Metallic Voice: Physiological and Acoustic Features

*Eliana Midori Hanayama, †Zuleica Antonia Camargo, ‡Domingos Hiroshi Tsuji, and §Sı́lvia Maria Rebelo Pinho,
São Paulo, Brazil

Summary. The metallic voice is usually confused with ring or nasality by singers and nontrained listeners, who are
not used to perceptual vocal analysis. They believe a metallic voice results from a rise in fundamental frequency. A di-
agnostic error in this aspect may lead to lowering pitch, an incorrect procedure that could cause vocal overload and fa-
tigue. The purpose of this article is to study the quality of metallic voice considering the correlation between information
of the physiological and acoustic plans, based on a perceptive consensual assumption. Fiberscopic video pharyngolar-
yngoscopy was performed on 21 professional singers while speaking vowel [e]—in normal and metallic modes to ob-
serve muscular movements and structural changes of the velopharynx, pharynx, and larynx. Vocal samples captured
simultaneously to the fiberscopic examination were acoustically analyzed. Frequency and amplitude of the first four
formants (F1, F2, F3, and F4) were extracted by means of linear predictor coefficients (LPC) spectrum and were statis-
tically analyzed. Vocal tract adjustments such as velar lowering, pharyngeal wall narrowing, laryngeal rise, aryepiglot-
tic, and lateral laryngeal constrictions were frequently found; there were no significant changes in frequency and
amplitude of F1 in the metallic voice; there were significant increases in amplitudes of F2, F3, and F4 and in frequency
for F2; metallic voice perceived as louder was correlated to an increase in amplitude of F3 and F4. Physiological adjust-
ments of velopharynx, pharynx, and larynx are combined in characterizing the metallic voice and can be acoustically
related to changes in formant pattern.
Key Words: Voice quality–Voice disorders–Speech acoustics–Speech perception.

INTRODUCTION
The scarce data in the literature about vocal performance
patterns and their correlation with physiological and acoustic
information make it difficult to establish efficient tools for vocal
description, assessment, and training.
Vocal quality refers to the perceptive aspect of the voice in
the psychological sphere, comprising several attributes that
are presented in a continuum and are described with the use
of terms based on auditive perception in each one of them,
such as normal/abnormal, good/bad, bright/dark, oral/nasal,
from one extreme to the other.1 The physical bases of the differ-
ences between the several vocal qualities correspond to a set of
adjustments in the vocal tract.1,2
From the phonetic point of view, vocal quality is determined
by the present acoustic parameters, related to the vibration of
the vocal folds and configuration of the supraglottic vocal
tract.2
Aerodynamic, acoustical, and perceptual studies compared
different vocal qualities in singing such as belt, twang, opera,3
country western, operatic, and Broadway vocal styles,4,5 pop,
jazz, and blues6 and have found distinguished acoustical fea-
tures among different vocal qualities.
Accepted for publication December 18, 2006.
Presented at XII Brazilian Congress of Speech Therapy, October 7, 2004, Iguassu Falls,
ISSN 1807-3115.
From the *Center for Specialization in Clinical Speech Pathology and Audiology, CE-
FAC, São Paulo, Brazil; yIntegrated Laboratory on Acoustic Analysis and Cognition,
LIAAC of the Department of Linguistics of the School of Communication and Philosophy
of the Pontifical Catholic University of São Paulo, São Paulo, Brazil; zDivision of Otorhi-
nolaryngology Clinic, Clinics Hospital, School of Medicine, University of São Paulo, São
Paulo, Brazil; and the xINVOZ, Voice Institute: Communication and Professional Voice,
São Paulo, Brazil.
Address correspondence and reprint requests to Eliana Midori Hanayama, R Mario Para-
nhos Pederneiras, 156, São Paulo, SP, Brazil, CEP 04072-060. E-mail: midorihanayama@
gmail.com
Journal of Voice, Vol. 23, No. 1, pp. 62-70
0892-1997/$36.00
Ó 2009 The Voice Foundation
doi:10.1016/j.jvoice.2006.12.006
One of the polemical issues that involve vocal quality is the
metallic voice, usually confused with ring, nasality, or high
pitch by singers and lay people, who are not used to perceptual
vocal analysis.7 The occurrence of voice metallization was as-
sociated to cul-de-sac resonance in cases of hearing impair-
ment. There is proximity between these attributes, which
makes the occurrence of mistaken diagnostic assessments
very common.8
Barnes, in his thesis, analyzed the vocal quality of students
from the University of Iowa in 1932, and found 27.2% of
them had a metallic voice.9
The impressionistic term metallic voice is a label that has
been used to characterize a strident, irritating, penetrating,
weeping, and fine voice,9 a rough voice,10 and as bright, clean,
sharp, biting voice.2
Metallization was related to hypertonicity of pharyngeal con-
strictors and to the elevation of the larynx, which decrease both
the length and the width of the pharynx11,12 and contract the
walls of the vocal tract. The walls, in turn, become a rigid sur-
face that emphasizes high acoustic frequencies, thus generating
a tense, metallic, strident voice.13 The resonance focus becomes
pharyngeal. In addition to pharyngeal constrictions, there
seems to be an elevation in the back of the tongue, positioning
of the lips in a smiling posture, and aryepiglottic constric-
tion,7,14 and velar lowering.12
Speakers may consider the metallic voice an undesirable
characteristic, because it is considered sharp and irritating.2
However, it may be an efficient resource for vocal projection
in the theater and in noisy environments with poor acoustics,
giving the actor confidence,12 in addition to being useful as
a dramatic resource in the characterization of a character, in
some types of singing, such as the American and the Brazilian
country music.7
The present study examined the metallic vocal quality from
the standpoint of the correlations between the information of
Eliana Midori Hanayama, et al Metallic Voice 63

