Professional Documents
Culture Documents
Summary: Objectives. To analyze the discrimination ability of acoustic, auditory parameters, and perception of
vocal effort during professional and social voice use, and the correlations of these parameters with the vocal demands.
Study Design. Longitudinal study.
Methods. Seventy-three subjects participated in the study: 31 females aged from 28 to 65 years (G1; professional
voice users) and 42 females aged from 31 to 59 years (G2, social voice users; ). All the subjects were subjected to acoustic voice analysis including F0 median, semiamplitude interquartile, quantile 99.5%, and skewness; first F0 derivate
mean, standard deviation (SD), and skewness; intensity skewness; spectral slope mean, SD, and skewness; long-term
average spectrumfrequency SD, perceptual parameters (GRBASI scale), and self-perception of vocal effort, before
and after 2 hours and 30 minutes of voice use. Statistical analyses were completed via multivariate discriminant analysis
and canonical correlation analysis.
Results. Discriminant analysis of acoustic, perceptual, and self-rating variables and analysis of the grouped parameters did not differentiate the samples before and after vocal use. Higher levels of canonical correlation were found for the
professional voice group after voice use, with a correlation between perceptual analysis and acoustic measures.
Conclusions. The current measures could not discriminate the differences of the type of vocal demands, professional
or social.
Key Words: VoiceSpeech acousticsDysphoniaFacultySpeechLanguage and hearing sciences.
INTRODUCTION
Teachers are often referred to as vocal athletes because of the
significant vocal demands placed on them in the workplace.1,2
The American College of Sports Medicine guidelines
established clinical protocols for physical training and
conditioning.3 The guidelines outline the fundamental tenet of
training to include progressive overload, or a gradual increase
in the physical stresses associated with physical exercise.3 The
correlation between vocal loading and vocal fatigue is not
well established. Extended periods of vocal fold vibration and
collision of vocal fold tissues during continuous speech and
more so during increased sound pressure levels in teachers,4
may lead to vocal fatigue. Titze suggests that intense voice
use leads to vocal fatigue and subsequent changes in blood circulation, tension, viscosity, and the composition of fluids, with
deleterious effect on phonatory efficiency.5
The literature defines vocal fatigue as a negative adaptation as
a consequence of prolonged vocal use,6 and this negative adaptation can be observed in auditory-perceptual, acoustic, aerodynamic, or physiological evaluation, indicating an undesirable
effect on functional phonatory physiology.7 Recent literature
suggests that after vocal loading, the most sensitive parameters
indicative of hyperfunctional voice production are acoustic
Accepted for publication June 16, 2014.
From the *Department of Speech-language Pathology, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil; yIntegrated Laboratory on Acoustic Analysis
and Cognition, Postgraduate Studies Program in Applied Linguistics and Language
Studies of the Pontifical Catholic University of S~ao Paulo, S~ao Paulo, Brazil;
zOtorhinolaryngology Service of Federal University of Minas Gerais, Belo Horizonte,
Minas Gerais, Brazil; and the xDepartment of Quantitative Methods, Faculty of Economics
and Administration of the Pontifical Catholic University of S~ao Paulo, S~ao Paulo, Brazil.
Address correspondence and reprint requests to Ana Cristina C^ortes Gama, Federal University of Minas Gerais, Av. Alfredo Balena, 190/249, Belo Horizon, CEP 30130-100,
Brazil. E-mail: anacgama@medicina.ufmg.br
Journal of Voice, Vol. 29, No. 2, pp. 260.e45-260.e50
0892-1997/$36.00
2015 The Voice Foundation
http://dx.doi.org/10.1016/j.jvoice.2014.06.012
^ rtes Gama, et al
Ana Cristina Co
260.e46
Perceptual analysis
Perceptual analysis was performed by two speech language pathologists with 10 years of experience in work in a voice clinic
and the samples were randomized, and therefore, the reviewers
were blinded to the group (G1 vs G2) and also the recording
condition (pre vs post loading).
The GRBASI scale was used for perceptual analysis; Gdegree of dysphonia, R-roughness, B-breathiness, A-asthenia,
S-stiffness, and the last parameter I-instability inserted by
Dejonckere et al (1996).20 With this scale, each speech pathologist evaluated the parameters according on a 0 to 3 scale with
0 reflecting absence of disturbance, 1 for slight, 2 for moderate,
and 3 for intense. The speech language pathologists performed
the perceptual analysis with Coby CV-3000 (Coby Electronics,
Lake Success, NY) earphones and were allowed to hear the
voice recordings as many times as needed.
