Journal of Voice

Vol. 16, No. 4, pp. 480–487

© 2002 The Voice Foundation

Harmonics-to-Noise Ratio: An Index of Vocal Aging

Carole T. Ferrand
Hofstra University, Hempstead, New York

Summary: Distinguishing between vocal changes that occur with normal aging
and those that are associated with disease is an important goal of research in
voice. Several acoustic measures have been used in an attempt to illuminate the
integrity of the vocal mechanism, including harmonics-to-noise ratio (HNR),
jitter, and fundamental frequency (F0). HNR is a measure that quantifies the
amount of additive noise in the voice signal; jitter reflects the periodicity of vo-
cal fold vibration. In this study, measures of HNR, jitter and F0 were used to
compare vocal function in three groups of normally speaking women: young
adults, middle-aged adults, and elderly adults. Significant differences in HNR
emerged between the elderly women and the other two groups. F0 differences
were also apparent between the elderly group and the two younger groups; there
were no significant differences in jitter between the three groups. HNR was
found to be a more sensitive index of vocal function than jitter. The significant
lowering of HNR evident in the elderly speakers may be attributable in part to
medications taken by the majority of these elderly subjects. Key Words: Har-
monics-to-noise ratio—Jitter—Fundamental frequency—Vocal function—El-
derly adults.

INTRODUCTION to change with aging, increasing with advancing age

Given the importance of distinguishing between
vocal changes that occur with normal aging and
those that are associated with disease, finding a ro-
bust acoustic measure that reliably signals vocal per-
formance in older adults remains an ongoing chal-
lenge. Fundamental frequency (F0) has been shown
Accepted for publication March 25, 2002.
Address correspondence and reprint requests to Dr. Carole
Ferrand, Department of Speech-Language-Hearing Sciences,
106 Davison Hall, Hofstra University, Hempstead, NY 11549,
USA.
Portions of this work were presented at the American
Speech-Language-Hearing Association Annual Convention,
San Francisco, CA, November 1999.
e-mail: sphctf@hofstra.edu
in elderly men, and decreasing in elderly women.1–4
By itself, however, F0 does not yield a sufficiently de-
tailed picture of vocal fold vibratory patterns to dif-
ferentiate between normal and disordered vocal func-
tion. Other acoustic measures such as jitter have
yielded inconclusive results in comparisons of
younger and older adults’ vocal function. For in-
stance, Linville and Fisher5 reported higher mean jit-
ter values in elderly women than in younger women.
However, jitter levels did not contribute to the over-
all estimation of the speaker’s age. Wilcox and Horii6
found significant jitter differences between young
and elderly men, while Ringel and Chodzko-Zajko7
indicated that jitter values were more related to phys-
iological health than to chronological age. Linville8
found considerable overlap in jitter levels among

