The Dysphonia Severity Index:

An Objective Measure of Vocal

Quality Based on a Multiparameter
Approach

Floris L. Wuyts
Marc S. De Bodt The vocal quality of a patient is modeled by means of a Dysphonia Severity Index
University of Antwerp (DSI), which is designed to establish an objective and quantitative correlate of the
Antwerp, Belgium perceived vocal quality. The DSI is based on the weighted combination of the
following selected set of voice measurements: highest frequency (F0-High in Hz),
Geert Molenberghs lowest intensity (I-Low in dB), maximum phonation time (MPT in s), and jitter (%).
Limburgs Universitair Centrum The DSI is derived from a multivariate analysis of 387 subjects with the goal of
Diepenbeek, Belgium describing, purely based on objective measures, the perceived voice quality. It is
constructed as DSI = 0.13 × MPT + 0.0053 × F0-High – 0.26 × I-Low – 1.18 ×
Marc Remacle Jitter (%) + 12.4. The DSI for perceptually normal voices equals +5 and for
University of Louvain severely dysphonic voices –5. The more negative the patient’s index, the worse is
Yvoir, Belgium his or her vocal quality. As such, the DSI is especially useful to evaluate therapeu-
tic evolution of dysphonic patients. Additionally, there is a high correlation
Louis Heylen between the DSI and the Voice Handicap Index score.
University of Antwerp
KEY WORDS: voice quality, voice assessment, acoustic, voice range profile,
Antwerp, Belgium
index
Benoite Millet
Brussels, Belgium

V
Kristiane Van Lierde ocal performance has increasingly gained interest in our society,
University of Gent which is evolving into a service-oriented community. This grow-
Gent, Belgium ing interest has consequently induced a lot of multidisciplinary
research concerning voice assessment and therapy with a comprehen-
Jan Raes
sive battery of tests focusing on qualitative and quantitative aspects of
University of Brussels
vocal performance.
Jette, Belgium
The medical diagnosis of vocal fold pathology is mainly based on an
Paul H. Van de Heyning endoscopic exam of the vocal folds and upper airway tract. Voice dys-
University of Antwerp function on the other hand is assessed by perceptual judgment and ob-
Antwerp, Belgium jective measures, such as acoustic and aerodynamic characteristics.
However, perceptual evaluation is one of the most controversial topics
in voice research. Review of literature reveals a wide variety of rating
scales (Gelfer, 1988; Hammarberg, 1992; Hirano, 1981; Laver, 1980;
Wendler, Rauhut, & Krüger, 1986; Wilson, 1987; Wirz & Mackenzie Beck,
1995) and reliability data fluctuating from study to study (Bassich &
Ludlow, 1986; Blaustein & Bar, 1983; Kreiman, Gerratt, & Berke, 1994;
Kreiman, Gerratt, Kempster, Erman, & Berke, 1993). So far, there is no
internationally accepted perceptual judgment protocol, but the GRBAS
scale (Grade of hoarseness, R for roughness, B for breathiness, A for
astheny, and S for strain) proposed by the Japan Society of Logopaedics

796 Journal
Journal of of Speech,
Speech, Language,
Language, andand Hearing
Hearing Research• •Vol.
Research Vol.
4343• •796–809
796–809• •June
June 2000 • ©American Speech-Language-Hearing Association
2000
1092-4388/00/4303-0796