Acoustic analysis
Acoustic analysis of the vocal samples (read sentences and
semispontaneous speech) was done by the script Expression
Evaluator21 applied to the free software Praat22 for automatic
extraction of acoustic measures. The script generated the acoustic parameters of F0 (median, semiamplitude interquartile,
quantile 99.5%, and skewness), first F0 derivative (mean, SD,
and skewness); intensity (skewness), spectral slope (mean,
SD, and skewness) and long-term average spectrum (LTAS) frequency (SD) measures.
Vocal self-perception
The subjects in both G1 and G2 were asked to evaluate their
effort during sentence reading before and after 2.5 hours of
voice use via a 10 point visual analogue scale (VAS). Zero
represented the absence of effort and 10 maximum effort.
A visual image was provided to compliment the scale
(Figure 1).
Statistical analysis
Statistical analysis of acoustic parameters, perceptual analysis,
and vocal self-perception was done by multivariate discriminant analysis and a canonical correlation analysis to evaluate
the strength of the association and to discriminate the parameters related to vocal loading in both G1 and G2.
Discriminant analysis of the acoustic parameters, perceptual analysis, and vocal self-perception was done both separately and grouped to evaluate the possibility of prediction
and discrimination of the samples in each group and at
each moment of vocal use. Canonical correlation analysis
260.e47
TABLE 1.
Values of Mean, Median, and Standard Deviation (SD) of the Acoustic Measurements in Group of Social Vocal Use Before
and After the Interval of 2.5 Hours
Before
After
Acoustic Measures
Mean
Median
SD
Mean
Median
SD
Median of F0
Semiamplitude interquartile of F0
Quantile 99.5% of F0
Skewness of F0
Mean of first F0 derivate
SD of first F0 derivate
Skewness of first F0 derivate
Skewness of intensity
Mean of spectral slope
SD of spectral slope
Skewness of spectral slope
LTAS-frequency (SD)
210.30
64.33
0.61
0.04
0.11
0.20
0.16
4.50
0.30
0.28
1.30
14.40
213.00
58.80
0.49
0.05
0.17
0.02
0.20
4.40
0.30
0.28
1.30
14.30
24.34
28.96
0.43
0.10
0.62
0.01
0.01
1.90
0.05
0.05
0.10
3.90
211.90
59.64
0.59
0.50
0.11
0.02
0.20
4.20
0.20
0.28
1.30
14.25
215.40
54.00
0.41
0.05
0.17
0.02
0.15
4.20
0.30
0.28
1.28
14.30
22.01
27.89
0.44
0.10
0.66
0.11
3.80
1.80
0.06
0.05
0.10
3.85
RESULTS
Tables 1 and 2 show the acoustic measures in G1 and G2, and
Tables 3 and 4 show the values of the perception of vocal
effort in G1 and G2.
Discriminant analysis of the acoustic parameters, perceptual
analysis, and vocal self-perception and the analysis of the
grouped parameters (acoustic parameters, perceptual analysis,
and vocal self-perception) before and after vocal use in G2 is
shown in Table 5.
TABLE 2.
Values of Mean, Median, and Standard Deviation (SD) of the Acoustic Measurements in the Group of Professional Voice
Use Before and After an Interval of 2.5 Hours
Before
Acoustic Measures
Median of F0
Semiamplitude interquartile of F0
Quantile 99.5% of F0
Skewness of F0
Mean of first F0 derivate
SD of first F0 derivate
Skewness of first F0 derivate
Skewness of intensity
Mean of spectral slope
SD of spectral slope
Skewness of spectral slope
LTAS-frequency (SD)
After
Mean
Median
SD
Mean
Median
SD
193.0
67.2
0.770
0.094
0.116
0.026
0.30
4.85
0.23
0.26
1.29
14.5
192.6
58.8
0.850
0.100
0.190
0.020
0.20
4.80
0.22
0.25
1.28
14.3
21.1
34.5
0.470
0.090
0.750
0.015
3.65
2.10
0.06
0.06
0.08
4.4
203.5
73.7
0.789
0.092
0.186
0.027
0.18
4.60
0.22
0.25
1.30
13.6
202.2
62.4
0.850
0.100
0.250
0.020
0.10
4.60
0.21
0.24
1.30
13.3
22.0
38.4
0.450
0.093
0.744
0.015
3.30
2.03
0.06
0.06
0.09
4.3
^ rtes Gama, et al
Ana Cristina Co
TABLE 3.
Values of the Perception of Vocal Effort in Group of Social
Vocal Use Before and After the Interval of 2.5 Hours
Before
After
Values
0
1
2
3
4
5
6
7
8
9
10
23
05
04
05
03
02
0
0
0
0
0
54.8
11.9
09.5
11.9
07.1
04.8
0
0
0
0
0
24
03
05
02
02
02
02
02
0
0
0
57.1
07.1
11.9
04.8
04.8
04.8
04.8
04.8
0
0
0
TABLE 5.