480
 HARMONICS-TO-NOISE RATIO
groups of young, middle-aged, and elderly women.
Elderly women, on average, did not display more jit-
ter than younger women.
Another acoustic measure that may be a more sen-
sitive index of vocal function is the harmonics-to-
noise ratio (HNR). HNR quantifies the relative
amount of additive noise in the voice signal.9 Addi-
tive noise arises from turbulent airflow generated at
the glottis during phonation.10 Inadequate closure of
the vocal folds allows excessive airflow through the
glottis, giving rise to turbulence. The resulting fric-
tion noise is reflected in a higher noise level in the
spectrum.11 Noise in the signal may also result from
aperiodic vocal fold vibration. The ratio thus reflects
the dominance of harmonic (periodic) over noise
(aperiodic) levels in the voice, and is quantified in
terms of dB. Yumoto, Sasaki, and Okamura12 report-
ed that adult speakers with normal voice quality ob-
tained HNRs of 7.4 dB and above when producing
isolated vowels; other research suggests that values
between 11–13 dB are normal for adults.13,14 Per-
ceptually, HNR reflects voice quality. HNR seems to
be one of the parameters that can be used to relate
physiological aspects of voice production to a per-
ceptual impression of the voice, because the degree
of spectral noise is related to the quality of the vocal
output.11 Indeed, it has been reported that HNR is a
significant predictor of voice samples rated perceptu-
ally as rough.15–17
A limited number of studies have investigated
HNR in elderly speakers. One of the few studies fo-
cusing on spectral characteristics in the voices of old-
er adults found that HNR decreased in subjects aged
60 to 99 years old, reflecting a growing voice insta-
bility with normal aging.18 Ramig et al19 reported
that older speakers in poor physical condition
showed more spectral noise than older speakers in
good phsyical condition or than younger speakers.
However, rather than quantifying the degree of noise,
Ramig et al used spectrograms to visually determine
the level of spectral noise. Mueller1 made the as-
sumption that because hoarseness is prevalent in the
voices of elderly speakers, a lower HNR would be
predicted. However, this assumption needs to be sup-
ported and quantified in order to distinguish between
vocal changes that are within normal limits and those
that may be due to pathology.
It is important to establish a normative acoustic
database for laryngeal function in older adults. HNR
481
might be a more sensitive index of vocal function in
older speakers than is jitter, as the ratio reflects both
cycle-to-cycle variability in frequency and amplitude,
as well as additive noise generated at the glottis.20 As
such, it could be a valuable adjunct in quantifying and
differentiating between possible deterioration in voice
function in adults who are aging normally and in
those with voice problems.
The first purpose of this study, therefore, was to
compare HNRs in three age groups of women:
younger, middle-aged, and older adults. The second
purpose was to compare HNR and jitter measures in
terms of sensitivity to potential differences in vocal
function in different age groups. The third purpose
was to determine F0 differences among the three
speaker groups.
METHOD
Subjects
Forty-two adult women with normal voices served
as subjects, with 14 speakers in each of three groups:
young adults, middle-aged adults, and elderly adults.
Young adults ranged in age from 21 to 34 with an av-
erage age of 25 years. Young adults were recruited
from the student population at Hofstra University in
Hempstead, NY. Middle-aged adults ranged in age
from 40 to 63, with an average age of 50 years. The
middle-aged speakers were recruited from faculty
members at Hofstra University, as well as friends and
relatives of the examiners. Older adults ranged in age
from 70 to 90, with an average age of 77 years. Old-
er speakers were recruited from senior centers and
retirement homes in the Hempstead, NY area. All
older adults lived independently. No subject had re-
ceived any voice therapy or professional voice train-
ing. All subjects denied a history of neurologic dis-
ease or dementia. All the elderly subjects passed a
hearing screening at 250, 500, 2000, and 4000 Hz at
25 dB HL. The young and middle-aged adults all re-
ported that they had normal hearing. All speakers
were free of colds or other respiratory illness on the
day of testing. All subjects were naive to the purpose
of the study, and no subject had participated in a
voice study before. Two of the young adults, 5 of the
middle-aged adults, and 12 of the elderly adults were