the physiological and acoustic plans, based on a perceptive con- was obtained. The laryngeal sequence added a laryngostrobo-
sensual assumption. scopy to record the movement of the vocal folds in detail.

METHODS Sample selection

Initial procedures All audio samples captured in the recorder were analyzed from
Twenty-four professional classical or popular singers, healthy the perceptive-auditory point of view by three speech therapists
and with no vocal complaints, with ages ranging from 18 to with experience in vocal quality assessment and familiarized
50 years, were requested to utter the vowel [e], because it is with metallic mode patterns. The samples were recorded ran-
used in nasofibroscopy and it is produced with the tongue in domly to be classified as per the perceptive difference between
high anterior position, favoring visualization of the regions an- the two modes (small, medium, or large). The study only in-
alyzed. The vowel was performed in two vocal qualities—oral cluded the samples of those subjects considered by all three
and metallic, hereinafter referred to as oral mode and metallic judges as efficient in the production of the metallic mode,
mode.1 with clear distinction among the patterns uttered.
Three days before the collection of samples, subjects were Following this criteria, the samples of three subjects were ex-
submitted to a background survey and asked to utter the vowel cluded. To analyze the physiological and acoustic correlations
in the metallic mode without being provided with a model. of the metallic mode, the following subjects were assessed:
When they mentioned they did not know this utterance mode, seven male singers (four tenors and three baritones) and 14 fe-
one of the researchers provided the metallic mode to be imi- male singers (four sopranos, nine mezzo-sopranos, and one
tated, without indicating any muscle adjustments for its produc- contralto), with professional experience from 2 to 36 years
tion. When examiners did not detect the striking perceptive (Table 1).
characteristic of the metallic voice, new utterances were re- Seven subjects presented vocal nodules in the otorhinolaryn-
quested and the target vocal pattern was obtained. Subjects gological exam, but were included in the study, similarly to the
were asked to train the metallic voice at home until the day be- study of Lovetri et al.15 The presence of nodules did not inter-
fore the collection. fere in the production of this vocal quality in the study samples.
All subjects signed a term of free informed consent duly ap-
proved by the ethics committee of the School of Medicine of the Acoustics analysis
University of São Paulo under number 105/02. Audio samples were digitalized with the Computer Speech Lab-
oratory (CSL), model 4300B (Kay Elemetrics Corporation),
Recording procedures
For capturing the voice itself, vowels were initially uttered in
TABLE 1.
the oral mode and, next, in the metallic mode, in the C4 tone
Age, Gender, Years of Experience in Singing, Singing
for female singers, and G3 for male singers. An alternated utter- Style, and Voice Class of the Subjects
ance of the two vocal patterns was requested in sequence, as
many times as necessary to obtain a stable utterance that main- Age Experience
tained the jaw in a fixed position, and constant loudness in both Subject # (yr) Voice Class Gender (yr) Style
situations. 1 27 Soprano F 7 c
During utterances, images of the vocal tract were captured by 2 32 Soprano F 19 c
means of nasopharyngolaryngoscopy, with the use of a flexible 3 50 Soprano F 12 c
fiberscope (Machida, Orangeburg, NY), video camera (Sony, 4 28 Soprano F 8 p
Japan), and VCR (Sony model Super VHS). All utterances 5 24 Mezzo-soprano F 10 p
6 23 Mezzo-soprano F 7 p
were recorded in a professional analogical recorder (Sony
7 18 Mezzo-soprano F 14 p
model TC-D5M), chrome tape, simultaneously to the conduc- 8 33 Mezzo-soprano F 7 c
tion of the fibroscopy, with a microphone (Sony model ECM- 9 33 Mezzo-soprano F 12 p
260F) positioned 5 cm from the subject’s mouth. In addition, 10 37 Mezzo-soprano F 25 p
laryngostroboscopy was performed with a rigid laryngoscope 11 20 Mezzo-soprano F 6 p
(Machida), connected to a source of stroboscopic light (Bruel 12 37 Mezzo-soprano F 16 p
& Kjaer, Naerum, Denmark). 13 23 Mezzo-soprano F 8 p
Nasopharyngolaryngoscopic tests were performed by an oto- 14 22 Contralto F 5 p
rhinolaryngologist who specialized in assessing and diagnosing 15 32 Tenor M 13 c
voice and larynx disorders. Three sequences were recorded in 16 31 Tenor M 9 c
different levels of the vocal tract: the first at the velopharynx, 17 19 Tenor M 7 p
18 28 Tenor M 14 p
then at the pharyngolarynx, and lastly, at the larynx.15 The nasal
19 44 Baritone M 36 p
fiberscope was kept fixed during the utterances of the two 20 33 Baritone M 7 p
modes, oral and metal, for each of the levels of the vocal tract 21 24 Baritone M 2 p
assessed. Each sequence requested more than one utterance in
m, male; f, female; p, popular; c, classical.
each vocal mode, until a stable image, not showing variation,
64 Journal of Voice, Vol. 23, No. 1, 2009