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 and Phoniatrics (Hirano, 1981) seems to be the most
widely used (De Bodt, 1997; Dejonckere & Wieneke,
1992). This is probably because it is a simple and short
scale, appropriate for daily use by both laryngologists
and speech pathologists. Still, the GRBAS scale has a
number of drawbacks in that there are several sources
of variability, such as the professional background and
experience of the judges (De Bodt, Wuyts, Van de
Heyning, & Croux, 1997), examiner bias due to knowl-
edge of the patient’s history (Wuyts, De Bodt, & Van de
Heyning, 1998), and the type of sample—such as run-
ning speech versus sustained vowel (Revis, Giovanni,
Wuyts, & Triglia, 1997, 1999). An attempt to refine the
grading (0 = normal, 1 = slight, 2 = moderate, 3 = se-
vere) with a continuous scale failed (Wuyts, De Bodt, &
Van de Heyning, 1999). In that study the perceptual
evaluation of 14 pathological voices was performed by
29 listeners using the GRBAS scale. In order to evalu-
ate scale-effects on the judgments, two versions of the
scale were presented: the original 4-point scale and a
visual analog scale. Each listener used the same voice
samples for both versions of the scale, with an interval
of 2 weeks between judgments. The amount of agree-
ment was found to be higher with the original 4-point
scale than with the visual analog version for all scale
items G, R, B, A, and S. Although a visual analog scale
seems to enable a finer judgment of voice quality, this
study shows that with the increased freedom of judg-
ment, the interrater agreement decreases considerably.
Next to the perceptual evaluation, a number of du-
ration and aerodynamic measures (such as maximum
phonation time, phonation quotient, airflow, subglottic
pressure, etc.) have been used for the characterization
of voice quality (Hirano, Hibi, Terasawa, & Fujiu, 1986).
The introduction of computer-based systems have addi-
tionally facilitated the use of acoustical analysis of voice
samples (Baken, 1987; Rabinov, Kreiman, Gerratt, &
Bielamowics, 1995).
The relation between perceptual judgment and
acoustic measurement has been investigated by several
authors. Most studies have investigated the quantita-
tive correlation between isolated acoustic variables, such
as jitter, shimmer, or harmonics/noise ratio, with the
perceptual judgment (Askenfeld & Hammarberg, 1986;
Crevier-Buchman, 1998; Dejonckere & Lebacq, 1996;
Dejonckere, Remacle, et al., 1996; Hammarberg et al.,
1980; Hillenbrand, Cleveland, & Erickson, 1994; Kreiman
et al., 1994; Rabinov et al., 1995). Other authors have
applied multiple linear regression analysis to investi-
gate the relationship between combinations of acoustic
variables and the perceptual ratings (Eskenazi, Childers,
& Hicks, 1990; Wolfe, Fitch, & Cornell, 1995). Wolfe et
al. (1995) indicated that none of the acoustic variables
were strongly correlated with the dysphonia ratings, and
neither was a combination of variables successful for
Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
the prediction of dysphonia. Eskenazi et al. (1990)
showed, however, a fair prediction of the degree of dys-
phonia based on pitch amplitude and harmonics-to-noise
ratio. Wuyts, De Bodt, Bruckers, and Molenberghs
(1996) showed that the best Spearman’s rank correla-
tion between acoustic variables from the Multi Dimen-
sional Voice Program (Kay Elemetrics) and the GRBAS
score in a large group of subjects (N = 494) did not ex-
ceed 0.53 for any of the GRBAS scale items. If we con-
sider the correlation between, for instance, jitter (%)
and the GRBAS scale items, then the following correla-
tion coefficients were reported: 0.46, 0.37, 0.31, 0.17, and
0.23 respectively. Although all these correlations are sig-
nificant (p < 0.001), it is not possible to draw clinically
useful conclusions about voice quality based on single
variables that exhibit such low correlations with the per-
ceived voice quality.
Voice research so far has not led to the construction
of or a consensus about a sensitive measure that unam-
biguously quantifies vocal quality. As a result of the
vague relationship among pathology, vocal quality, and
measurements, the clinician is frequently confronted
with contradictory data when assessing an individual
patient’s voice.
Significant differences between group means of sev-
eral variables obtained from normal subjects and dys-
phonic patients can only be elicited when large groups
of patients are considered (Andrews, 1995; Hirano, 1981,
1990; Wuyts et al., 1996). Indeed, the standard error of
the mean decreases with an increasing number of sub-
jects (standard error = standard deviation/√N). As such,
with large numbers of cases, the statistics become more
reliable and differences between samples become sig-
nificant. Yet, the standard deviation that indicates the
typical variation of one measurement (case) around the
mean is usually so large that it becomes difficult to de-
clare a specific case as abnormal. The range of normalcy
is often defined by the prediction interval (PI) [average
± 1.96 standard deviation]. For the maximum phona-
tion time for male subjects, for example, the PI extends
from 6.7 s to 37 s (Wuyts et al., 1996). Therefore, one
can only conclude that an individual patient performs
below normal when his or her MPT is lower than 6.7 s,
which is very low indeed. This large variability of most
characteristics is probably one of the major reasons why
most measurements fail to quantify overall voice qual-
ity (Van de Heyning et al., 1996).
The drawback of such a univariate approach can be
circumvented by applying multivariate statistics. The
essence of this method lies in the possibility of describ-
ing the behavior of a given dependent (outcome) vari-
able based on a combination of several independent vari-
ables. As such, more information is used at the same
time. The dependent variable can be either continuous
Wuyts et al.: Dysphonia Severity Index 797
 (e.g., 0–100) or categorical (e.g., 0: normal and 1: patho-
logical or severity ratings 0–3).
The use of a multivariate approach as a measure of
a complex phenomenon is not new in health care. The
Body Mass Index [weight (kg)/height2(m)], for instance,
is a good example of the power of combining variables.
If a man weights 90 kg, one can state that this is a nor-
mal weight only if information about his height is also
given. Indeed, if this man is only 1.6 m tall, he can be
considered obese. This is expressed by a BMI = 35 kg/m2
(BMI > 30 is obese) that mathematically combines
weight and height into one outcome measure. Weighted
sums are commonly used in multivariate statistical
methods such as logistic regression or discriminant
analysis. Additionally, stepwise analysis methods, which
have become available with the new statistical computer
software packages, enable selection of an optimal com-
bination of independent variables. Therefore, one can
start the analysis with, for example, 20 independent
variables and end up with only 3 variables if a specific
combination of these 3 variables is sufficient to describe
the dependent variable.
Multivariate techniques prove to be useful in voice
research as shown by the voice range profile index for
children (Heylen et al., 1998). The voice range profile
(VRP) is a two-dimensional representation of frequency
and intensity that cannot be trivially expressed by a
single value. This limits its clinical use because an in-
terpretation of the overall shape is necessary to assess