taking some form of medication on a regular basis,
incuding estrogen, antidepressants, blood pressure
Journal of Voice, Vol. 16, No. 4, 2002
482 CAROLE T. FERRAND
medications, and cholesterol-lowering medications.
All subjects reported that their voices sounded typi-
cal for them on the day of testing.
Procedures
Recording sessions were held in a sound-treated
laboratory at Hofstra University for the young speak-
ers; in a quiet schoolroom in a school that was closed
for the summer for middle-aged speakers, and in a
quiet room at a senior center housed in a church
building for the elderly speakers. Subjects vocalized
the vowel /!/ five times each, using a comfortable
pitch and loudness level, with each prolongation last-
ing 3 seconds, and a 2 to 3 second pause between
each trial. Voice samples for all subjects were tape
recorded by two graduate research assistants. Each
individual session lasted no longer than 20 minutes.
Instrumentation and acoustic analysis
A Sony Portable MiniDisc Recorder (model MZ-
B3) was used to collect all voice samples, with Max-
ell 74 recordable minidiscs. The recorder had a built-
in microphone, and was held by the examiner at a
distance of 6 inches from each subject’s mouth. The
recorder uses a sampling frequency of 44.1 kHz, and
has a frequency response of 20 to 20,000 Hz. Wow
and flutter are below measurable limits. The record-
able minidiscs use magneto-optical technology. The
laser inside the recorder applies heat to the minidisc,
demagnetizing the magnetic layer of the minidisc.
The recorder then applies a magnetic field to the lay-
er. This magnetic field corresponds exactly to the au-
dio signals generated by the connected source (Sony
Portable MiniDisc Recorder Manual, 1995). The
tape recorded samples were analyzed using a Kay El-
emetrics (Lincoln Park, NJ) Computerized Speech
Lab (model 4300). The samples were input directly
from the tape recorder to channel 2 of the CSL. The
voice analysis function of the CSL was used to obtain
HNR, as well as jitter and average F0. The sampling
rate on the CSL was set to 51.2 kHz, and the time
frame was set to 2 seconds. The first and last portions
of each vocalization were discarded in order to avoid
any perturbations due to onset or offset of vibration.
The CSL calculates the HNR based on the ratio of
harmonically related energy to noise in the speech
signal. The algorithm follows Yumoto’s method, ex-
cept that the signal is not preemphasized (Kay Ele-
Journal of Voice, Vol. 16, No. 4, 2002
metrics CSL manual). The impulse command was set
prior to the voicing analysis command. This com-
mand computes and marks the divisions between
each voicing impulse in the waveform, placing divid-
ing marks at the locations immediately following pos-
itive-going zero-crossings that precede the first posi-
tive-going amplitude peaks of the voicing impulses.
The process thus separates the voiced signal into its
periodic components. This command is used in prepa-
ration for subsequent pitch-synchronous analysis of
the data (Kay Elemetrics CSL manual).
Reliability
Each taped voice sample was analyzed twice by
each research assistant, in order to determine intra-
judge and interjudge reliability.
RESULTS
For more than 98% of the samples, neither HNR
nor jitter differed by more than 0.02 dB and 0.02%,
respectively.
Values for HNR, jitter, and F0 were averaged over
the five vowel prolongations for each individual. In-
dividual means were then averaged for each of the
three groups of speakers. Tables 1, 2, and 3 display
individual and group means and standard deviations
for HNR, jitter, and F0.
In terms of HNR, young and middle-aged subjects
obtained mean values of 7.82 dB and 7.86 dB, re-
spectively. Elderly women obtained an average of
5.54 dB. Intrasubject standard deviations (SD) for
the young women ranged from 0.71 to 5.63 dB, with
a mean of 2.27 dB. Individual SDs for the middle-
aged subjects were similar, ranging from 0.36 to 4.30
dB, with an average of 2.42 dB. For the elderly
speakers, intrasubject SDs ranged from 0.81 to 3.73
dB, with a mean of 1.95 dB. These small intrasubject
SDs demonstrate a good deal of intertrial consisten-
cy for all subject groups. Intersubject SDs were also
consistent among the three groups, with a value of
2.64 dB for the young group, 3.14 dB for the middle-
aged group, and 2.91 dB for the elderly group.
Table 2 indicates that mean jitter values were sim-
ilar for the three speaker groups: 0.69% for the young
adults, 0.57% for the middle-aged subjects, and 0.66%
for the elderly group. Intrasubject SDs for the five tri-
als ranged from 0.03 to 0.36% for the young group,
 HARMONICS-TO-NOISE RATIO 483