using the CSL software for Windows of the Voice and Speech
Laboratory of the Center for Specialization in Clinical Speech
Pathology and Audiology (CEFAC), in the sampling frequency
in 22 050 Hz, 16 bits. For sample analysis, the procedures per-
formed included acoustic inspection (wave format and wide-
band spectrogram), in which extracts of utterance stability
were selected to obtain the measures of frequency (Hz) and am-
plitude (dB) of the four first formants (F1, F2, F3, and F4). The
wideband spectrograms were generated with Hanning win-
dows, defining the extension of the superior frequency limit
at 5 kHz because higher frequency energy can lead to inaccu-
rate allocation of zeroes in the error prediction filter. The dura-
tion of the window was defined at 10 milliseconds.
Downsampling to 11 025 Hz was made and the filter order
was kept between 10 and 12 to obtain the spectral envelope.16
The intensity values of spectral peaks were normalized to
proportional values to eliminate variations due to different
levels of recording and different loudnesses of utterance at col-
lection.17,18 In this manner, the values referring to F1 intensity
were normalized to value 100 and, the others, to fractions of
100.19
Physiological analysis
For each sequence of nasofibroscopic images, both for the oral
mode utterance and metallic mode utterance, we selected and
froze the images that presented the highest degree of constric-
tion of all vocal tract structures involved in that segment, for
the three positions recorded: velopharynx, pharyngolarynx,
and larynx. All selected images were printed on film by
a Sony video image printer, model UP2300 (Japan), and were
then traced in transparencies, for the comparative analysis of
the movement of the structures.
For this analysis, two situations were defined: the initial po-
sition, observed during the oral mode utterance, and the analy-
sis position, observed during the metallic mode utterance.15 The
different relative postures of the structures of the vocal tract
were qualitatively analyzed.
The analysis of the physiological data involved the detection
or nondetection of the adjustments of the structures of the vocal
tract15 during oral and metallic mode utterances, based on the
comparison of the traces of photographic records, regarding
the soft palate, presence or absence of slight tension, presence
or absence of moderate tension, presence or absence of lower-
ing, medialization of the pharynx walls, changes in the height of
the larynx, aryepiglottic tightening, and lateral larynx tighten-
ing. When a certain adjustment was detected, a ‘‘1’’ was
marked, when not, a ‘‘0.’’
When the anatomical structures were not visible, the image
was classified as ‘‘nonvisible.’’
Statistical analysis
For the analysis of frequency acoustic data and formant ampli-
tude data (F1, F2, F3, and F4) of the oral mode and metallic
mode, and comparison per gender, we used the Student’s t
test for matching data, Mann-Whitney test, and Fisher Exact
Test. To check the correlation between the data referring to
the formants and the adjustments of the vocal tract, we used
Spearman correlation analysis. The significance level consid-
ered was 5%.
RESULTS
Table 2 shows the analysis of the nasopharyngolaryngofiber-
scopic images observed during metallic mode utterance on
three analysis positions: velopharyngeal, laryngopharyngeal,
and laryngeal sequences, in each subject.
The frequencies of adjustments (in number of occurrences
and %) seen are presented in Table 3. Most showed aryepiglot-
tic constriction (81%) and laryngeal raising (71%). Among ve-
lopharyngeal adjustments, the most frequent was lowering
(48%). Other relatively frequent adjustments included lateral
constriction (38%) and light tightening (24%). In nine subjects

TABLE 2.
Analysis of Nasopharyngolaryngofiberscopic Images Observed During Metallic Mode Utterance: Velopharyngeal,
Pharyngeal, and Laryngeal Movements in Each Subject
Sequence 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Velopharynx
No movement 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0
Light tightening 0 1 0 0 0 0 0 0 1 0 1 0 0 0 1 1 0 0 0 0 0
Moderate tightening 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0
Lowering 0 0 0 1 1 1 1 1 0 1 0 1 1 0 0 0 0 0 0 1 1
Laryngopharynx
Pharynx: medialization NV NV NV 1 1 0 1 NV 0 1 NV NV 1 1 NV NV 1 NV 1 1 1
Larynx: raising 1 1 1 1 1 0 1 1 0 1 0 1 1 0 0 0 1 1 1 1 1
Larynx: lowering 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0
Larynx
Aryepiglottic tightening 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 0 1 1 1 1
Lateral tightening 0 0 0 0 0 0 1 0 0 0 1 1 1 1 1 0 1 0 0 1 1
0, no adjustment in analysis position; 1, adjustment present in analysis position; NV, not visible.
Eliana Midori Hanayama, et al Metallic Voice 65