Confusion Matrix for the Estimation Sample Groups
Social Voice Use Before and After a Period of 2.5 Hours
From the Acoustic Parameters, Perceptual, Selfperception of Vocal Effort and All Parameters Together
Parameters
Acoustic
Perceptual
Self-perception
of vocal effort
All parameters
Discriminant analysis of the perceptual analysis parameters revealed that the perceptual analysis of vocal quality
separated the professional voice users before voice use
(Table 6). The samples were separated by breathiness (B)
(94.6%), degree of dysphonia (G) (82.1%), and roughness
(R) (82.1%).
Discriminant analysis of the three parameters grouped
(acoustic, perceptual analysis, and self-perception) differentiated 70.73% of the samples of the professional voice users group
before the vocal use and 61.10% at the after vocal use (Table 6),
demonstrating a difference between these moments. The samples were separated by the perceptual evaluation of breathiness
(B) (60.3%), degree of dysphonia (G) (60.3%) and roughness
(R) (60.3%), and acoustic parameter F0 median (59.4%).
Analysis of canonical correlation explored several levels of
correlations of the acoustic parameters, perceptual analysis,
and self-perception of vocal effort between the groups before
TABLE 4.
Values of the Perception of Vocal Effort in Group of
Professional Vocal Use Before and After the Interval of
2.5 Hours
Before
After
Values
0
1
2
3
4
5
6
7
8
9
10
12
05
05
03
01
03
01
01
0
0
0
38.7
16.1
16.1
09.7
03.2
09.7
03.2
03.2
0
0
0
04
03
05
07
08
01
02
01
0
0
0
12.9
09.7
16.1
22.6
25.8
03.2
06.4
03.2
0
0
0
260.e48
369
304
673
621
553
1174
269
265
534
441
313
754
377
446
823
125
197
322
476
486
962
307
435
742
746
750
1496
746
750
1496
745
751
1496
748
748
1496
Correct
(%)
49.46
59.47
54.48
83.24
26.27
54.68
36.11
64.71
50.47
58.96
58.16
58.56
DISCUSSION
Among professional voice users, teachers are certainly increasingly vulnerable to develop dysphonia.23,24 The development of
dysphonia in teachers is because of multiple factors including
environmental (noise, dust, and smoke), organizational
(excess and demand of work and lack of material) and
individual (age, allergy, poor hydratation, stress, and so
on.).25 Therefore, the health of teachers, including their voice,
TABLE 6.
Confusion Matrix for the Estimation Sample Group of
Professional Voice Use Before and After a Period of
2.5 Hours From the Acoustic Parameters, Perceptual,
Self-Perception of Vocal Effort and All Parameters
Together
Parameters
Acoustic
Perceptual
Self-perception
of vocal effort
All parameters
354
207
561
408
301
709
242
246
488
377
212
589
189
328
517
123
246
369
290
300
590
156
333
489
543
535
1078
531
547
1078
532
546
1078
533
545
1078
Correct
(%)
65.19
61.31
63.27
76.84
44.97
60.67
45.49
54.95
50.28
70.73
61.10
65.86
260.e49
is highly related to their environment and characteristics of academic management.26
The literature refers to the most common signs of dysphonia:
roughness, vocal fatigue, ardor, sore throat and cervical pain,
difficulty in sustaining the voice, variations in F0, absence of
vocal capacity and vocal projection, loss of vocal efficiency,
and less vocal resistance even aphonia.27 Vocal fatigue is a
prevalent symptom among teachers1,4 yet its definition,
characteristics, and mechanisms are still uncertain.7
Many authors have attempted to indentify characteristics
of vocal fatigue via acoustic14,28,29 and aerodynamic
analyses,15,28 as well as laryngeal13,14 and EMG12 characteristics; however no study has showed a strong correlation between
these parameters and vocal loading.
This study used long-term acoustic measures in continuous
speech (Tables 1 and 2), associated with the perceptual
analysis and vocal self-perception (Tables 3 and 4), in an
attempt to analyze if these parameters could be more
sensitive to the vocal fatigue process, functioning as an index
of fatigue.
When the literature is analyzed in the context of vocal
loading task, many questions remain. For example, the time
and type of vocal use has not been characterized. Analysis of
the duration of voice use varies significantly; from 15 minutes12;
30 minutes30; 45 minutes10; 2 hours28 until 8 hours.4 All of
these studies, however, failed to demonstrated results highly
related to vocal fatigue, with vast variability of results.