TABLE 1. Individual and Group Means and Standard TABLE 2. Individual and Group Means and Standard
Deviations (SD) for Young, Middle-Aged, and Elderly Deviations (SD) for Young, Middle-Aged, and Elderly
Women for HNR (in dB) Women for Jitter (%)
Young Middle-Aged Elderly Young Middle-Aged Elderly

Subject Mean SD Mean SD Mean SD Subject Mean SD Mean SD Mean SD

1. 5.12 5.62 5.84 1.99 4.98 3.73 1. 0.91 0.31 0.65 0.14 1.06 0.21
2. 7.42 2.42 1.51 2.13 7.03 2.72 2. 0.45 0.15 0.88 0.18 0.33 0.07
3. 8.80 0.92 5.92 2.94 1.71 2.03 3. 0.48 0.19 1.07 0.18 0.56 0.03
4. 8.61 0.79 7.97 2.62 6.80 2.24 4. 0.74 0.10 1.01 0.28 0.28 0.06
5. 9.68 1.53 8.72 1.05 1.21 2.66 5. 0.39 0.16 0.37 0.09 1.05 0.33
6. 7.04 2.39 10.05 4.30 7.09 2.37 6. 0.70 0.11 0.34 0.04 0.33 0.07
7. 6.56 0.83 9.07 0.36 10.59 0.81 7. 0.75 0.24 0.67 0.10 0.26 0.06
8. 8.32 2.26 8.09 3.53 4.40 2.44 8. 0.45 0.05 0.60 0.10 1.45 0.36
9. 9.50 0.71 9.39 2.52 9.56 0.89 9. 0.93 0.36 0.26 0.03 0.26 0.23
10. 2.85 2.71 7.85 3.08 5.71 1.89 10. 1.36 0.25 0.29 0.05 0.37 0.05
11. 3.62 5.63 13.92 1.35 6.49 2.50 11. 1.00 0.29 0.31 0.05 1.04 0.70
12. 8.89 1.82 2.54 0.93 7.60 1.11 12. 0.86 0.24 0.53 0.03 0.53 0.05
13. 12.29 1.70 9.80 3.34 2.25 1.72 13. 0.31 0.10 0.35 0.10 0.87 0.24
14. 10.83 3.40 9.31 3.71 2.20 1.20 14. 0.34 0.03 0.60 0.28 0.78 0.51
Mean 7.82 2.27 7.86 2.42 5.54 1.95 Mean 0.69 0.18 0.57 0.12 0.66 0.21
SD 2.64 1.47 3.14 1.17 2.91 0.84 SD 0.30 0.10 0.27 0.08 0.39 0.20

with an average SD of 0.18%. The middle-aged speak-
ers’ intrasubject SDs ranged from 0.03% to 0.28%,
with an average of 0.12%; those for the elderly sub-
jects ranged from 0.03% to 0.70%, with an average
of 0.21%. Intersubject SDs were 0.30% for the
young group, 0.27% for the middle-aged group, and
0.39% for the elderly group.
As can be seen in Table 3, the young adults ob-
tained a mean of 209.68 Hz for F0, the middle-aged
adults a mean of 204.49 Hz, and the elderly adults a
mean of 175.23 Hz. Intrasubject SDs for the young
speakers ranged from 0.51 to 17.92 Hz, with a mean
of 3.88 Hz. Values for the middle-aged speakers
ranged from 0.30 to 8.97 Hz, with an average SD of
2.60 Hz. Elderly speakers’ SDs ranged from 0.62 to
23.76 Hz, with a mean of 6.01 Hz. Similar to HNR
and jitter, these levels of intrasubject variability are
consistent among the three groups. Intersubject vari-
abiity for F0 was somewhat higher for the middle-
aged group (39.44 Hz) than for the young and elder-
ly groups (23.24 and 27.04 Hz, respectively).
Three separate one-way analyses of variance
(ANOVA) were performed for HNR, jitter, and F0,
with age as the grouping variable in each case. A sig-
nificant difference was found for HNR (F = 3.804,
p= 0.0307). Post hoc testing using Fisher’s least sig-
nificant differences (LSD) demonstrated significant
differences between the young and elderly values (p
= 0.0248), and between the middle-aged and elderly
values (p = 0.0204). There was no significant differ-
ence between the young and middle-aged groups for
HNR (p = 0.9356). Significant differences were also
found for F0 (F = 3.614, p = 0.0361). Post hoc test-
ing revealed significant differences between the
young and elderly values (p = 0.0163) and between
the middle-aged and elderly values (p = 0.0461). No
significant differences emerged for the young and
middle-aged groups for F0 (p = 0.6610). In terms of