TABLE 3.
Adjustments Observed in the Vocal Tract (in Number of
Occurrences and %) During Metallic Mode Utterance of
the Subjects
Vocal Tract Adjustment n %
Velopharynx
No movement 3 14
Light tightening 5 24
Moderate tightening 3 14
Lowering 10 48
Laryngopharynx
Pharynx: medialization 10 48
Larynx: raising 15 71
Larynx: lowering 2 9
Larynx
Aryepiglottic tightening 17 81
Lateral tightening 8 38

(43%), adjustment of the pharyngeal medialization was not vis-

ible.
Among the possible correlations and the adjustments ob-
served in the vocal tract, the relationship between velar adjust-
ments and pharyngeal medialization was analyzed. There is
evidence that there is a correlation between the occurrences
of both adjustments (P ¼ 0.624).
Figures 1–4 show nasofibroscopic images with examples of
the adjustments observed in each sequence.
Two male subjects presented larynx lowering.
Subject 9 (mezzo-soprano) presented only a slight tightening
of the velum palatinum and contraction of the posterior pharyn-
geal wall, without perceptible laryngeal adjustments.
Laryngostroboscopic images suggested a longer time for FIGURE 1. Nasopharyngolaryngofiberscopic images of velophar-
glottal closing. yngeal sequence: subject 12. (A) Oral mode utterance. (B) Metallic
mode utterance where velar lowering can be observed.

Comparison of frequency and amplitude values of

F1, F2, F3, and F4 during oral and metallic mode
Table 4 presents the average normalized amplitude values of F1,
F2, F3, and F4 in dB in oral and metallic mode utterances of sub-
jects.
The amplitude of formants F2, F3, and F4 showed a significant
increase in the metallic mode in relation to oral mode. F1 did
not present significantly different amplitude in the comparison
between the two modes.
Figures 5–7 present the LPC spectrums of the oral mode and
metallic mode utterances for each subject, which illustrate the
amplitude variation of F2, F3, and F4.
Table 5 presents the average frequency values of F1, F2, F3,
and F4 for oral and metallic mode utterances.
As to the frequency parameter of formants, F2 was signifi-
cantly greater in the metallic mode. Both F3 and F4 did not pres-
ent significant changes in the corresponding frequencies. F1
values were practically unchanged. As a result, it is possible
to say that the relative distances between F3 and F4 are also sim-
ilar between the vocal mode and metallic mode.
Figure 8 shows, as examples, the curves of the spectral enve-
lope referring to the samples of subject 9 (mezzo-soprano),
where it is possible to see the mentioned changes: raising of fre-
quency for F2 and raising of amplitude for F2, F3, and F4 ob-
served during metallic mode in relation to the oral mode
emission.
Subject 3 (soprano) presented, in the metallic mode, a marked
approximation of F3 and F4 frequencies, the latter practically
indefinable and with energy gain in this range (Figure 9).
There were no significant differences between men and
women subjects regarding utterances in the oral and metallic
modes. As to adjustments of the vocal tract, Mann-Whitney
test indicated a difference between men and women for the lar-
ynx lowering variable (P ¼ 0.040). However, Fisher Exact Test
showed that this difference was not significant (P ¼ 0.100).
DISCUSSION
Metallic voice has been described with many terms based on
perception.2,7–12 The present study considered the analysis of
66
FIGURE 2. Nasopharyngolaryngofiberscopic images of laryngo-
pharynx sequence: subject 12. (A) Oral mode utterance. (B) Metallic
mode utterance where laryngeal raising can be observed.
metallic mode voice defined as having pharyngeal resonance
and strident quality, according to the descriptions of Boone
and McFarlane,11 Berry,12 and Pinho.7
Vocal tract adjustments, determining for the vocal quality of
utterance, have been studied in the phonetic model, as a combi-
nation of intrinsic factors, related to the anatomy unique to a cer-
tain speaker, and extrinsic, related to long-term muscular
adjustments. These adjustments may be present in normal sys-
tems,2 as a vocal resource, or as a compensation to changes.
Some supraglottic adjustments were analyzed in this study
through the direct observation of the involved structures and
acoustic analysis.
Most subjects in this study, professional singers, showed that
they knew the vocal quality under study and produced it follow-
ing verbal command.
Simultaneous images of the velopharynx, pharynx, and lar-
ynx could not be obtained only through the nasopharyngolar-
yngoscopy. Adjustments made it difficult for the larynx to be
simultaneously observed. In this manner, it was not possible
Journal of Voice, Vol. 23, No. 1, 2009
FIGURE 3. Nasopharyngolaryngofiberscopic images of laryngo-
pharynx sequence: subject 13. (A) Oral mode utterance. (B) Metallic
mode utterance where pharyngeal wall medialization can be observed.
to obtain images of the many positions during the same utter-
ance.
The supraglottic adjustments observed during metallic mode
utterance presented a relevant occurrence in this study. Aryepi-
glottic constriction was seen in 81% of the subjects. This adjust-
ment was observed during singing in the twang, belting, and
opera styles, where the presence of the singer’s formant was
verified.3 The main function of the aryepiglottic fold was
thought to be related to the epiglottis in swallowing but it was
also related to speaking, when those folds become high,
smooth, and taut, producing a deep supraglottic cavity for the
regulation of resonance.13
Lateral constriction was observed in 38%. The ventricle of
Morgagni provides a space for free vibration of the vocal
fold, and it seems that it functions as a resonator. When pressed,
its volume was reduced by the ventricular folds, and this was re-
lated to the lack of suppression of sharp harmonics, which
would result in a rough and metallic voice. This phenomenon
would be related to vocal potency in singers who have small
Eliana Midori Hanayama, et al Metallic Voice 67

