Académique Documents
Professionnel Documents
Culture Documents
Summary: Objectives. Presbyphonia is the sequence of physiological events related to the process of senility of
the vocal folds. The aim of our analysis was to provide deeper knowledge of presbyphonia, raising awareness of this
condition as well as giving basic suggestions on how to treat related vocal alterations.
Study Design. This is a randomized study.
Methods. In 2015, we conducted a study on 182 subjects. Each participant underwent an ENT examination (video-
laryngo-stroboscopy and subjective acoustic analysis using the General degree of dysphonia; degree of voice Instability;
degree of voice Roughness; degree of voice Breathiness; degree of voice Asthenia; degree of voice Strain (GIRBAS)
scale) and a logopedic examination (anamnesis, medical history, and acoustic voice analysis using the free software
Praat).
Results. The comparison between the voice of young people and the seniors showed significant differences for the fol-
lowing Praat-analyzed acoustic parameters: modal fundamental frequency (F0) in women (P < 0,0001), fraction of locally
unvoiced frames (P < 0,0001), number of voice breaks (P < 0,0001), jitter local (P < 0,0001), jitter local abs (P < 0,0001),
jitter rap (P < 0,0001), jitter ppq5 (P < 0,0001), shimmer local (P < 0,0001), shimmer local dB (P < 0,0001), shimmer apq3
(P < 0,0001), shimmer apq5 (P < 0,0001), mean N/H (P < 0,0001), and mean H/N (P < 0,001), for both sexes.
Conclusions. The Praat was confirmed to be a useful tool to detect the existence of the variation of the speech pa-
rameters in relation to aging and to quantify statistically significant differences that show a general deterioration in the
voice quality, defined numerically. This might lead to a phoniatric treatment or speech therapy, which could improve
patients’ quality of life, leading to better vocal performance and social and communicative interaction.
Key Words: presbyphonia–aging voice–Praat–voice analysis–dysphonia.
analyses include the multiparametric voice assessment, the aerodigestive tract disorders, and surgical procedures) family
maximum phonation time test, and the phonetogram, whereas diseases; social history (including occupation, living
the semiobjective ones comprise the spectrogram analysis. arrangements, tobacco and alcohol consumption, environ-
Nowadays, the multiparametric voice assessment is easily con- mental pathogens exposure, and voice misuse and abuse);
ducted through a series of computer programs, among which the regular and acute medications.
“Praat” is an easily accessible and user-friendly software avail- • Acoustic voice analysis using the free software Praat.
able for free and developed by P. Boersma and D. Weenink from
the Phonetics Department of the University of Amsterdam SAMPLING AND INCLUSION/EXCLUSION CRITERIA
(www.Praat.org). We analyzed 182 subjects: 109 females (59.89%) and 73 males
Praat’s voice analysis output consists of a waveform of the (40.10%).
analyzed signal, a spectrogram, and a voice report. Inclusion criteria were: no physiological or organic anoma-
The Voice Report is useful for the evaluation of a series of lies shown during the video-laryngo-stroboscopy, and no evidence
vocal parameters related to: of common presbylarynx features such as vocal fold bowing and
prominence of vocal processes.
– F0 (median pitch, mean pitch, minimum pitch, maximum Exclusion criteria were neurological disorders affecting la-
pitch); ryngeal functions, chronic degenerative diseases, vocal cord
– temporary vocal interruption: Voicing (fraction of locally lesions (nodules, polyps, granulomas, cysts, and paralysis), dys-
unvoiced frames; number of voice breaks; degree of voice functional dysphonia due to an incomplete glottic closure
breaks) (posterior triangle or oval-like glottis), patients with allergies or
– micro perturbation of frequency: Jitter (jitter local; jitter other pharyngeal-laryngeal acute inflammations or alterations due
local absolute; jitter rap; jitter ppq5; jitter ddp); to pharmacological effects, and previous vocal-tract surgery.