a patient’s performance. Thus, several investigators have
restricted their interpretation to selected VRP features,
such as the dynamic intensity or frequency ranges. By
including several salient VRP characteristics in a con-
cise measure, defined as the VRP index, details regard-
ing the F0 intensity interaction can be addressed as well.
By using discriminant analysis, a weighted sum (defined
as the VRPIc) was constructed based on the stepwise
selection of only a few salient VRP features. The VRPIc
can be used to discriminate children with vocal pathol-
ogy from those with normal voices. Because the group
means of the VRPIc for healthy and dysphonic children
are scaled to +10 and –10 respectively, the VRPIc en-
ables the clinician to rate a child’s vocal performance
with reference to healthy and dysphonic children in gen-
eral. Although the index is an appropriate tool in voice
assessment, its use is limited to children and voice-
range-profile measurements only.
Few other authors have successfully demonstrated
the use of multivariate statistics to construct an index
that enables the discrimination of normal from dyspho-
nic voices. Piccirillo, Painter, Fuller, Haiduk, and
Fredrickson (1998) constructed an index based on voice-
range profile and aerodynamic measurements. Yet, they
included only those variables that are univariately
798 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000
Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
significantly different in dysphonic as opposed to nor-
mal populations. However, stepwise multivariate tech-
niques look for optimal combinations of variables regard-
less of their univariate significance level. Finally, the
authors state that their index is preferentially a classi-
fication tool rather than a predictive or treatment effec-
tiveness index.
Recent work by Callan, Kent, Roy, and Tasko (1999)
introduces the concept of self-organizing maps for the
classification of normal and disordered voices. They show
that this method is superior to stepwise discriminant
analysis. The self-organizing map was trained using
spasmodic dysphonia exemplars, pre- and posttreatment
functional dysphonia exemplars, and normal voices. The
self-organizing maps used only a selected set of acoustic
variables. No aerodynamic or VRP variables were in-
cluded. An overall correct classification of 76% was
achieved, which is a promising result. However, in our
view, the scope of such studies is most beneficial if a
outcome measure that describes vocal quality is offered,
rather than a classification method. In clinical practice,
it is the clinician who classifies the patient in one or
another diagnostic category.
In conclusion, there is a need for an objective mea-
sure that describes vocal quality. Preferentially, it should
reflect the multidimensional nature of voice and it must
be robust. This implies that many different investiga-
tors can use it without any effect on the result. Given
the fact that perceptual evaluation is the gold standard,
because it is based on experts’ opinion, the outcome
measure should be as much as possible in agreement
with clinicians’ perception. In this regard, the aerody-
namic, voice-range profile, and acoustic characteristics
of the voice singly do not adequately describe vocal qual-
ity. The present study attempts to construct such a mul-
tidimensional measure that reflects the overall vocal
quality based on an integration of voice-range profile,
aerodynamic, and acoustic measurements. To incorpo-
rate the perceptual nature of voice assessment, the in-
dex is based on perceptual severity ratings of vocal qual-
ity, rather than on the differentiation between normal
and pathologic voices. We call this measure an index
because Stewart and Ware (1992) define an index as “the
aggregation of two or more distinct health measures into
an overall summary measure.” Because it represents an
objective correlate of perceived voice quality, we refer to
it as the Dysphonia Severity Index (DSI).
Materials and Methods
Subjects
Voice samples were extracted from a database con-
structed for the Belgian Study Group on Voice Disorders
 (Van de Heyning et al., 1996). This database contained
the data of patients visiting the voice clinics of four uni-
versity centers in Belgium and two affiliated voice cen-
ters over a period of 16 months. The methods used for
the data collection are described in detail elsewhere (De
Bodt, Wuyts, Raes, & Qiu, 1996). For each patient a stan-
dardized protocol was followed. Data from a group of
387 adults (age range: 18–80 years; 53% female, 47%
male) were used for the calculation of the outcome mea-
sure. In order to avoid effects of vocal mutation as well
as maturation, we put the lower limit at 18 years old
because several authors suggest this as an appropriate
limit (Lee, 1980; Thurman & Klitzke, 1994). Among the
subjects were 68 controls (43 women, 25 men), with no
vocal fold pathology or any vocal complaint, recruited
at random in the different participating centers. A group
of 319 patients visiting the voice clinic with complaints
regarding their voice were included. Table 1 shows the
number of cases per pathology. The assessment included
history and clinical antecedents, stroboscopic evaluation,
perceptual assessment, acoustic analysis, voice range,
aerodynamic measurements, and self-rating. Criterion
validity of the index was assessed on a separate group
of 40 patients subsequently visiting the voice clinic of
the Antwerp University Hospital.
Data Recording
The acoustic parameters were obtained by the Multi
Dimensional Voice Program (MDVP, model 4305, Kay
Table 1. Number of cases per diagnostic category obtained from
the database of the Belgian Study Group on Voice Disorders.
Diagnosis N
Elemetrics Corp.) from the Computerized Speech Lab.
A midvowel segment on a sustained /a/—at habitual loud-
ness and pitch—was used. For reasons of general ap-
plicability, only jitter (%), shimmer (%), and noise-to-
harmonics ratio (NHR) were included for further
analysis because these are the most commonly and
widely used acoustical variables (De Bodt, 1997). The
aerodynamic measurements are maximum phonation
time (MPT in s), vital capacity (VC in cc), and phona-
tion quotient (PQ in cc/s). Voice range measurements
are (a) Lowest intensity (I-Low), (b) Highest intensity
(I-High), and (c) Intensity range (I-range = I-High – I-
Low). The I-range defines the dynamic intensity range
without reference to vocal frequency. Intensities were
recorded in dB(A) units. Frequency characteristics (in
Hz) yielded (a) Lowest F0 (F0-Low), (b) Highest F0 (F0-
High), (c) F0 range (F0 range = F0-High – F0-Low), and
(d) Semitone range (ST-range).
The perceptual ratings were performed using the
GRBAS scale as described by Hirano (1981), where the
G stands for Grade of hoarseness, R for roughness, B
for breathiness, A for astheny, and S for strain. Each
scale item is rated as 0 (normal), 1 (slight), 2 (moder-
ate), or 3 (severe).
Additionally, the presence of vocal abuse, allergy,
chronic obstructive lung disease, infection of the upper
respiratory tract, stress, nasal airway obstruction, and
smoking were recorded. Subjects were placed into one
of the four groups depending on their G score (0–3).
The variables used for the statistical analysis con-
sist of jitter (%), shimmer (%), NHR, F0-High (Hz), F0-
Low (Hz), F0-range (Hz), ST-range, I-High (dB), I-Low
(dB), I-range (dB), MPT (s), VC (cc), PQ (cc/s), and the G
score.