FIGURE 5. Comparison of normalized amplitude variation of F2 in

dB between oral and metallic mode emissions for each subject.

a natural increase of loudness observed in the metallic mode.

FIGURE 4. Nasopharyngolaryngofiberscopic images of larynx se-
quence: subject 20. (A) Oral mode utterance. (B) Metallic mode utter-
ance where aryepiglottic tightening can be observed.
ventricles.10 Ring quality is a vocal quality whose perception is
related also to increase in high spectral energy in the singer for-
mant range.20 Its physiological correlation is, among other fac-
tors, a narrower epilaryngeal tube area, that refers to the cross-
sectional area of the supralaryngeal portion that extends from
vocal folds to the aryepiglottic folds.21,22
For both emission modes, the target was to obtain the best
possible quality and yield in relation to loudness. There was
This increase is related to the increase in the amplitude of for-
mants.18 It was considered important not to interfere in this pro-
cess with an objective control of intensity, because the increase
in loudness is a striking characteristic of the metallic voice. For
this reason, measures were normalized, thus ensuring the possi-
bility of data comparison. The comparison of the acoustic anal-
ysis of emissions in the oral and metallic modes of the present
study found that in the metallic mode there is a significant in-
crease in amplitude of F3 and F4, in relation to F1, which prob-
ably suffers the interference of these supraglottic adjustments.
It is worth noting that F1 amplitude did not show significant in-
crease in the metallic mode emission. This acoustic feature
probably provides efficiency to metallic voices, when potency
is necessary in certain environments and situations. Another
finding is the approximation between F3 and F4 peaks in one
subject (soprano). When the frequencies of the two formants
are closer, the peaks of both increase their amplitude.23
There seems to be greater muscular mobilization during ut-
terance in the metallic mode. Fant23 valued the degree of vocal
fold closing related to rise in vocal amplitude. The vocal source
during the tense quality voice utterance described by Laver2
seems to be similar to the metallic voice. The laryngostrobo-
scopic images of the present study suggested increased glottal
closing time, which could be related to the increased loudness
observed. However, data related to such increases were not ob-
tained by means of this limited procedure for verifying the de-
gree of glottis closing. The statistically significant rise in the
amplitude of F2, F3, and F4, in relation to F1, must be related
to the degree of glottal abduction/adduction.6,24

TABLE 4.
Average Normalized Values of Levels of Formants (F1, F2,
F3, and F4) + SD in dB During Oral and Metallic Mode
Utterance of Subjects
Amplitude (dB) Oral Metallic P
F1 14 ± 5 16 ± 5 0.126
F2 71 ± 10 85 ± 11 <0.001
F3 63 ± 11 80 ± 10 <0.001
FIGURE 6. Comparison of normalized amplitude variation of F3 in
F4 46 ± 12 63 ± 17 <0.001
dB between oral and metallic mode emissions for each subject.
68 Journal of Voice, Vol. 23, No. 1, 2009

FIGURE 7. Comparison of normalized amplitude variation of F4 in

dB between oral and metallic mode emissions for each subject.

