Documentos de Académico
Documentos de Profesional
Documentos de Cultura
520
frequency levels17,20,21 (eg, Awan and Awan17 recently reported
that the knee was present in 76% of the males and 36% of the females that they studied).
In humans, most of the acoustic energy is generated by the
cessation of the transglottal airflow.22 The relative duration of
the closed phase during a glottal cycle has a direct influence on
the shape of the glottal airflow pulse and hence the generation of
stronger high-frequency partials in the radiated sound source spectrum.23,24 Therefore, it may be expected that differences in the
relative duration and the quality or type of vocal fold contact
during the glottal cycle (as discussed in the aforementioned
literature review regarding adult male vs female EGG
waveforms) will have an effect on the spectral characteristics of
the acoustic voice signal. Holmberg et al25 examined the relationships between aerodynamic, acoustic, and EGG measures of voice
in consonant-vowel syllable and vowel production. Results
showed weak but significant negative relationships between
EGG adduction quotient and harmonic-to-harmonic (H1-H2)
and harmonic-to-formant (H1-F1 and H1-F3) amplitude differences, whereas weak positive correlations were observed between
H1-F3 noise and F1-F3 noise. These authors hypothesized that
gradual closing movements of a somewhat abducted membranous
portion of the vocal folds would result in relatively sinusoidal
glottal waveforms and small adduction quotients and that the corresponding acoustic spectrum would be characterized by a first
harmonic with relatively high amplitude, a steep overall spectral
slope, and an attenuated F1 peak amplitude (because of a relatively
wide F1 bandwidth). However, it was reported that strong relationships (rs > 0.70) between EGG adduction quotients and the aforementioned spectral characteristics were observed in 35% of the
subjects. Winkler et al26 examined the effect of aging on EGG,
acoustic, and perceptual analyses of the voice. Results indicated
a characteristic spectral distribution in the // vowel of male voices. Long-term average spectra (LTAS) of the male voices
perceived as old had relatively little harmonic energy between 2
and 4 kHz, whereas younger male voices showed distinct spectral
harmonics up to 4 kHz. The spectral characteristics for the young
males coincided with EGG waveforms that showed a steep increase of vocal fold contact in the closing and a more extended
opening phase with the instant of glottal opening clearly visible
in the EGG signal as a knee. In contrast, the EGG waveforms
for the older males were nearly sinusoidal with reduced steepness
of the increasing signal slope. Bestebreurtje and Schutte27 examined EGG and spectral characteristics of the belting singing style
in a single subject study. Results indicated that the loud bright
sound of the belting style is achieved by enhanced higher harmonics in the acoustic spectrum and a corresponding raising of
the CQEGG above 52% (the authors speculated that this was an
apparent threshold value for belting).
In addition to the examination of isolated spectral harmonic
amplitudes, one of the methods that may be used to summarize
the spectral characteristics of the acoustic voice signal is by
examining the spectral moments of the LTAS.28 In the computation of spectral moments, the spectral values are treated as a
random probability distribution, with the first spectral moment
being the spectral mean (a measure of the spectrums central
tendency in hertz; the center of gravity); the second spectral
METHODOLOGY
The Bloomsburg University of Pennsylvania Institutional
Review Board approved the methodology used in this study.
Shaheen N. Awan, et al
Participants
Participants consisted of 24 adult males and 24 adult females
between the ages of 18 and 30 years (mean age 21.33 years,
SD 1.49), with typical voice characteristics of pitch, loudness, and quality as determined by a certified speechlanguage pathologist. The mean fundamental frequency (F0)
for the adult male subjects was 112.22 Hz (SD 16.99);
mean F0 for the adult females was 210.00 Hz (SD 16.16).
All subjects passed a hearing screening (25 dB at 0.5, 1, 2,
and 4 kHz).
Task and equipment
Participants produced the vowel // at their comfortable pitch
and loudness for approximately 3 seconds in duration. Each
vowel was repeated three times. During vowel productions,
EGG signals were recorded using a KayPENTAX Model
6103 Electroglottograph (KayPENTAX Inc., Montvale, NJ).
During EGG recording, the time-varying degree of admittance
of an electrical signal presented at the level of the vocal folds
was recorded via two gold-plated EGG electrodes placed on
either side of subjects laryngeal prominence. The signal was
high-pass filtered at 20 Hz to remove direct current (DC) float
and low-frequency artifacts. Sigma Creme electrode gel (Parker
Laboratories, Fairfield, NJ) was used to enhance conductivity,
and optimal positioning of the electrodes was checked by monitoring the EGG signal amplitude in real-time using a Protek
6502 Oscilloscope (Protek Test & Measurement, Norwood,
NJ). During vowel production, the acoustic signal was recorded
using a Shure SM10 A microphone (Shure Inc., Niles, IL)
placed approximately 45 off-center at a constant one inch
(2.5 cm) distance from the participants mouth. EGG and acoustic signals were digitized simultaneously using KayPENTAX
Model 4500 Computerized Speech Lab (CSL) version 3.4.1
and Real-Time EGG software (KayPENTAX) at 44.1 kHz,
16 bits of resolution per channel.