Journal of Voice, Vol. 16, No. 4, 2002

484 CAROLE T. FERRAND

TABLE 3. Individual and Group Means and Standard proximately 17 to 19 dB for their male and female
Deviations (SD) for Young, Middle-Aged, and Elderly speakers. It is not clear why this large discrepancy
Women for F0 (in Hz) exists in the values for HNR reported in the literature,
Young Middle-Aged Elderly but possible candidates are different algorithms used
to calculate the HNR, different recording procedures,
Subject Mean SD Mean SD Mean SD differences in instrumentation, and differences in
1. 237.93 2.32 146.06 3.90 190.19 14.64
subject samples.
Results of the present study suggest that HNR ap-
2. 195.78 17.92 172.30 3.32 129.13 0.62
pears to be stable in the young and middle adult
3. 168.57 0.87 268.66 1.71 184.31 10.94
years, with HNR decreasing during the elderly years.
4. 213.80 1.98 248.94 2.60 142.98 1.86 The well-documented age-related decrements in la-
5. 218.25 2.09 185.77 2.29 156.84 4.94 ryngeal function such as ossification of cartilage, de-
6. 208.77 4.26 250.77 3.29 185.07 6.24 generation of muscle and connective tissue as well as
7. 166.93 1.28 185.55 0.30 210.56 1.12 respiratory changes (decreased elastic recoil of lung
8. 218.99 3.02 171.35 0.75 230.57 3.86 tissue and reductions in vital capacity24) may con-
9. 194.73 0.51 206.90 0.78 156.91 1.04
tribute to older women’s lower HNRs by contributing
to instability in vocal fold vibration and thus interfer-
10. 214.73 8.80 172.24 0.45 150.31 1.73
ing with the periodicity of vibration. Another factor
11. 202.86 8.08 235.78 2.58 165.61 4.37
that may play a role in the increasing predominance
12. 213.04 0.62 156.60 8.97 180.03 2.29 of noise over harmonic components in the voice, may
13. 248.27 1.67 232.52 3.11 182.82 6.78 be the medications taken by many elderly persons.
14. 232.84 0.95 229.41 2.33 187.84 23.76 Twelve of the 14 elderly subjects in the present study
Mean 209.68 3.88 204.49 2.60 175.23 6.01 were taking some kind of medication on a regular ba-
SD 23.24 4.81 39.44 2.17 27.04 6.50 sis, even though all the subjects considered them-
selves to be in good health, with no respiratory,
phonatory, or neurologic problems. All the older
adults would be considered to be “successfully” ag-
jitter, no significant differences were demonstrated ing. Five of the 14 subjects in the middle-aged group
between any of the groups (F = 0.630, p = 0.5378).
were taking some form of medication, while only
two of the 14 subjects in the young group were tak-
DISCUSSION ing medications. Table 4 shows the medications tak-
The major focus of this study was to determine en by subjects in each group.
whether HNR values changed with age. The second Virtually all medications including antihistamines,
purpose was to determine whether HNR values and decongestants, diuretics, hypertensives, and so on,
jitter values yielded similar results in three groups of have some laryngeal effects, the most common of
women; the third purpose was to compare F0 values which are excessive vocal tract dryness and thick-
between the three groups. ened mucosal secretions.25,26 Sataloff et al27 noted
The HNR values in the present study are compara- that normal mucosal secretions and free movement
ble to some values reported in the literature, but low- of the vibratory margins are critical for normal
er than others. Bertino et al21 and Yumoto, Gould, phonation. If “vocal tract lubrication is impaired by
and Baer22 reported values of 7.23, and 7.3 dB, re- dehydration, or by shifting the normal balance of
spectively, for adult speakers. These are very similar serous and mucinous secretions, alterations in phona-
to the values for the young (7.82 dB) and middle- tion occur” (p. 223). Other medications can interfere
aged (7.91 dB) speakers in the present study. On the directly with voluntary movement which can affect
other hand, Awan and Frenkel9 reported values of ap- laryngeal motor control. For example, antidepres-
proximately 15 dB for the adult speakers in their sants such as Zoloft and Paxil can cause not only dry-
study, and Horii and Fuller23 reported values of ap- ness of the mouth, but tremor. Antianxiety drugs such