There was no statistically significant change in the values of

frequency or amplitude of F1 in the analysis of the absolute non-
normalized values of the samples of the metallic mode. These
aspects deserve further studies through other methods such as FIGURE 8. LPC spectrum of oral and metallic emission—subject 9.
electroglottography.25
To analyze in more depth the physiological bases underlying
in the pharynx region and did not have any other adjustment in
pharyngeal findings, it is worth mentioning that the palatoglos-
the vocal tract. Isometric contraction of the pharyngeal muscles
sal muscle and the palatopharyngeal muscle can be considered
would change its resonance properties, giving the voice the me-
as related to the pharyngeal resonator.26 The palatopharyngeal
tallic and strident quality28 determined by the emphasis in high
muscle is connected to the aryepiglottic fold, which in turn is
harmonics.2,11,13 This should explain why the subject had a me-
related to the oblique arytenoid muscle, adductor of the vocal
tallic mode perceived as a major difference in comparison to the
folds. Thus, its contraction may be associated with the vibration
oral mode and spectral curve clearly characteristic of this mode,
mode of the vocal folds and with the contraction of the laryng-
even without the remaining adjustments in the tract. The impor-
opharynx and of the upper portion of the larynx.2
tance attributed by Greene26 to the participation of the pharynx
Pressed voice is another quality of voice and it can be con-
in vocal quality is worth highlighting.
fused with the ring voice. Its perception is related to increased
Velar lowering was observed in 10 subjects (48%). This ad-
glottal adduction. As the ring voice, it is acoustically related to
justment is related to the contraction of the pharyngeal pillars2
an increase in high-frequency energy.6,22 Similar physiological
and determines perceivable differentiated effects according to
adjustments may be present during the metallic mode utterance.
the height of vowels. The acoustic effect is the elevation in
Among the 12 subjects in whom the narrowing of the pharyn-
the frequency of F1 and the lowering of F2,23 which did not oc-
geal walls could be visualized, 10 had this adjustment (48% of
cur in the present study. Of these 10 subjects with velar lower-
total subjects). Among the 10 subjects, seven had velar lower-
ing, only one did not have evident laryngeal raising, which may
ing showing evidence that there is a correlation between these
be related to a physiological interconnection between these two
adjustments, which, although not significant, may be due to
adjustments, determined, partially, by the same muscle, the pal-
the small n for this variable. In a review of his research on sing-
atopharyngeal.2 Laryngeal raising is a tendency for untrained
ing voices, Sundberg27 made comments about the strong neural
speakers/singers when they increase effort, so that they shorten
relationship between the velar and the pharyngeal regions that
the vocal tract and increase formant frequencies. It was related
need to be further clarifie.
to production of more strident vocal timbre.29
Findings such as isometric pharyngeal constriction and
changes in the vertical velar position were almost imperceptible
in the present study. Some subjects had a constriction of the
pharyngeal walls, which kept the pharynx volume unaltered.
Subject 9 (mezzo-soprano) had this clear isometric contraction