Signal analyses
For each trial of the vowel //, a central 1 second portion of the
combined two-channel acoustic + EGG signal was edited for
further analyses. Because the EGG signal can be adversely
affected by the presence of low-frequency modulations as well
as random broadband noise, the channel containing the EGG
signal was bandpass filtered using a phase-preserving Blackman
filter (pass band of 201500 Hz) using the CSL main program,
consistent with filtering procedures used in a number of previous
studies.18,40 EGG signals were then analyzed using a custom
program written using SciLab version 5.3.3 open source
software for numerical computation (Scilab Enterprises,
Versailles, France) as previously described and validated in
Awan and Awan.17 The analysis algorithm uses the derivative
method for determining the CQEGG, in which the derivative
maxima corresponds to the vocal fold contacting point and the
derivative minima corresponds to the vocal fold decontacting
point.17,18,41 In addition to CQEGG, mean fundamental
frequency (F0) estimates were also obtained for each trial of
the vowel // via autocorrelation-guided peak picking of the
CQEGG maxima.
521
522
TABLE 1.
Correlations Between the CQEGG and Spectral Moments Across All subjects (N 48)
Dependent Variable
Contact quotient
Pearson correlation
Sig. (2-tailed)
Index of determination
Spectral mean
Pearson correlation
Sig. (2-tailed)
Index of determination
Spectral standard deviation
Pearson correlation
Sig. (2-tailed)
Index of determination
Spectral skewness
Pearson correlation
Sig. (2-tailed)
Index of determination
Spectral kurtosis
Pearson correlation
Sig. (2-tailed)
Index of determination
Contact
Quotient
1
Spectral Mean
Spectral Standard
Deviation
0.24
0.094
0.06
0.40*
0.006
0.16
0.37y
0.01
0.14
0.36y
0.011
0.13
0.06
0.672
<0.01
0.44*
0.002
0.19
0.37y
0.01
0.14
0.003
0.98
<0.01
0.29y
0.045
0.08
Spectral Skewness
Spectral Kurtosis
0.87*
<0.001
0.76
1
Shaheen N. Awan, et al
523
TABLE 2.
Summary Statistics for Contact Quotient (CQEGG) and Spectral Moments in the Adult Male and Female Subgroups
Dependent Variable
Gender
Mean
Standard Deviation
Female
Male
Female
Male
Female
Male
Female
Male
Female
Male
24
24
24
24
24
24
24
24
24
24
36.22
46.96
736.04
670.63
376.97
307.13
2.05
3.21
23.35
42.84
10.63
9.17
120.44
101.57
70.22
66.58
1.41
1.82
23.17
40.24
TABLE 3.
Summary Statistics for Contact Quotient (CQEGG) and Spectral Moments in the Concave Down (Knee) and Concave Up (No
Knee) Subgroups
Dependent Variable
Profile
Mean
Standard Deviation
Concave up
Concave down
Concave up
Concave down
Concave up
Concave down
Concave up
Concave down
Concave up
Concave down
24
24
24
24
24
24
24
24
24
24
38.68
44.50
745.16
661.51
336.9
347.14
2.31
2.96
33.46
32.73
12.74
8.70
101.17
114.76
80.97
72.79
1.64
1.76
39.31
28.45
524
FIGURE 2. Mean LTAS for high versus low CQ subjects (CQ categorization was determined using the overall mean CQ across all subjects
(CQEGG 41.59%) as the cutoff point). Raw LTAS (A) and Loess smoothed LTAS (B) are displayed. High CQ subjects were observed to have significantly lower mean spectral standard deviation, significantly higher spectral skewness and significantly higher spectral kurtosis versus low CQ
subjects.
Figure 3 provides a histogram showing the range of CQEGG
values within each adult male and female subgroup.