Regarding the type of voice use, the literature suggests value
in performing voice analysis in real work conditions14; in
reading at high intensity,12,28 and with sustained vowels.16
However, indexes of vocal fatigue have not been characterized,
demonstrating that vocal fatigue is a multifactorial process
influenced by anatomical and physiological idiosyncrasies.7,10
This study used an interval of 2 hours and 30 minutes,
because it represents the period of continuous vocal use in Brazilian schools. The speech tasks were considered consistent
with professional voice use of teachers or social vocal use, in
real-life scenarios. Therefore, environmental factors were not
controlled, although they interfere in vocal production.26
Our results demonstrate that long-term acoustic measures
and self-perception of vocal use, when analyzed jointly did
not strongly differentiate social vocal use (Table 5) and professional vocal use (Table 6). Concerning social vocal use, the
discriminant analysis of the perceptual analysis of the vocal
quality separated 83.34% of the voice samples before vocal
use (Table 5), demonstrating that the voices differentiated
only before social vocal use, being separated by the perceptual
parameters of degree of dysphonia (G) and roughness (R). This
result was also observed in the group of professional voice
users, separating 76.84% the vocal samples before vocal use,
related to the breathiness (B), degree of dysphonia (G), and
roughness (R) parameters (Table 6).
Discriminant analysis of the three parameters associated
(acoustic, perceptual analysis, and self-perception) showed
that the results, when grouped, separated 70.73% of the vocal
samples in the group of professional voice users before vocal
use (Table 6). The samples were separated by the perceptual
^ rtes Gama, et al
Ana Cristina Co
2. Gama ACC, Bicalho VS, Valentim AF, Bassi IB, Teixeira LC,
Assunc~ao AA. Vocal self-perception of speech therapy after discharge: a
prospective study. CEFAC. 2012;14:714720.
3. Kraemer WJ, Adams K, Cafarelli E, et al. American College of Sports Medicine position stand. Progression models in resistance training for healthy
adults. Med Sci Sports Exerc. 2002;34:364380.
4. Laukkanen AM, Ilomaki I, Leppanen K, Vilkman E. Acoustic measures and
self-reports of vocal fatigue by female teachers. J Voice. 2008;22:283289.
5. Titze IR. Principles of Voice Production. 2nd ed. Englewood Cliffs, NJ:
Prentice Hall; 1994:361367.
6. Scherer RC, Titze IR, Raphael BN, Wood RP, Ramig LA, Blager RF. Vocal
fatigue in a trained and an untrained voice user. In: Baer T, Sasaki C,
Harris K, eds. Laryngeal Function in Phonation and Respiration. San
Diego, CA: Singular Publishing Group; 1991:533555.
7. Welham NV, Maclagan MA. Vocal fatigue: current knowledge and future
directions. J Voice. 2003;17:2130.
8. Gelfer MP, Andrews ML, Schmidt CP. Effects of prolonged loud reading on
selected measures of vocal function in trained and untrained singers. J
Voice. 1991;5:158167.
9. Vilkman E, Lauri E, Alku P, Sala E, Shivo M. Effects of prolonged oral
reading on F0, SPL, subglottal pressure, and amplitude characteristics of
glottal flow waveforms. J Voice. 1999;13:303315.
10. Kelchner LN, Lee L, Stemple JC. Laryngeal function and vocal fatigue after prolonged reading in individuals with unilateral vocal fold paralysis. J
Voice. 2003;17:513528.
11. Boucher VJ, Lamontagne M. Effects of speaking rate on the control of
vocal fold vibration: clinical implications of active and passive aspects of
devoicing. J Speech Lang Hear Res. 2001;44:10051014.
12. Boucher VJ, Ayad T. Physiological attributes of vocal fatigue and their
acoustic effects: a synthesis of findings for a criterion-based prevention
of acquired voice disorders. J Voice. 2010;24:324336.
13. Stemple JC, Stanley J, Lee L. Objective measures of voice production in
normal subjects following prolonged voice use. J Voice. 1995;9:127133.
14. Doellinger M, Lohscheller J, McWhorter A, Kunduk M. Variability of
normal vocal fold dynamics for different vocal loading in one healthy subject investigated by phonovibrograms. J Voice. 2009;23:175181.
15. Solomon NP, DiMattia MS. Effects of a vocally fatiguing task and systemic
hydration on phonation threshold pressure. J Voice. 2000;14:341362.
16. Chang A, Karnell MP. Perceived phonatory effort and phonation threshold
pressure across a prolonged voice loading task: a study of vocal fatigue. J
Voice. 2004;18:454466.
17. Lowell SY, Colton RH, Kelley RT, Mizia SA. Predictive value and discriminant capacity of cepstral- and spectral-based measures during continuous
speech. J Voice. 2013;27:393400.
18. Vintturi J, Alku P, Lauri ER, Sala E, Sihvo M, Vilkman E. Objective analysis of vocal warm-up with special reference to ergonomic factors. J Voice.
2001;15:3653.
19. Camargo Z, Madureira S. Voice quality analysis from a phonetic perspective: voice profile analysis scheme profile for Brazilian Portuguese
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
260.e50