– microperturbation of amplitude: Shimmer (shimmer local; According to these criteria, 40 subjects could not be in-
simmer local dB; shimmer apq3; shimmer apq5; shimmer cluded in the analysis: 27 females (67.5%) and 13 males (32.5%).
apq11; shimmer ddp); Consequently, we designed a clinical study to assess the voice
– spectral energy balance between harmonic and dishar- of 142 patients.
monic components: signal noise (mean autocorrelation, Based on the references in recent scientific articles,8 we chose
mean H/N, mean N/H).7 the limit of 60 years of age to divide the sample in two different
groups. We also considered the voice differences according to
Despite the wide range of studies conducted on older pe- the gender, comparing younger and elderly people in relation to
ople’s voices, there is still a lack of objective and univocal criteria their sex. Thus, we subdivided our sample in four subcatego-
to clearly define presbyphonia. ries: (group 1) 21 young men aged between 21 and 54 years (age
Running the “Praat” software has been successful in provid- mean = 35.79); (group 2) 39 elderly men aged between 63 and
ing an objective benchmark, which confirms the symptoms 93 years (age mean = 74.43); (group 3) 35 young women aged
previously described from a purely subjective point of view. between 20 and 55 years (age mean = 38.49); and (group 4) 47
The final aim of this analysis was to provide deeper knowl- elderly women aged between 61 and 84 years (age mean = 72.76).
edge of presbyphonia, raising awareness of this condition as well
as giving basic suggestions on how to treat related vocal alterations. Materials
For the recording, we used:
MATERIALS AND METHODS
Between May and September 2015, we conducted a study on – a personal computer;
182 subjects recruited from the “Università dell’età Libera” (lit- – the Praat software;
erally translated as “Free Age University”) in Colle di Valdelsa. – An omnidirectional microphone with a G&BL condens-
After signing an informed consent module, each participant un- er (Frequency Response 20 Hz ~ 16kHz, Impedance 2.0
derwent a phoniatrics and logopedics examination at “St. Maria KΩ), suitable for all sound cards, positioned at a dis-
Le Scotte” Hospital, in Siena, Italy. The examination was con- tance of 20 cm from the lips and with an axial tilt of
ducted in two different phases: 45 °, thus eliminating disruption related to the airflow.
ENT examination:
Praat software: Setting
• Video-laryngo-stroboscopy, to evaluate laryngeal mor- Before the examination, it is necessary to configure how the spec-
phology, motility, and vocal folds vibration during speech. trogram is computed and displayed according to “view range,”
• Subjective acoustic analysis using the GIRBAS scale. “window length,” and “dynamic range.” The standard range rec-
ommended by P. Boersma and D. Weenink8 is from 75 Hz to
Logopedic examination: 500 Hz. As advised, for our analysis, we set two different pitch
ranges:
• Chief complaint and history of the present illness, child-
hood diseases, past medical and surgical history (focusing – one specifically for female voices (100–500 Hz);
on allergies, gastroesophageal reflux disease (GERD), upper – the other specifically for male voices (75–300 Hz).
ARTICLE IN PRESS
Chiara Mezzedimi et al. Presbyphonia: a Study on 142 Patients with Praat 3
The recording was conducted by the same speech therapist shimmer apq5 (P < 0,0001), mean N/H (P < 0,0001), mean H/N
in the very same conditions, in the same room, with a sam- (P < 0,001), for both sexes.
pling rate of 22 050 Hz9 with a background noise of less than Differences in intensity absolute mean (P = 0,2225) for both
50 dB. men and women and F0 in young and old males (P = 0,3272)
were not statistically significant (Figure 1; in every figure the
Recording methods and analysis procedures abscissa shows the patients divided in four groups according to
As suggested by the Praat10 e-guidebook, after listening to an age: (group 1) young males, (group 2) elder males, (group 3)
example given by the examiner, participants were asked to pro- young females, and (group 4) elder females.