Control 68
Incomplete closure 48
Statistics
Vocal nodules 30 The correlation between the G score and other vari-
Reinke’s edema 30 ables was calculated with Spearman Rank statistics
Chronic laryngitis 27
because the G is a categorical variable. The normalcy of
Excessive muscular tension 26
the variables for the different groups was investigated
Paralysis in abduction 23
Tumor 22
using the Kolmogorov-Smirnov test. The equality of vari-
Sulcus glottidis and scar 21 ances for the four groups (i.e., the patients character-
Paralysis in adduction 20 ized with G0 to G3) was investigated by means of the
Mucosal cyst 14 generalized linear model ANOVA for all the variables
Granuloma 10 included. The Dysphonia Severity Index (DSI) was con-
Acute laryngitis 8 structed to be analogous to Fisher’s discriminant analy-
Haemoragy and trauma 7 sis (Fisher & Van Belle, 1993), a standard approach that
Spasmodic dysphonia 6 is used to differentiate two or more populations on the
Ventricular phonation 5 basis of several variables. Given the different subject
Polyps 4
populations, we set up a rule, based on the measure-
Psychogenic aphonia 3
ments of these subjects, whereby a new subject may be
Other 15
correctly assigned to one of the populations. When re-
Total 387 stricted to two populations and two variables, the

