintained over some ranges of pitch and
loudness.’((18), p. 253) These two approaches of the
notion of vocal register are sometimes implicitly
combined. In a recent thesis on registers in singing
(17), the singer and voice teacher Donald Miller
points out the distinction between registration de-
fined ‘as an exclusive feature of the voice source’ and
the ‘integrated approach to registration, including
adjustments of the vocal tract, as well as those of the
source’. He chooses the latter approach, but defines
the ‘natural registers’ as registers related mainly to
6 N. Henrich
glottal source adjustments. These natural registers
are designated ‘chest’ and ‘falsetto’, and the author
refers to the ‘chest vibratory pattern’ and ‘falsetto
vibratory pattern’ when emphasizing the voice
source characteristics of a given register. According
to this author, the female voice is divided into five
registers: chest and belting registers, which both have
the chest vibratory pattern, and middle , upper and
flageolet registers, which all have the falsetto vibra-
tory pattern. The male voice is divided into four
registers: chest , full head and mezza voce registers,
which all have the chest vibratory pattern, and
falsetto register, which has the falsetto vibratory
pattern. The corresponding frequency ranges are
shown in Figure 1.
Conclusion: what are the vocal registers?
In his time, Garcia was distinguishing vocal registers
on the basis of a ‘mechanical principle’ which he
related to a laryngeal event. Following Garcia’s path,
other authors, such as Benhke, Hollien or Roubeau,
have explored the concept of laryngeal-related vocal
registers with the help of more and more sophisti-
cated experimental techniques: laryngoscopy, cine-
matography, electromyography, electroglottography
(EGG), etc. The understanding of laryngeal vibra-
tory mechanisms has become more refined.
Yet, a vocal register classification based only on
laryngeal phenomena does not take into account the
great timbre variations which can be achieved in
singing. This may explain that, even today, the
notion of vocal registers in singing remains very
often a matter of debate among the singing voice
community. Whereas some authors define the regis-
ters exclusively on the basis of an underlying
laryngeal mechanical principle, others prefer to
define them on the basis of sound voice quality. In
this context, to find a consensual position on the
labelling of registers is inevitably problematic, but
the core of this problem is less the terminology itself
than the ‘failure to identify and specify the mechan-
isms distinguished by the terms’, as pointed out by
Miller ((17), p. 32).
In the next section, we will see how the problem of
identifying the mechanisms related to singing voice
registers has been partly solved, at least at the
laryngeal level.
Transition phenomena: how to identify a
laryngeal mechanism
The reason for defining at least two registers in
human phonation comes from the discontinuities or
transition phenomena, which can occur voluntarily
or involuntarily during the production of voiced
sounds. Different kinds of transition phenomena
are listed in the literature, in particular periodicity,
timbre, and laryngeal transitions. They will be briefly
presented here. We will then focus on the laryngeal
vibratory mechanisms, the consistency of which is
evinced by laryngeal transition phenomena.
Detection of transition phenomena
A transition phenomenon may result from psychoa-
coustic properties of human perception. The ‘peri-
odicity transition’ distinguished by Titze (18,19) is
an example of such a transition. It refers to ‘changes
in vocal quality that occur whenever glottal pulses
are perceived as individual events rather than a
continuous auditory stimulus’ (19). According to
this author, it accounts for the distinction between
pulse (or vocal fry ) and chest registers. A given voice
production is perceived as pulsed when its funda-
mental frequency F
0
is lower than a 70-Hz ‘cross-
over’ frequency, and perceived as non-pulsed above
this frequency. In the case of a F
0
/n subharmonic
pattern, the crossover frequency is n
/70 Hz.
According to Titze, the reason for this perceptual
fact lies in formant energy damping (related to the
formant bandwidth) over the fundamental period:
waveforms are perceived as pulse-like if the formant
energy during one excitation is sufficiently damped
out before the next excitation occurs. This author
defines the pulse-chest registers boundaries on the
basis of this periodicity transition.
Transition phenomena may also result from no-
ticeable timbre variations (17,19,20), which may or
may not be related to a laryngeal adjustment.
According to Titze (18,19), timbre transitions are
characterized by an abrupt voice quality change,
which is associated with a spectral energy change in
the high-frequency part of the sound spectrum, i.e.,
a modification in the spectral slope. This spectral
change may come from a variation in the glottal flow
derivative discontinuity and the return phase, which
can be more or less abrupt. From the acoustic point
of view, it may result from an interference between a
subglottal resonance and the vocal fold driving
pressure.
According to Miller (17), either these transitions
have a laryngeal source, or they result from reso-
nance strategies, also called ‘formant tuning’. In this
latter case, a vocal tract resonance is adjusted to
match a given harmonic in frequency, reinforcing the
energy in this harmonic frequency band. The tuning
process may induce the presence of strong acoustic
standing waves, which in return can affect the vocal
fold vibratory movement, yet in a more subtle way
than in the case of a laryngeal adjustment (17).
Mirroring the voice from Garcia to the present day 7