nounce the vowel “a” with an intensity of voice similar to a
normal conversation, with no changes in intensity and frequen- DISCUSSION
cy, for at least 5 seconds. The test was repeated three times. The aim of our research is to define any voice changes in
We then further proceeded recording with Praat and analyz- presbyphonia, objectifying them through an analysis con-
ing the three middle seconds of each spectrogram. We did not ducted with Praat. Our review of the scientific literature showed
take into consideration the beginning and the end of the pho- that the “aging voice” is generally identified as a “gradual weak-
nation, when the most relevant disturbances occur. Moreover, ening” of the voice. Clinical signs of presbyphonia are voice
with a segment of 3 seconds, it is certain that at least 110 cycles fatigue, hoarseness, difficulty in singing, decreased intensity, varia-
of vibration of the vocal folds are recorded.7 tions and reduction of tonal range, low phonation duration, and
For each subject, we evaluated the mean of the values ob- general deterioration of voice quality. Referring to GIRBAS scale,
tained in the three recordings. we can say that the deterioration is expressed by six param-
According to Società Italiana Foniatria e Logopedia (SIFEL) eters: general degree of dysphonia, degree of voice instability,
protocol,4 we also carried out the maximum phonation time (MPT) degree of voice roughness, degree of voice breathiness, degree
evaluation, asking our subject to repeat “a” three times and taking of voice asthenia, and degree of voice strain. Because there is
into account only the maximum value obtained from the three a correlation between GIRBAS scale parameters and acoustic
tests. spectral ones,11 we established a correspondence between the lit-
We therefore achieved an objective measurement with respect erature data and the objective evidence obtained through the use
to F0, fraction of locally unvoiced frames, number of voice breaks, of Praat (Table 1).
jitter local, jitter local absolute, jitter rap, jitter ppq5, shimmer The F0 modal analysis shows a lower frequency both in older
local, shimmer local dB, shimmer apq3, shimmer apq5, noise males and females, with a significant difference reported only
to harmonic ratio, and harmonic to noise ratio. Once the re- in women (Figure 2). This could be due to the postural change
cordings had been completed, we reported our data in Excel of the elderly, with forward flexion of the upper spine, which
format, divided according to the before-mentioned four categories. drags down the vocal cords. Even if the difference was not sig-
As regard the absolute intensity, we referred to the SIFER4 nificant, it should be noted that, as regards men, our results differ
protocol for spectroacoustic analysis, although it does not refer from the ones obtained by other authors.12,13 They reported that
specifically to Praat. However, in our analysis, we also took into F0 increases in older men, whereas it remains constant or de-
account anamnesis and videolaryngostroboscopy to compare our creasing in older women.14,15
subjects. Jitter (Figure 3) expresses the regularity of the vibratory cycle,
the perturbation of the F0 mean and is related to the degree of
RESULTS hoarseness (Roughness). It is significantly higher in the elderly
The nonparametric Kruskal-Wallis test has been applied to all of both sexes. Our results justify hoarseness as a clinical sign
the Voice Report parameters except for “Mean N/H,” which of presbyphonia.
follows a normal distribution (analysis of variance). Equally significant results (Figures 4–6) were obtained ana-
The aim of the analysis was to determine whether there were lyzing other parameters that test the same characteristics (jitter
significant differences in the medians between the two groups local abs, rap, ppq5) to reduce the error due to inadequate fun-
(young voices and aged voices) so as to infer and establish sys- damental period extraction.
tematic difference in the two vocal classes with fixed probability. Shimmer (Figure 7) expresses the perturbation of the glottic
The arithmetic means of all the categories analyzed have been vibration amplitude, which is the amplitude of the sound wave,
taken into account except for those following a non-Gaussian and it is related to the degree of voice breathiness and to inten-
distribution for which median values and 95% confidence in- sity variations. Higher values of shimmer justify what has been
tervals for the median have been considered (Table 1). observed in video-laryngo-stroboscopy: the “bowing” of the vocal
The comparison between the voice of young people and those folds because of the atrophy of the vocal muscles.
of the seniors showed significant differences for the following Thus, seniors’ voice is hypophonic and “blown.” Similar results
Praat-analyzed acoustic parameters: modal fundamental fre- (Figures 8–10) emerged studying other parameters (shimmer local
quency (F0) in women (P < 0,0001), fraction of locally unvoiced dB, apq3, and apq5) that analyze similar vocal characteristics.
frames (P < 0,0001), number of voice breaks (P < 0,0001), jitter The intensity (Figure 1) is associated with the degree of vocal
local (P < 0,0001), jitter local abs (P < 0,0001), jitter rap asthenia; according to our results, the intensity absolute mean
(P < 0,0001), jitter ppq5 (P < 0,0001), shimmer local (P < 0,0001), is lower in seniors than in younger subjects, although not with
shimmer local dB (P < 0,0001), shimmer apq3 (P < 0,0001), a significant difference.