Wuyts et al.: Dysphonia Severity Index 799

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Fisher’s discriminant rule can be visualized as a dividing
line between the populations, represented as a contour
plot of the two variables in the plane (see Figure 1).
With three variables, a separation plane in three-
dimensional space is obtained. Because our data are not
normally distributed, a semiparametric approach was
applied that consisted of a proportional odds logistic
regression (Hosmer & Lemeshow, 1989). This technique
is less based on assumptions than the Fisher’s Linear
Discriminant analysis for these sets of variables, which
are particularly skewed. To select relevant variables for
the separation process, a stepwise selection procedure,
based on calculating the likelihood ratio test statistics,
was applied. Whereas the classical logistic regression
procedure allows the classification into only two groups,
proportional odds logistic regression allows more than
two groups. Because in this study we attempted to in-
corporate a perceptual severity rating into the multidi-
mensional measure, the grade of hoarseness, expressed
by the G score from the GRBAS scale and ranging from
0 to 3, was used to identify separate groups of subjects.
In other words, the entire group of subjects was divided
according to their G score into four groups. Only the G
was chosen because it embodies best the overall sever-
ity rating of vocal quality, whereas the other items re-
flect only parts of it. By means of proportional odds lo-
gistic regression a discrimination rule was determined
that enables the classification of a new patient into the
groups G0 to G3, based on the values of a specific set of
variables. This classification rule can then be trans-
formed to build the outcome measure denoted as the
“DSI.” The calculations are performed by the statistical
package Statistical Analysis Software system (SAS In-
stitute Inc, Cary, NC).
Figure 1. Illustration of the distribution of two variables for two
different populations (a & b) on two axes. The Fisher discrimant
rule can be regarded as the line that best separates both popula-
tions. The DSI can be visualized by the dashed line connecting
both population centers.
800 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000
Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
To assess the error on the DSI, due to measurement
errors of the variables that make up the DSI, we per-
formed a reliability study on 6 healthy subjects (3
women/3 men) without any vocal complaint. On five con-
secutive days, each of the variables was measured twice
a day. For each person we calculated the standard de-
viation for each variable, because the standard devia-
tion reflects the error due to the natural variability and
the method of measurement. Then, by taking the aver-
age of the earlier calculated standard deviations for all
individuals, we obtained an estimation of the typical
measurement error on the separate variables. Finally,
the delta method (Welsh, 1996) was used to calculate
the cumulative error on the DSI.
Voice Handicap Index
The psychosocial handicapping effect of the voice
disorders, as perceived by the patient, was measured by
means of the Voice Handicap Index (VHI; Jacobson et
al., 1997). The VHI questionnaire, which contains 30
items, assesses the patient’s judgment about the rela-
tive impact of his or her voice disorder upon daily ac-
tivities. It can also be useful as a component of measur-
ing functional outcomes in behavioral, medical, and
surgical treatments of voice disorders. It is statistically
robust, has a high internal consistency, and the test-
retest stability is strong (Jacobson et al., 1997). The VHI
therefore reflects the complaint of the patient regard-
ing his or her voice, which is in essence his or her rea-
son to seek professional help in a voice clinic. A high
correlation between the DSI and the VHI can be re-
garded as reflecting good agreement between the
patient’s perception of his or her voice problem, the ob-
jective measurements, and the clinician’s perception
through the intrinsic incorporation of the grade (G) in
the DSI. Thus, a significant correlation between the DSI
and the VHI can be considered a measure of criterion
validity of the DSI.
Forty consecutive patients for whom the DSI was
obtained completed the VHI questionnaire. The Pear-
son’s correlation coefficient was then calculated for DSI
versus VHI.
Results
Table 2 lists the averages, standard deviations, and
ranges for all investigated variables for the group of 68
healthy subjects and the 319 dysphonic cases.
The Spearman Rank correlation coefficients ρ be-
tween the G and the other variables for 387 cases are
listed in Table 3. The ρ2 values are additionally reported
because these values express the percentage of vari-
ability of the data that is explained by the association
 Table 2. Descriptive data on the acoustic, voice range, and aerodynamic variables for the group of 387 subjects.

Jitter (%) Shimmer (%) NHR

Subjects M SD Min Max M SD Min Max M SD Min Max N

Healthy 0.73 0.54 0.23 2.97 3.03 1.16 0.91 6.71 0.121 0.020 0.07 0.16 68
Dysphonic 2.55 2.33 0.21 14.12 6.39 4.18 1.37 19.87 0.204 0.151 0.06 1.17 319

F0-High (Hz) F0-Low (Hz) F0-Range (Hz) ST-Range

Subjects M SD Min Max M SD Min Max M SD Min Max M SD Min Max N

Healthy 794 240 329 1397 125 35 58 185 669 226 237 1280 32 5 22 44 68
Dysphonic 442 213 104 1108 115 39 55 300 328 203 26 985 22 8 0 49 319

I-High (dB) I-Low (dB) I-Range (dB)

Subjects M SD Min Max M SD Min Max M SD Min Max N

Healthy 97 7 85 112 51 2 43 57 46 7 35 61 68
Dysphonic 89 10 56 117 55 4 44 77 34 10 7 62 319

MPT (s) VC (cc) PQ (cc/s)

Subjects M SD Min Max M SD Min Max M SD Min Max N

Healthy 18.9 6.7 9.0 43.0 3788 1020 932 6300 216 68 75 379 68
Dysphonic 12.4 6.4 1.0 41.0 3131 995 400 6300 307 172 57 1400 319

Table 3. Spearman Rank Correlation coefficients (ρ) between G and the acoustic, aerodynamic, and voice range measurements. The second
row (ρ2) indicates the percentage of variability of the data that is explained by the association between the G and the other variables.

Jitter Shimmer F0-High F0-Low F0- ST- I-High I-Low I-range MPT VC PQ
(%) (%) NHR (Hz) (Hz) range range (dB) (dB) (dB) (s) (cc) (cc/s)

ρ 0.57 0.42 0.39 –0.42 –0.05 –0.45 –0.45 –0.39 0.34 –0.48 –0.38 –0.20 0.25
ρ2 0.32 0.17 0.15 0.18 0.00 0.20 0.20 0.15 0.12 0.23 0.14 0.04 0.06