Journal of Voice, Vol. 16, No. 4, 2002

 HARMONICS-TO-NOISE RATIO 485

TABLE 4. Medications Taken by HNR values between the elderly group and the other
Young, Middle-Aged, and Elderly two groups supports the hypothesis that HNR is more
Subjects sensitive to subtle differences in vocal function than
Young is jitter. A similar finding was reported by Awan and
Frenkel.9 These researchers calclulated the relation-
Claritin Antiallergan ship between jitter and HNR, and found more vari-
Rhinocort Allergic or perennial rhinitis ability in jitter values at different frequencies than
was observed with HNR levels. The authors found
Middle-Aged occasions in which small or no changes in jitter were
Claritin Antiallergan present in signals that showed substantial changes in
HNR. This led them to suggest that although jitter
Toprol Hypertension
and HNR are related, HNR may more accurately re-
Procardia Antianginal
flect the degree of additive noise in a signal.
Estraderm Estrogen (estradiol) Further, results from several types of investigations
Provera Progesterone highlight the inability of the jitter measure to distin-
Prempro Estrogen/progesterone guish between vocal function in different groups.
Altace Hypertension Wilcox and Horii6 found that older subjects obtained
Celebrex Anti-inflammatory higher levels of jitter than younger subjects (0.73%
and 0.55%, respectively). However, both these values
Elderly are well within normal limits. Brown, Morris, and
Estrace Estrogen Michel28 analyzed young and elderly womens’ pro-
Procardia Antianginal
longations of the vowel /a/. Jitter values were not sig-
nificantly different for the groups at their normal
Ultram Central acting analgesic
pitch and loudness levels (0.78% and 0.75%). Biever
Zoloft Antidepressant
and Bless29 found no significant jitter differences be-
Synthroid Hypothyroidism tween young and elderly women sustaining the vow-
Prilosec GERD el /i/. Stroboscopically, however, there were signifi-
Toprol Hypertension cant differences in vibratory patterns, with the
Celebrex Anti-inflammatory elderly group showing increased aperiodicity, incom-
Zocor Cholesterol plete midmembranous glottal closure, mucosal wave
alterations, and reduced amplitude of vibration. The
fact that the acoustic measure of jitter was not relat-
ed to the observed aperiodicity suggests that jitter
as Xanax and Valium can have side effects including may not accurately reflect vocal fold vibratory pat-
weakness, ataxia, and decreased motor perfor- terns. Omori et al30 studied subjects with normal and
mance.27 In addition, drug interactions can cause pathologic voices. Subjects sustained the /a/ vowel,
side effects that can result in vocal changes, such as and the authors calculated degree of jitter as well as
thinning and dehydration of the vocal folds.28 Thus, subharmonics in the voice. Several voices which
given that the majority of elderly women in the pres- were rated as perceptually rough and which demon-
ent study were taking medications, with many taking strated the presence of subharmonics, exhibited jitter
multiple medications, it seems plausible that laryn- levels within normal limits. Similarly, Zyski et al31
geal function was affected, possibly contributing to reported that 40% of pathological voices in their
the observed lower HNR values. study demonstrated normal jitter values.
The jitter values found in the present study (0.69% Consistent with many other studies,2–4 the elderly
young, 0.57% middle-aged, 0.66% elderly) are con- women in the present study obtained a significantly
sistent with those in the literature for normally speak- lower average F0 than the middle-aged and young
ing adults. The lack of significant differences in jitter women. Brown, Morris, and Murry24 studied /a/ pro-
values as opposed to the significant differences in longations of young, middle-aged, and elderly men