ARTICLE IN PRESS
4 Journal of Voice, Vol. ■■, No. ■■, 2016
TABLE 1.
Means, CI for the Means, and Standard Deviations for Each Acoustic Parameter Analyzed with Praat
Men Women
Young Elderly Young Elderly
Intensity (dB) Arithmetic mean 67,78 67,73 67,68 64,98
95% CI for the mean 64,81–70,76 64,66–70,80 65,29–70,07 62,45–67,52
SD 6,52 9,47 6,95 8,63
Frequency (Hz) Arithmetic mean 141,64 139,56 217,54 177,98
95% CI for the mean 131,0–152,29 126,91–152,22 207,59–227,49 165,12–190,84
SD 23,38 39,02 28,97 43,8
Fraction unvoiced Median 0 0 0 0
95% CI for the median 0,00–0,00 0,00–0,68 0,00–0,00 0,00–0,30
SD 0 8,58 0 2,97
N° Voice breaks Median 0 0 0 0
95% CI for the median 0,00–0,00 0,00–0,00 0,00–0,00 0,00–0,00
SD 0 0,97 0 0,9
Jitter local(%) Median 0,284 0,65 0,275 0,493
95% CI for the median 0,234–0,377 0,345–0,770 0,244–0,304 0,406–0,690
SD 0,095 1,102 0,077 0,525
Jitter local abs (μs) Median 20,876 45,447 13,765 33,999
95% CI for the median 16,039–26,937 24,906–63,11 11,9–14,606 24,612–46,812
SD 8,837 120,708 2,731 37,391
Jitter rap (%) Median 0,189 0,325 0,178 0,262
95% CI for the median 0,164–0,223 0,169–0,429 0,146–0,203 0,227–0,409
SD 0,047 0,662 0,046 0,289
Jitter ppq5 (%) Median 0,183 0,362 0,173 0,261
95% CI for the median 0,149–0,256 0,189–0,416 0,156–0,207 0,231–0,354
SD 0,078 0,55 0,042 0,298
Shimmer local (%) Median 3,254 6,71 3,119 6,306
95% CI for the median 2,591–3,865 5,285–7,883 2,424–3,311 5,155–7,852
SD 1,083 3,516 0,846 4,013
Shimmer local (Db) Median 0,317 0,616 0,246 0,574
95% CI for the median 0,230–0,373 0,41–0,704 0,21–0,299 0,488–0,71
SD 0,1 0,313 0,086 0,359
Shimmer apq3 (%) Median 1,764 3,567 1,761 2,99
95% CI for the median 1,631–1,996 2,441–4,261 1,559–1,969 2,664–3,743
SD 0,271 1,786 0,537 1,957
Shimmer apq5 (%) Median 1,936 4,069 2,046 3,962
95% CI for the median 1,509–2,62 2,986–5,021 1,495–2,304 3,138–4,921
SD 0,84 2,145 0,805 2,614
Mean N/H Median 0,009 0,053 0,011 0,041
95% CI for the median 0,005–0,026 0,024–0,079 0,008–0,018 0,028–0,058
SD 0,015 0,116 0,012 0,084
Mean H/N (dB) Arithmetic mean 23,483 15,669 22,793 16,011
95% CI for the mean 22,334–24,633 14,241–17,098 21,766–23,819 14,646–17,376
SD 2,525 4,407 2,988 4,649
MPT (s) Median 18,14 10,3 16,4 9,73
95% CI for the median 15,606–19,54 9,489–14,191 14,413–18,491 8,54–11,385
SD 3,311 5,775 3,515 4,492
Abbreviation: CI, confidence interval; SD, standard deviation; MPT, maximum phonation time.
The fraction of locally unvoiced frames (Figure 11) and the Mean N/H (Figure 13) indicates the amount of noise in re-
number of voice breaks (Figure 12) are associated with the degree lation to the harmonics of the signal: the higher its value, the
of voice stability over time. Statistical analysis has led to a highly more the overall degree of voice quality is lowered. Our results
significant difference: the voice of an elderly person does not showed that the parameter value is significantly higher in those
remain stable over time, as in young individuals, and the F0 with more than 60 years of age in both sexes. This confirms the
suffers interruptions in both men and women. reliability of the previous data because the noise is due to changes
ARTICLE IN PRESS
Chiara Mezzedimi et al. Presbyphonia: a Study on 142 Patients with Praat 5
FIGURE 1. Intensity (ordinate) in the four groups (abscissa). Cat 1: FIGURE 4. Jitter local abs (ordinate) in the four groups (abscissa).
young males 21–54 years old (mean age = 35,79); Cat 2: elderly males
63–93 years old (mean age = 74,43); Cat 3: young females 20–55 years
old (mean age = 38,49); Cat4: elderly females 61–84 years old (mean
age = 72,76).