Spectral mean. No significant main effects of profile (F(1,
44) 3.97, P 0.053; h2 0.074) or gender (F(1,
44) 1.17, P 0.29; h2 0.022) and no significant interaction
between gender and profile (F(1, 44) 1.13, P 0.29;
h2 0.021). Although nonsignificant, the main effect of EGG
profile was considered to have an effect size that indicated a
moderate degree of practical significance, with the concave
down/knee group observed to have lower average spectral
mean than the no knee profile group (661.51 Hz vs
745.16 Hz, respectively; Table 3).
Spectral SD. Analyses showed significant main effects of
gender (F(1, 44) 18.19, P < 0.001; h2 0.29) and profile
(F(1, 44) 5.11, P 0.029; h2 0.082), whereas the interaction
of gender and profile was statistically nonsignificant (F(1,
FIGURE 3. Histogram showing the relative distribution of CQEGG values for the adult female and male subgroups.
Shaheen N. Awan, et al
525
FIGURE 4. Mean LTAS for concave down/knee versus concave up/no knee EGG decontacting phase profile subjects. Raw LTAS (A) and Loess
smoothed LTAS (B) are displayed. The concave down/knee group was observed to have significantly greater spectral standard deviation and a strong
trend (moderate strength effect size) for a lower spectral mean than concave up/no knee subjects.
526
FIGURE 5. Mean LTAS for concave adult male versus female subjects. Raw LTAS (A) and Loess-smoothed LTAS (B) are displayed. Men were
observed to have significantly greater spectral skewness and spectral kurtosis than women, whereas women presented with significantly greater spectral standard deviation versus men.
finding was particularly prominent and statistically significant in
women (rs 0.64 and 0.49 for correlations between spectral
mean and spectral kurtosis and skewness, respectively) although
weak and nonsignificant in men. Calculation of coefficients of
variation (CV) indicated that there was greater relative variation
within the adult female versus male group for the spectral mean
(CVs 16.36% vs 15.15%, respectively), kurtosis
(CVs 99.23% vs 93.93%, respectively), and particularly skewness (CVs 68.78% vs 56.70%, respectively). As previously
mentioned, increased variance and range of data often aids in
revealing correlational trends and may have resulted in the stronger correlations observed between these aforementioned spectral
moments in adult females versus males. Across all subjects, spectral skewness and spectral kurtosis were significantly and
strongly correlated. Spectral energy concentration in the lower
harmonics (ie, more positively skewed) tends to result in a
more peaked spectrum (ie, higher kurtosis) with the possibility
of a heavier tail in the spectral distribution (spectral energy
may move from the shoulders of the spectral distribution to
the higher frequency region). In contrast, as energy becomes
more evenly distributed throughout the spectrum, the spectrum
becomes flatter and spectral kurtosis is reduced.
Effect of EGG profile on spectral moments
The presence of concave down (knee) versus concave up and
other decontacting profiles (no knee) was observed to be
associated with moderate strength effects (h2 values >0.06)
for spectral mean (statistically nonsignificant at P 0.053)
and spectral SD (statistically significant at P 0.029). Subjects
with concave down/knee profiles were observed to have a
strong trend for lower spectral means than subjects with
concave up and other decontacting profiles. Given the profile
of concave down/knee-shaped EGG waveforms, it may be expected that the decontacting point is more abrupt in these subjects as opposed to the more gradual linear decontacting of the
concave up waveforms.4,20 More abrupt decontacting was also
observed to coincide with longer duration vocal fold contact
(increased CQEGGs) and may also be consistent with
Shaheen N. Awan, et al
527
CONCLUSIONS
The results of this study clearly show that degree of vocal fold
contact and differences in decontacting phase profile have an
effect on the spectral characteristics of the voice. However,
the strength of correlations between CQEGG values and measures of spectral moments tended to be relatively weak and
528
only accounted for approximately 1316% of the variation in
spectral distribution characteristics. These findings are consistent with the relatively weak relationships between CQEGG
and spectral measures reported by Holmberg et al25 and recent
observations by Alipour et al49 indicating that degree of glottal
adduction may influence certain spectral characteristics in relatively unpredictable ways. Although somewhat stronger associations were observed between EGG profile (knee vs no knee)
and spectral characteristics (such as a tendency for lower spectral means (moderate effect size) but significantly greater spectral SD (moderate effect size) observed in the acoustic spectral
distributions associated with concave down/knee-shaped
EGG waveforms), the results of this study stress the importance
of the transformative role of the supraglottal vocal tract in producing an acoustic output that maintains some of the characteristics of the glottal source, but which modifies the source
characteristics in ways not completely accounted for by single
parameters such as CQEGG or EGG profile. Future studies that
examine EGG and acoustic measures with larger groups of participants may potentially allow for more clarity regarding the
presence of possible correlations between EGG measures and
spectral moments. In addition, future studies may compare
and contrast the spectral characteristics of the EGG waveform
itself (as in the study by Libeaux64) with the spectral characteristics of the airborne acoustic signal that reflect the characteristics of the vocal tract filter.