Journal of Voice, Vol. 16, No. 4, 2002

486 CAROLE T. FERRAND
and women. The older women demonstrated signifi-
cantly lower F0 levels than the young and middle-
aged speakers, with means of 209 Hz (young), 198
Hz (middle-aged), and 189 Hz (elderly). These val-
ues are consistent with those obtained in the present
study: 209.68 Hz (young), 204.49 Hz (middle-aged),
and 175.23 Hz (elderly). This decrease in average F0
has often been attributed to vocal fold edema resulting
from postmenopausal changes. Biever and Bless29
noted that edema not only adds mass to the vocal
folds, but also changes their depth and contour, and
these changes correlate with reports of pitch lowering
in postmenopausal women. Sataloff et al27 indicated
that as women reach menopause, the loss of estrogen
causes substantial changes in the mucous membranes
of the vocal tract as well as in the muscles. According
to Boone,26 the most noticeable change in the female
voice after menopause is the lowering of F0 as a result
of increased systemic androgen. The ovary secretes
minimal or no estrogen, but continues to secrete an-
drogen,27 which causes unsteadiness of the voice,
rapid changes of voice quality, and lowering of the
F0.27 These hormonal changes may contribute not on-
ly to lowered F0, but also to the less efficient func-
tioning of the laryngeal system in elderly women,
with the resultant decrease in HNR values.
Subjects in the present study were required to
phonate the vowel /a/ at a comfortable pitch and
loudness level. This procedure is consistent with
much of the research investigating acoustic aspects
of voice production. Brown et al24 operationally de-
fined the term comfortable as “. . . the level of effort
that is expended by a speaker as reflected by such vo-
cal parameters as F0 and/or intensity” (p. 299). The
implication is that there is a typical level of vocal ef-
fort which provides the speaker with an internal ref-
erence in producing repeated samples.32 Brown et
al33 quantified the amount of intrasubject variability
in F0 and intensity within and across experimental
sessions in young, middle-aged, and elderly men and
women. They determined each speaker’s variability
from his or her own mean for production of /a/, and
found minimal intrasubject variability for all speaker
groups. The authors concluded that the variability in
F0 and intensity within and across days was small,
and may represent tolerable limits of variation, when
subjects are required to phonate at a comfortable lev-
el.24 Similarly, Sussman and Sapienza34 required
Journal of Voice, Vol. 16, No. 4, 2002
their subjects to produce vowels at a comfortable lev-
el. Subjects were not instructed to match a particular
intensity level because the researchers wanted to
avoid “any unnatural laryngeal variations” (p. 149)
that may have affected their typical F0 control. The
authors reported that their subjects demonstrated
consistent F0 and jitter levels both within and across
trials. On the other hand, Huang et al35 found a rela-
tionsip between vocal effort and jitter, such that jitter
levels decreased with increases in vocal effort. Based
on this finding they suggested that it is important to
control vocal effort for research or clinical purposes.
However, their tiny sample size (n = 3) precludes
drawing any strong conclusions from their data. In-
terestingly, Brown et al28 found that both young and
elderly speakers demonstrated signficantly more jit-
ter when the vowel was produced either more loudly
or more softly when compared to the subject’s com-
fortable loudness level.
The different acoustic environments in which data
were collected for the different groups is a limitation
of the present study. However, in light of the fact that
jitter values for the three groups were similar, it
seems unlikely that this factor played a major role in
yielding different results for F0 and HNR for the
young, middle-aged, and elderly subjects.
In sum, it appears that the HNR measure is better
able to discriminate between the voices of elderly
and younger adults than is jitter. Future research
should expand on these findings by adding a percep-
tual analysis of the voice quality of the three groups,
as well as correlating voice quality, HNR values, and
number and types of medications being taken by in-
dividual subjects.
Acknowledgments: This work was supported in part by a
Hofstra College of Liberal Arts and Sciences Faculty Re-
search and Development Grant. The author wishes to
thank Jillian Resch and Michael Sayers for their outstand-
ing assistance in collecting the data.
REFERENCES
1. Mueller PB. The aging voice. Sem Sp Lang. 1997;18:159–168.
2. Sataloff RT, Caputo Rosen D, Hawkshaw M, Spiegel J. The