TABLE 5.
Average Frequency Values of Formants (F1, F2, F3, and
F4) + SD in Hz During Oral and Metallic Mode Emission of
Subjects
Frequency (Hz) Oral Metal P
F1 443 + 84 443 + 68 >0.999
F2 2038 ± 245 2186 ± 250 <0.001
F3 2708 ± 217 2840 ± 341 0.052
F4 3872 ± 425 3973 ± 510 0.158
FIGURE 9. LPC spectrum of oral and metallic emission—subject 3.
Eliana Midori Hanayama, et al Metallic Voice
It is worth emphasizing that velar lowering does not mean ve-
lopharyngeal closing deficit, which can be confirmed by the fact
that no attenuation of formants in the acoustic analysis was ob-
served. In fact, velopharyngeal opening that could be caused by
velar lowering is not also necessarily related to nasal quality of
the vowel timbre.30
There are almost always other phenomena parallel to pharyn-
geal constriction, such as change in laryngeal height and ten-
dency to velar lowering, which may cause a certain nasalizing
feeling.2 This should be related to the frequent confusion in
the perceptive evaluation of the metallic quality with nasaliz-
ing. The metallic quality, sometimes mistaken for nasalizing,
may be perceived in western country singing and in Brazilian
country singing.3
The vocal quality is clearly influenced by the rising ampli-
tude of F2 and F3, when compared to the wider amplitude of
F1.31
The voices of classical singers and the speech of many indi-
viduals with ring are characterized by an increased amplitude in
F3 and F4.19,32 The fact that the sample of the present study had
a significant rise in amplitude of both formants may indicate
that the ring component accompanies the metallic voice.
Another phenomenon related to ring is the potency observed
in lyrical singing. Professional classical singers have certain re-
inforced frequency bands, around 3 kHz, the singer’s formant,
enabling him/her to be heard, perfectly, even while accompa-
nied by a symphonic orchestra, and it is acquired through tech-
nique and experience.33–35 Laboratory experience has related
this phenomenon to an anteroposterior glottis narrowing also
observed in the present study.21
In the country style, on the other hand, in which the voice is
used with a feature similar to spoken voice, singers do not have
the singer’s formant36 despite the tendency to raise the fre-
quency of F2. Voices considered efficient for professional use
were related, in acoustic analyses, to the concentration between
3 and 4 kHz frequencies, called ‘‘speaker’s formant.’’17
As to the behavior of the frequencies of formants in the me-
tallic mode, there was a significant shift of F2 frequency to
a higher region. Formants F1, F3, and F4 did not change signif-
icantly in frequency. F2 is extremely sensitive to anteroposterior
positioning of the tongue. The tongue anteriorizing movement
that narrows the region of the hard palate and expands the pha-
ryngeal cavity leads to an elevation of this formant.23,37 When
an adult woman adopts a ‘‘little girl’s’’ voice, the elevation is
mentioned and may be a slight tongue anteriorizing and, some-
times, laryngeal raising.2 In this situation, the frequency of F2
becomes very high and near F3, a behavior that was observed
in the present study. Probably the significant frequency rise of
F2 observed in the metallic mode may be related to this kind
of tongue adjustment. The adjustment was not evaluated in
this study, although it may have occurred. Tongue dislocation
is related to extremely complex synergetic muscular move-
ments. Findings of tests such as MRI or 3D images captured
by electron beam computed tomography, where the vocal tract
configuration is a function of changes in vocal modes, can be
assessed by comparing differences in area functions along the
length of the vocal tract29 allied to filming lip behavior. This
69
would be perfect and would greatly help the understanding of
the behavior of F2, leaving, therefore, a gap to be explained
by future research.
Lindblom and Sundberg37 reported effects of frequency low-
ering of all formants to the low larynx position, with a more pro-
nounced effect in F2. In theory, the opposite situation, that is,
larynx raising, could be related to an increase in F2. However,
the laryngeal raising observed in the present study did not cor-
relate significantly to increase in F2.
No significant differences were detected in the data observed
when comparing male and female subjects. The only variable
with a significant difference in a statistical test, although not
confirmed by another, was larynx lowering, observed in two
of seven male subjects, and it was not observed in any female
subject. Future research should assess the differences between
men and women in the incidence of this kind of adjustment.
The metallic voice is commonly associated with the raise in
regular fundamental frequency by nontrained listeners. A diag-
nostic error in this aspect leads to an incorrect therapeutic pro-
cedure of lowering regular frequency, which in turn would
cause vocal overload and fatigue.7 The factors that in fact deter-
mine the high pitch perceived by the listener, and that could be
observed in this study, could be related to the emphasis in the
amplitude in the high spectral region of F2, F3, and F4.
The functional diagnosis should, therefore, consider that the
physiological correspondences to the perceptive-auditory find-
ing of the metallic mode were pharyngeal constriction and func-
tional reduction of vocal tract cavities due to tension of
structures14 and also of remaining adjustments observed in
this study. In the future, lingual and labial adjustments, not an-
alyzed in this study, should be taken into consideration.
The use of metallic voice in some singing situations or styles
and the adjustments observed in this study, especially epilarynx
narrowing, seem to be the most effective configurations to be
used, from the vocal hygiene point of view.
CONCLUSION
Regarding physiological and acoustic characteristics of the me-
tallic mode in relation to the oral mode:
1. There was a relevant incidence of adjustments of the vo-