between the G and the other variables. The significance
levels p were for all correlation coefficients smaller than
0.001 except for F0-Low, which did not exhibit a signifi-
cant relationship with G.
Except for I-High (dB) and the vital capacity (cc),
none of the variables were normally distributed. This
justifies the use of proportional odds logistic regression,
which defined the combination of the following variables
as indicators of the degree of hoarseness (G) when used
in a specific linear combination: F0-High (Hz), I-Low (dB),
MPT (s), and Jitter (%). The DSI, being the discriminat-
ing rule calculated by the logistic regression, consists of
a linear combination of these four variables, where each
variable has a different weight. The equation is:
DSI = 0.13 × MPT (s) + 0.0053 × F0-High (Hz)
– 0.26 × I-Low (dB) – 1.18 × Jitter (%)
+ 12.4
This DSI is the weighted combination of variables that
reflects best the degree of hoarseness as expressed by
the G from the GRBAS scale.
The relation between G and the DSI is represented
in Figure 2. The more negative this DSI is for a patient,
the worse his or her vocal quality. The more it is posi-
tive, the better it is. The initially obtained regression-
generated coefficients were post hoc multiplied with a
scale factor in order to construct a practical scale where
+5 corresponds to the average DSI of the G0 group and –5
to the average DSI of the G3 group. Table 4 represents
the DSI value after scaling for the different G scores.
Measurement errors on the individual components
of the DSI inevitably give rise to an error on the final
outcome measure. We calculated this error on the DSI
as 0.64, based on an average standard deviation of 1.6
seconds for MPT, 39 Hz for F0-High, 1.7 dB(A) for I-Low,
and 0.3% for jitter.
To estimate the reliability of the DSI, Table 5 rep-
resents the classification success of this method. This
table shows the agreement between the observed and
predicted perceived voice quality as expressed by G. In
50% of the cases a perfect agreement is obtained. When

Wuyts et al.: Dysphonia Severity Index 801

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 2. Dysphonia Severity Index versus Grade of hoarseness
defined in the GRBAS scale.
equal probability—the diagonal and one-off diagonal
cells represent in that case 10 out of 16 cells = 62.5%).
Table 6 reports the values for the DSI and its vari-
ables for the male and female subjects of the control
group. The sex effect is canceled out because of the op-
posite behavior of the MPT and F0-High for female and
male subjects (see Table 2), so only one version of the
DSI needs to be used for both sexes.
Figures 3 to 6 illustrate the DSI and the four vari-
ables for several cases, treated for hoarseness by either
voice therapy or surgery or a combination of both. The
changes of the individual variables and the global effect
on the DSI are clearly illustrated by several cases. The
shaded areas indicate the limits of normal, based on the
95% prediction intervals PI (Wuyts et al., 1996).
The Pearson’s correlation coefficient of the relation
between DSI and VHI is –0.79 (p < 0.001). Figure 7 shows
the relationship between both measures.

Table 4. Average of the DSI values (± standard error of the mean)

for the subjects characterized by G0 to G3 score.
Discussion
G score DSI ± SE Acoustic, aerodynamic, and voice-range measure-
ments of a patient’s voice are often within normal lim-
G0 5.00 ± 0.23
G1 1.02 ± 0.25 its. This is because these limits are in general very broad,
G2 –1.4 ± 0.3 as illustrated by the ranges in Table 2. Only values that
G3 –5.0 ± 0.8 deviate markedly from normal may be conclusive for
clinical purposes.
In this study we present an objective measure of
Table 5. Classification table of observed and predicted class
membership, indicating the power of the proportional odds logistic
vocal quality, whereby the multidimensional character
regression calculation. of voice is incorporated by means of the multivariate
statistics. The DSI consists of a specific weighted com-
Observed bination of the highest fundamental frequency, the low-
est intensity, the maximum phonation time, and the jit-
G0 G1 G2 G3 ter. The DSI was transformed in such a way that for the
G0 12 3 0 0 group of voices characterized by a G0 the DSI corresponds
Predicted

G1 44 91 61 6 to +5, and for those patients with a G3 it corresponds to

G2 0 52 83
G3 0 0 8 7
we consider, however, the cases where the neighboring
classification is also acceptable—for example a G2 scored
as a G1 or vice versa—then we obtain a “diagonal plus
one off-diagonal” classification of 98.4%. A purely ran-
dom classification would produce in that case only 62.5%
(i.e., all cells of the 4 × 4 matrix would be filled with
20 –5. The more negative this DSI is for a patient, the more
his or her voice can be regarded as dysphonic. The higher
it is, the better is his or her vocal quality. However, the
DSI is not per se limited to the interval +5, –5. In our
clinical practice we sometimes obtain values of –6 and
more. This is usually caused by high jitter values. In-
spection of the weights of the DSI reveals that when the
jitter is above 4%, the DSI will shift at least 5 points on
its scale towards negative values. In extreme cases, the
user should be aware of certain restrictions concerning

Table 6. Average values (± SE) of the DSI and its components for female and male subjects of the control
group.