three ages of voice: the aging adult voice. J Voice. 1997;11:
156–160.
3. Russell A, Penny L, Pemberton C. Speaking fundamental fre-
quency changes over time in women: a longitudinal study. J
Speech Hear Res. 1995;38:101–109.
 HARMONICS-TO-NOISE RATIO
4. Linville SE. The sound of senescence. J Voice. 1996;10:
190–200.
5. Linville SE, Fisher H. Acoustic characteristics of perceived
versus actual vocal age in controlled phonation by adult fe-
males. J Acoust Soc Am. 1985;78:40–48.
6. Wilcox KA, Horii Y. Age and changes in vocal jitter. J
Gerontol. 1980;35:194–198.
7. Ringel RL, Chodzko-Zajko WJ. Vocal indices of biological
age. J Voice. 1987;1:31–37.
8. Linville SE. Maximum phonational frequency range capa-
bilities of women’s voices with advancing age. Folia Phoni-
atr. 1987;39:297–301.
9. Awan SN, Frenkel ML. Improvements in estimating the har-
monics-to-noise ratio of the voice. J Voice. 1994;8:255–262.
10. Hillenbrand J. A methodological study of perturbed and ad-
ditive noise in synthetically generated voice signals. J
Speech Hear Res. 1987;30:448–461.
11. de Krom G. A cepstrum-based technique for determining a
harmonics-to-noise ratio in speech signals. J Speech Hear
Res. 1993;36:254–266.
12. Yumoto E, Sasaki Y, Okamura H. Harmonics-to-noise ratio
and psychophysical measurement of the degree of hoarse-
ness. J Speech Hear Res. 1984;27:2–6.
13. Murry T, Brown WS Jr, Rothman H. Judgments of vocal
quality and preference: acoustic interpretations. J Voice.
1987;1:252–257.
14. Till JA, Jafar M, Crumley R, Law-Till CB. Effects of initial
consonant, pneumotachographic mask, and oral pressure
tube on vocal perturbation, harmonics-to-noise, and intensi-
ty measurements. J Voice. 1992;6:217–223.
15. Martin D, Fitch J, Wolfe V. Pathologic voice type and the
acoustic predictions of severity. J Speech Hear Res. 1995;
38:765–771.
16. Eskenazi L, Childers DG, Hicks DM. Acoustic correlates of
vocal quality. J Speech Hear Res. 1990;33:298–230.
17. de Krom G. Some spectral correlates of pathological
breathy and rough voice quality for different types of vowel
fragments. J Speech Hear Res. 1995;38:794–811.
18. Decoster W, Debruyne F. The ageing voice: changes in fun-
damental frequency waveform stability and spectrum. Acta
Otorhinolaryngol Belg. 1997;51:105–112.
19. Ramig LA, Titze IR, Scherer C, Ringel SP. Acoustic analy-
sis of voices of patients with neurologic disease: rationale
and preliminary data. Ann Otol Rhinol Laryngol. 1988;97:
164–172.
487
20. Walton JH, Orlikoff RF. Speaker race identification from
acoustic cues in the vocal signal. J Speech Hear Res. 1994;
37:738–745.
21. Bertino G, Bellomo A, Miani C, Ferrero F, Staffieri A.
Spectrographic differences between tracheal-esophageal
and esophageal voice. Folia Phoniatr Logo. 1996;48:
255–261.
22. Yumoto E, Gould WJ, Baer T. Harmonics-to-noise ratio as
an index of the degree of hoarseness. J Acoust Soc Am.
1982;71:1544–1550.
23. Horii Y, Fuller BF. Selected acoustic characteristics of voic-
es before intubation and after extubation. J Speech Hear
Res. 1990;33:505–510.
24. Brown WS Jr, Morris RJ, Murry T. Comfortable effort level
revisited. J Voice. 1996;10:299–305.
25. Sataloff RT. Medications and their effect on voice. ASHA
Special Interest Divisions Voice Voice Disord Newsletter.
1995;5:5–11.
26. Boone DR. The three ages of voice: the singing/acting voice
in the mature adult. J Voice. 1997;11:161–164.
27. Sataloff RT, Hawkshaw M, Caputo Rosen D. Medications:
effects and side effects in professional voice users. In: Sat-
aloff RT, ed. Vocal Health and Pedagogy. San Diego, Calif:
Singular Publishing Group, Inc.; 1998: 223–235.
28. Brown WS Jr, Morris RJ, Michel JF. Vocal jitter in young
adult and aged female voices. J Voice. 1989;3:113–119.
29. Biever DM, Bless DM. Vibratory characteristics of the vo-
cal folds in young adult and geriatric women. J Voice.
1989;3:120–131.
30. Omori K, Kojima H, Kakani R, Slavit DH, Blaugrund SM.
Acoustic characteristics of rough voice: subharmonics. J
Voice. 1997;11:40–47.
31. Zyski B, Bull G, McDonald W, Johns M. Perturbation
analysis of normal and pathologic larynges. Folia Phoniatr.
1984;36:190–198.
32. Murry T, Brown WS Jr, Morris RJ. Patterns of fundamental
frequency for three types of voice samples. J Voice. 1995;9:
282–289.
33. Brown WS Jr, Morris RJ, DeGroot T, Murry T. Reliability
of single sample experimental designs: comfortable effort
level. J Voice. 1998;12:453–459.
34. Sussman JE, Sapienza C. Articulatory, developmental, and
gender effects on measures of fundamental frequency and
jitter. J Voice. 1994;8:145–156.
35. Huang DZ, Minifie FD, Kasuya H, Lin SX. Measures of vo-
cal function during changes in vocal effort level. J Voice.
1995;9:429–438.
Journal of Voice, Vol. 16, No. 4, 2002