cal tract such as velar lowering, pharyngeal wall medial-
ization, laryngeal raising, aryepiglottic, and lateral
tightening.
2. There was no significant change in F1, neither in relation
to frequency nor in relation to amplitude.
3. There was a significant change in the amplitude of F2, F3,
and F4.
4. There was a significant dislocation of F2 in the direction
of high frequencies.
5. Higher intensity was significantly correlated with the el-
evation in the amplitude of F3 and F4.
The analyses performed in this study provided data that may
help toward the correct approach of this vocal mode and will
70
support future studies that should be included in the analysis of
the glottis source and of labial and lingual adjustments.
REFERENCES
1. Colton RH, Estill JA. Elements of voice quality: perceptual, acoustic and
physiologic aspects. Speech and Language: Advances in Basic Research
and Practice. New York, NY: Academic Press Inc.; 1981. 311–403.
2. Laver J. The Phonetic Description of Voice Quality. Cambridge, MA: Cam-
bridge University Press; 1980.
3. Yanagisawa E, Estill J, Kmucha ST, Leder SB. The contribution of aryepi-
glottic constriction to ‘‘Ringing’’ voice quality—a videolaryngoscopic
study with acoustic analysis. J Voice. 1989;3:342-350.
4. Stone RE, Cleveland TF, Sundberg J, Prokop J. Aerodynamic and acousti-
cal measures of speech, operatic, and Broadway vocal styles in a profes-
sional female singer. Speech, Music and Hearing. In: TMH-QPSR, Vol.
43. Stockholm, Sweden: KTH; 2002. 17–29.
5. Stone RE, Cleveland TF, Sundberg J. Acoustic and aerodynamic character-
istics of country-western, operatic and Broadway singing styles compared
to speech. J Acoust Soc Am. 2003;113:2242-2243.
6. Sundberg J, Thalén M, Alku P, Vilkman E. Estimating perceived phonatory
pressedness in singing from flow glottograms. Speech, Music and Hearing.
In: TMH-QPSR, Vol. 43. Stockholm, Sweden: KTH; 2002. 89–96.
7. Pinho SMR. Avaliação e Tratamento da Voz. Fundamentos em Fonoaudio-
logia. Rio de Janeiro: Guanabara Koogan; 1998. 3–37.
8. Wilson DK. Children’s voice problems. Voice Problems of Children. 3rd ed.
Maryland: Williams & Wilkins; 1987. 2–15.
9. Van Dusen CR. A laboratory study of the metallic voice. J Speech Dis.
1941;6:137-140.
10. Van den Berg JW. On the role of the laryngeal ventricle in voice production.
Folia Phoniatr. 1955;7:57-69.
11. Boone DR, McFarlane SC. The Voice and Voice Therapy. 4th ed. New Jer-
sey: Prentice-Hall; 1988.
12. Berry C. Voice and the Actor. New York, NY: MacMillan; 1991.
13. Kaplan HM. Anatomy and Physiology of Speech. New York, NY:
McGraw-Hill; 1971.
14. Pinho SMR. Fonoterapia nos distúrbios vocais. In: Campos CAH,
Costa HOO, eds. Tratado de Otorrinolaringologia. São Paulo: Roca;
2002:596-605.
15. Lovetri J, Lesh S, Woo P. Preliminary study on the ability of trained singers
to control the intrinsic and extrinsic laryngeal musculature. J Voice.
1999;13:219-226.
16. Vieira MN. Automated measures of dysphonias and the phonatory effects of
asymmetries in the posterior larynx [doctorate]. Edinburgh: University of
Edinburgh; 1997.
17. Nawka T, Anders LC, Cebulla M, Zurakowski D. The speaker’s formant in
males voices. J Voice. 1997;11:422-428.
Journal of Voice, Vol. 23, No. 1, 2009
18. Nordenberg M, Sundberg J. Effect on LTAS of vocal loudness variation.
Logop Phoniatr Vocol. 2004;29:183-191.
19. Oliveira-Barrichelo VM, Heuer RJ, Dean CM, Sataloff RT. Comparison of
singer’s formant, speaker’s ring, and LTA spectrum among classical singers
and untrained normal speakers. J Voice. 2001;15:344-350.
20. Ekholm E, Papagiannis GC, Chagnos FP. Relating objective measurements
to expert evaluation of voice quality in western classical singing: critical
perceptual parameters. J Voice. 1998;12:182-196.
21. Titze IR, Story BH. Acoustic interactions of the voice source with the lower
vocal tract. J Acoust Soc Am. 1997;101:2234-2243.
22. Bergan CC, Titze IR, Story B. The perception of two vocal qualities in
a synthesized vocal utterance: ring and pressed voice. J Voice. 2004;18:
305-317.
23. Fant G. Acoustic Theory of Speech Production. Paris: Mouton; 1970.
24. Henrich N, D’Alessandro C, Castellengo M, Doval B. Glottal open quotient
in singing: measurements and correlation with laryngeal mechanisms, vo-
cal intensity, and fundamental frequency. J Acoust Soc Am. 2005;117:
1417-1430.
25. Baken RJ. Electroglottography. J Voice. 1992;6:98-110.
26. Greene MCL. The Voice and Its Disorders. Philadelphia, PA: Pitman Med-
ical; 1972.
27. Sundberg J. Research on the singing voice in retrospect. Speech, Music and
Hearing. In: TMH-QPSR, Vol. 45. Stockholm, Sweden: KTH; 2003. 11–22.
28. Hardcastle WJ. Physiology of Speech Production. London: Academic
Press; 1976.
29. Tom K, Titze IR, Hoffman EA, Story BH. Three-dimensional vocal tract
imaging and formant structure: varying vocal register, pitch, and loudness.
J Acoust Soc Am. 2001;109:742-747.
30. Birch P, Gumoes B, Stavad H, Prytz S, Bjorkner E, Sundberg J. Velum be-
havior in professional classic operatic singing. J Voice. 2002;16:61-71.
31. Borden GJ, Harris KS, Raphael LJ. Speech Science Primer—Physiology,
Acoustics and Perception of Speech. 3rd ed. Baltimore, MD: Williams &
Wilkins; 1994.
32. Sundberg J. Vocal tract resonance. In: Sataloff RT, ed. Professional Voice:
The Science and Art of Clinical Care. New York, NY: Raven Press Ltd.;
1991:49-68.
33. Sundberg J. Articulatory interpretation of the ‘‘singing formant.’’ J Acoust
Soc Am. 1974;55:838-844.
34. Sundberg J. Level and center frequency of the singer’s formant. J Voice.
2001;15:176-186.
35. Pillot C, Quattrocchi S. Acoustic measurements, perceptive judgements
and physiological correlates of the ‘‘singing-formant’’ in male and female
opera singers. Rev Laryngol Otol Rhinol (Bord). 1996;117:335-339.
36. Cleveland TF, Sundberg J, Stone RE. Long-term average characteristics of
country singers during speaking and singing. J Voice. 2001;15:54-60.
37. Lindblom EF, Sundberg JEF. Acoustical consequences of lip, tongue, jaw,
and larynx movement. J Acoust Soc Am. 1971;50:1166-1179.