MPT (s) F0-High (Hz) I-Low (dB) Jitter (%) DSI

Female (N = 43) 16.9 ± 0.7 905 ± 31 51.3 ± 0.2 0.79 ± 0.10 5.22 ± 0.26
Male (N = 25) 22.2 ± 1.7 602 ± 34 50.4 ± 0.5 0.63 ± 0.06 4.7 ± 0.4

802 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 3. Follow-up of a 58-year-old male patient with dysphonia due to a paralysis of the n recurrens following surgery of the lower
cranial nerves. Voice therapy during 8 months. The time indications are (1) data 15 days postoperatively (p.o.): G2R1B2A2S0, start of voice
therapy. (2) Data 5 months p.o.: G2R1B1A1S0. (3) Data 8 months p.o.: end of voice therapy, G0R0B0A0S0. (4) Data 16 months p.o.:
G0R0B0A0S0, follow-up assessment. (5) Data 18 months p.o.: G0R0B0A0S0, follow-up assessment. Shaded area represents 95% Prediction
Interval (mean ± 1.96 SD).

the nonperiodicity of some pathologic voice samples, as
reported by Titze (1995).
The choice of variables in the DSI is entirely deter-
mined by the stepwise logistic regression procedure. How-
ever, it seems logical that the highest frequency is among
the chosen ones. In more than 50% of the dysphonic
patients the vocal cords are afflicted with an excess mass
(vocal nodules, edema, etc.; see Table 1). This extra mass,
usually heterogeneously distributed along the cords,
hampers the higher vibratory rates, which is reflected
by a decreased F0-High. Likewise the presence of nod-
ules, edema, and so forth increases the glottal resistance
such that a greater driving pressure will be necessary
to initiate and maintain vocal fold vibration (Colton,
1994). Consequently the lowest intensity will be in-
creased in several dysphonic patients. Similar effects
for F0-High and I-Low are found in VRP studies of chil-
dren with vocal nodules (Heylen et al., 1998). Perturba-
tion measures, such as jitter, are per se intended to as-
sess the degree of irregularity of the vocal cord vibration,
within certain limits. It is likely that a perceived dys-
phonia will result in an increased perturbation measure.
The reason for the choice of MPT as a relevant variable
included in the DSI may lie in the fact that MPT can be
regarded as a phonatory ability measure (Hirano, 1981)
that reflects the efficiency of several mechanisms nec-
essary for voice production, such as subglottic pressure,
airflow resistance, closure of the vocal folds, and so forth.
Piccirillo et al. (1998) elaborated a concept of a vo-
cal function index, but only the classification between
normal and dysphonic was emphasized. Using logistic
regression, they found that a weighted combination of
estimated subglottic pressure, airflow at lips, vocal effi-
ciency, and maximum phonation time was able to dis-
criminate between healthy and pathologic voices. It is
noteworthy that the MPT emerges from both their study
and our work as an important variable for the overall
assessment of voice quality.
In order to validate their index Piccirillo and cowork-
ers compared it with the GRBAS score for a group of 33
patients with limited vocal dysfunction (Piccirillo,
Painter, Fuller, Haiduk, & Fredrickson, 1998). The cor-
relation coefficient they found between their index and
the G was 0.58, whereas we found a value of 0.996 for

Wuyts et al.: Dysphonia Severity Index 803

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 4. Follow up of a 54-year-old patient who underwent a surgical reduction of Reinkes oedema. The patient is an amateur singer with
bad vocal hygiene. The time indications are (1) 3 weeks postsurgery (p.o.): G2R2B1A1S1, start of voice therapy, and vocal hygiene program.
(2) Three months p.o.: G1R1B0A1S1. (3) Six months p.o.: G1R1B1A0S0, end of voice therapy. The patient shows a bad compliance in vocal
hygiene; patient keeps on smoking and singing. (4) Twenty months p.o.: G1R1B0A0S1, no changes in vocal habits or laryngeal condition.

the correlation between DSI and G. This is because in
our approach the DSI is based on the G score itself rather
than on the discrimination between normal and patho-
logic voices. Additionally, the relationship between pa-
thology and dysphony is not obvious, because a severe
pathology does not always strictly imply a severely dys-
phonic voice, and vice versa. Therefore we have adopted
the perceptual rating as a landmark for the classifica-
tion rule.
The classification table (Table 5) illustrates the ef-
ficiency of the applied method. An ideal classification
tool would produce values only on the diagonal, which
is of course not achieved in practice. When a group of
judges is scoring a number of patients, the interobserver
agreement is at best “good,” but never excellent, as ex-
pressed by the kappa statistic (De Bodt et al., 1997).
This means that a certain variation or test-retest error
exists in the G score itself. Therefore, it is realistic to ex-
pect that in some cases a perceived G1 might as well have
been a G2 or vice versa. This is translated by off-diagonal
elements in the classification table. It implies that when
voices are perceived as being characterized by a specific
G value, there is considerable probability that their cal-
culated G, based on the DSI, falls into an adjacent cat-
egory. This is apparently the case as indicated by Table 5.
Moreover, the DSI seems to classify better subjects with
G1 and G2 than with G0 or G3. Still, only 6 out of 387 cases
are really misclassified by more than one scale point. Fi-
nally, the DSI is not meant as a classification tool.
Fortunately, the effect of sex is included implicitly
in the DSI, so that a separate DSI for males and fe-
males need not be used. As seen in Table 6, the opposite
behavior of F0-High and MPT for both sexes cancels out
so that DSIs for both male and female subjects are iden-
tical. Indeed, the difference between the average female
and male DSI is not significant; in general the error on
the DSI is estimated as 0.6.
To assess the clinical impact of the DSI and its ease
of use we present some follow-up cases. Figures 4 to 7
illustrate changes in vocal quality with different thera-
pies. For some cases the pathology was still present after
therapy, but the vocal function had improved, according