Acknowledgment
This study was partially supported by the European Social Fund
project OP VK CZ.1.07/2.3.00/30.0004 POST-UP (CTH).
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
REFERENCES
1. Titze I. Physiologic and acoustic differences between male and female voices. J Acoust Soc Am. 1989;85:16991707.
2. Chen Y, Robb M, Gilbert H. Electroglottographic evaluation of gender and
vowel effects during modal and vocal fry phonation. J Speech, Lang Hear
Res. 2002;45:821830.
3. Hanson HM, Chuang ES. Glottal characteristics of male speakers: acoustic
correlates and comparison with female data. J Acoust Soc Am. 1999;106:
10641077.
4. Higgins MB, Schulte L. Gender differences in vocal fold contact computed
from electroglottographic signals: the influence of measurement criteria. J
Acoust Soc Am. 2002;111:1865.
5. Echternach M, Dippold S, Sundberg J, Arndt S, Zander MF, Richter B.
High-speed imaging and electroglottography measurements of the open
quotient in untrained male voices register transitions. J Voice. 2010;24:
644650.
6. Hanson HM. Glottal characteristics of female speakers: acoustic correlates.
J Acoust Soc Am. 1997;101:466.
7. Orlikoff RF, Golla ME, Deliyski DD. Analysis of longitudinal phase differences in vocal-fold vibration using synchronous high-speed videoendoscopy and electroglottography. J Voice. 2012;26:816.e13816.e20.
8. S
odersten M, Herteg
ard S, Hammarberg B. Glottal closure, transglottal
airflow, and voice quality in healthy middle-aged women. J Voice. 1995;
9:182197.
9. Holmberg EB, Hillman RE, Perkell JS. Glottal airflow and transglottal air
pressure measurements for male and female speakers in soft, normal, and
loud voice. J Acoust Soc Am. 1988;84:511529.
10. Fabre P. Un procede electrique percuntane dinscription de laccolement
glottique au cours de la phonation: glottographie de haute frequence; premiers resultats. [A non-invasive electric method for measuring glottal
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
closure during phonation: high frequency glottogr]. Bull Acad Natl Med.
1957;141:6669.
Baken RJ. Electroglottography. J Voice. 1992;6:98110.
Scherer R, Druker D, Titze I. Electroglottography and direct measurement
of vocal fold contact area. In: Fujimura O, ed. Vocal Fold Physiology: Voice
production, Mechanisms and Functions, Vol. 2. New York: Raven Press;
1988:279290.
Orlikoff R. Assessment of the dynamics of vocal fold contact from the electroglottogram: data from normal male subjects. J Speech Hear Res. 1991;
34:10661073.
Rothenberg M, Mahshie JJ. Monitoring vocal fold abduction through vocal
fold contact area. J Speech Hear Res. 1988;31:338351.
Titze IR. A four-parameter model of the glottis and vocal fold contact area.
Speech Comm. 1989;8:191201.
Childers DG, Hicks DM, Moore GP, Alsaka YA. A model for vocal fold
vibratory motion, contact area, and the electroglottogram. J Acoust Soc
Am. 1986;80:13091320.
Awan SN, Awan JA. The effect of gender on measures of electroglottographic contact quotient. J Voice. 2013;27:433440.
Herbst C, Ternstrom S. A comparison of different methods to measure the
EGG contact quotient. Logoped Phoniatr Vocol. 2006;31:126138.
Robb MP, Simmons JO. Gender comparisons of childrens vocal fold contact behavior. J Acoust Soc Am. 1990;88:13181322.
Herbst CT, Fitch WTS, Svec JG. Electroglottographic wavegrams: a technique for visualizing vocal fold dynamics noninvasively. J Acoust Soc Am.
2010;128:30703078.
Cecconello LA. Qualitative and quantitative aspects of the electroglottography in singers with different glottic contact patterns. In: 28th World
Congress of the International Association of Logopedics and Phoniatrics.
Cordoba, Argentina: Sanatorio del Salvador; 2010:779782.
Fant G. Quarterly progress and status report: glottal source and excitation
analysis. STL-QPSR. 1979;20:85107.
Doval B, DAlessandro C, Henrich N. The spectrum of glottal flow models.
Acta Acustica. 2006;92:10261046.
Titze I. Principles of Voice Production. 2nd ed. Salt Lake City, UT: National
Center for Voice and Speech; 2000.
Holmberg EB, Hillman RE, Perkell JS, Guiod PC, Goldman SL. Comparisons among aerodynamic, electroglottographic, and acoustic spectral measures of female voice. J Speech Hear Res. 1995;38:12121223.