FIGURE 3. Jitter local (ordinate) in the four groups (abscissa). FIGURE 6. Jitter ppq5 (ordinate) in the four groups (abscissa).
ARTICLE IN PRESS
6 Journal of Voice, Vol. ■■, No. ■■, 2016
FIGURE 7. Shimmer local (ordinate) in the four groups (abscissa). FIGURE 10. Shimmer apq5 is the amplitude perturbation quotient—
(ordinate) in the four groups (abscissa).
FIGURE 9. Shimmer apq3 is the amplitude perturbation quotient— FIGURE 12. Number of voice breaks (ordinate) in the four groups
(ordinate) in the four groups (abscissa). (abscissa).
ARTICLE IN PRESS
Chiara Mezzedimi et al. Presbyphonia: a Study on 142 Patients with Praat 7
FIGURE 13. Mean N/H (ordinate) in the four groups (abscissa). FIGURE 15. Maximum phonation time (ordinate) in the four groups
(abscissa).
be obtained by conducting further studies to verify the effec- 7. Maryn Y, Corthals P, De Bodt M, et al. Perturbation measures of voice: a
tiveness of a rehabilitation program, via a comparison between comparative study between Multi-dimensional Voice Program and Praat.
Folia Phoniatr Logop. 2009;61:217–226.
the values found. 8. Vaz Freitas S, Melo Pestana P, Almeida V, et al. Integrating voice evaluation:
correlation between acoustic and audio-perceptual measures. J Voice.
2015;29:390, e1-7.
REFERENCES 9. Brockmann M, Storck C, Carding PN, et al. Voice loudness and gender
1. Reiss M, Reiss G. The aging voice: how to differentiate disease from normal effects on jitter and shimmer in healthy adults. J Speech Lang Hear Res.
changes. Geriatrics. 1998;53:76–79. 2008;51:1152–1160.
2. Vaca M, Mora E. Cobeta I. The aging voice: influence of respiratory 10. Woo P, Casper J, Colton R, et al. Dysphonia in the aging: physiology versus
and laryngeal changes. Otolaryngol Head Neck Surg. 2015;153:409–413. disease. Laryngoscope. 1992;102:139–144.
3. Vaca M, Cobeta I, Mora E, et al. Clinical Assessment of Glottal Insufficiency 11. Sondhi S, Khan M, Vijay R, et al. Acoustic analysis of speech under stress.
in age-related dysphonia. J Voice. 2016;doi:10.1016/j.jvoice.2015.12.010; Int J Bioinform Res Appl. 2015;11:417–432.
892–897. 12. Harnsberger JD, Shrivastav R, Brown WS Jr, et al. Speaking rate and
4. Ricci Maccarini A, Lucchini E. La valutazione soggettiva ed oggettiva della fundamental frequency as speech cues to perceived age. J Voice. 2008;22:58–
disfonia: il protocollo sifel. 2002 Available at: www.ssli.it/wp-content/ 69. [Epub 2006 Sep 11].
uploads/2013/10/protocollo-sifel.disfonia.doc. Accessed July 2016. 13. Hollien H, Shipp T. Speaking fundamental frequency and chronologic age
5. Kendall K. Presbyphonia: a review. Curr Opin Otolaryngol Head Neck Surg. in males. J Speech Hear Res. 1972;15:155–159.
2007;15:137–140. 14. Benjamin BJ. Frequency variability in the aged voice. J Gerontol.
6. Deliyski DD, Shaw HS, Evans MK. Influence of sampling rate on accuracy 1981;36:722–726.
and reliability of acoustic voice analysis. Logoped Phoniatr Vocol. 15. Mysak E. Pitch and duration characteristics of older males. J Speech Hear
2005;30:55–62. Res. 1959;2:46–54.