Winkler R, Bruckl M, Sendlmeier W. The aging voice: an acoustic, electroglottographic and perceptive analysis of male and female voices. In: 15th
ICPhS. Barcelona; 2003:14.
Bestebreurtje ME, Schutte HK. Resonance strategies for the belting style:
results of a single female subject study. J Voice. 2000;14:194204.
Buder E, Kent R, Kent J, Milenkovic P, Workinger M. Formoffa: an automated formant, moment, fundamental frequency, amplitude analysis of
normal and disordered speech. Clin Linguist Phon. 1996;10:3154.
Forrest K, Weismer G, Milenkovic P, Dougall RN. Statistical analysis of
word-initial voiceless obstruents: preliminary data. J Acoust Soc Am.
1988;84:115123.
Kardach J, Wincowski R, Metz DE, Schiavetti N, Whitehead RL,
Hillenbrand J. Preservation of place and manner cues during simultaneous
communication: a spectral moments perspective. J Commun Disord. 2002;
35:533542.
Harwardt C, Gottsman F, Noubours S. On the relationship between vocal
effort and spectral moments. In: SiMPE 2011 Proceedings.;14.
DeCarlo L. On the meaning and use of kurtosis. Psychol Methods. 1997;2:
292307.
Dytham C. Choosing and Using Statistics: A Biologists Guide. 3rd ed. Oxford, UK: Wiley Blackwell; 2011.
Lowell SY, Colton RH, Kelley RT, Hahn YC. Spectral- and cepstral-based
measures during continuous speech: capacity to distinguish dysphonia and
consistency within a speaker. J Voice. 2011;25:e223e232.
Colton RH, Paseman A, Kelley RT, Stepp D, Casper JK. Spectral moment
analysis of unilateral vocal fold paralysis. J Voice. 2011;25:330336.
Tanner K, Roy N, Ash A, Buder EH. Spectral moments of the long-term
average spectrum: sensitive indices of voice change after therapy? J Voice.
2005;19:211222.
Shaheen N. Awan, et al
37. Houtz DR, Roy N, Merrill RM, Smith ME. Differential diagnosis of muscle
tension dysphonia and adductor spasmodic dysphonia using spectral moments of the long-term average spectrum. Laryngoscope. 2010;120:749757.
38. Dromey C. Spectral measures and perceptual ratings of hypokinetic dysarthria. J Med Speech Lang Pathol. 2003;11:8594.
39. Uchanski R, Geers A. Acoustic characteristics of the speech of young
cochlear implant users: a comparison with normal-hearing age-mates.
Ear Hear. 2003;24(1 Suppl.):90S1055S.
40. Vieira MN, McInnes FR, Jack MA. Comparative assessment of electroglottographic and acoustic measures of jitter in pathological voices. J Speech
Lang Hear Res. 1997;40:170182.
41. Henrich N, dAlessandro C, Doval B, Castellengo M. On the use of the derivative of electroglottographic signals for characterization of nonpathological phonation. J Acoust Soc Am. 2004;115:1321.
42. Owen S, Froman R. Focus on qualitative methods uses and abuses of the
analysis of covariance. Res Nurs Health. 1998;21:557562.
43. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed.
Hillsdale N.J: L. Erlbaum Associates; 1988.
44. Ferguson CJ. An effect size primer: a guide for clinicians and researchers.
Prof Psychol Res Pract. 2009;40:532538.
45. Kraemer H, Morgan G, Leech N, Gliner J, Vaske J, Harmon R. Measures of clinical significance. J Am Acad Child Adolesc Psychiatry. 2003;42:15241529.
46. Schuele C, Justice L. The importance of effect sizes in the interpretation of
research: primer on research: part 3. ASHA Lead. 2006.
47. Thalheimer W, Cook S. How to calculate effect sizes from published
research: a simplified methodology. Somerville, MA: A Work-Learning
Research; 2002.
48. Holm SA. Simple sequentially rejective multiple test procedure. Scand J
Stat. 1979;6:6570.
49. Alipour F, Scherer R, Finnegan E. Measures of spectral slope using an
excised larynx model. J Voice. 2012;26:403411.
50. Kreiman J, Shue Y-L, Chen G, et al. Variability in the relationships among
voice quality, harmonic amplitudes, open quotient, and glottal area waveform shape in sustained phonation. J Acoust Soc Am. 2012;132:26252632.
51. Awan SN, Giovinco A, Owens J. Effects of vocal intensity and vowel type
on cepstral analysis of voice. J Voice. 2012;26:670.e15670.e20.
529