804 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 5. Follow up of a 34-year-old female patient who underwent a vocal fold medialization (thyroplasty, Isshiki II) due to laryngeus
superior paralysis. The time indications are (1) Pre-operative assessment: G2R0B2A2S0. (2) Ten days postsurgery (p.o.): worsening of quality
due to oedema, G3R0B2A2S0, start of voice therapy. (3) Four weeks p.o.: G1R0B1A0S0 . (4) Six weeks p.o.: G1R0B1A0S0. (5) Eight weeks p.o.:
G1R0B1A0S0. (6) Ten weeks p.o.: G1R0B1A0S0. The patient is very satisfied about her current voice quality. End of voice therapy. (7) Eight
months p.o.: G1R0B1A0S0. A control follow-up assessment reveals a stable voice quality.

to the otolaryngologist and the patient. In other cases,
therapy may have improved one variable, but when other
variables became worse, the DSI reflected this overall
change.
It takes 10 to 15 min to collect clinical measures
(MPT, etc.) from a patient and to calculate the index
using the above-mentioned equation. For this calcula-
tion a desktop calculator or spreadsheet is sufficient.
Additionally, the use of anchor points of –5 and +5 fa-
cilitate the clinical use of the DSI. These aspects con-
tribute largely to the ease of use of the DSI. In the Ap-
pendix a recommended clinical recording protocol that
yields the DSI is described.
Additionally we want to address the content and
criterion validity of the DSI. Content validity refers to
whether the index measures what it is intended to—
that is, the degree of dysphonia. The four variables used
in the DSI are individually all clear indicators of dys-
phonia, because their averages are significantly differ-
ent for patients with vocal pathology as opposed to nor-
mal subjects (Wuyts et al., 1996). Additionally, the
selection of these variables is based on a statistical
stepwise procedure that constructs a rule to classify
voices that are characterized by the scores G0 to G3,
which in turn represents the degree of dysphonia as
perceived by the voice specialist. Considering these facts,
it seems reasonable that the DSI meets the criteria of
content validity.
Criterion validity refers to the accuracy of the DSI.
How does it compare to a “gold standard”? Auditory-
perceptual judgments are typically the final arbiter in
clinical decision-making and often provide the standards
against which instrumental measures are evaluated
(Kent, 1996). Inherently, the construction of the DSI is
based on such a standard, being the Grade of the widely
used perceptual GRBAS scale. To compare the DSI with
an external measure, we correlated the DSI with the Voice
Handicap Index. The high correlation between both mea-
sures adds to the criterion validity of the DSI. Moreover,
this high correlation indicates that the DSI reflects not
only the vocal quality of the patient but also reflects to
a great extent the handicap as perceived by the patient.

Wuyts et al.: Dysphonia Severity Index 805

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 6. Follow up of a 42-year-old female patient with dysphonia due to bilateral Reinke’s edema of vocal folds, phonosurgically treated
with bilateral cordotomy. Additionally the patient received anti-reflux medication but no functional voice therapy. The time indications are (1)
Preoperative data: G2R2B2A1S0. (2) Two months postoperative (p.o.): G2R1B2A1S0. (3) Three months p.o.: G1R0B1A1S0 .

Improvement of the DSI can most probably be
achieved by the use of acoustic variables that are de-
rived from running speech samples rather than from a
sustained vowel. Also other types of variables, such as
the airflow at lips, subglottic pressure, and so forth,
might be included. Other methods, such as neural-net-
work approaches and self-organizing maps, might prove
to be superior to multivariate statistical tools, such as
logistic regression or discriminant analysis (Callan et
al., 1999). Inevitably, again, a multicenter study with
several hundred subjects is needed for the development
of a new DSI-like outcome measure.
Daily clinical use of the DSI for the past 18 months
has shown the authors that the DSI is a practical tool to
describe voice quality in a well-balanced way. It plays a
valuable part in the global assessment of a dysphonic
patient. A model for voice assessment may consist of four
components (De Bodt, 1997): laryngeal inspection, per-
ceptual evaluation (e.g., GRBAS), subjective evaluation
by the patient him- or herself (e.g., VHI), and the DSI.
It is our belief that such a four-component model is suf-
ficient in a clinical setting to assess voices in a scientifi-
cally relevant way within a limited amount of time.
The parameters included in the DSI have become
quite accessible in most voice clinics throughout the
world. The DSI is objective because no perceptual input
is required for its calculation. The DSI’s small measure-
ment error (0.6 on 10 points) and the fact that a multi-
center database was used for its construction underlie
its robustness. These factors, together with the fact that
the DSI is based on aerodynamic, voice range, and acous-
tic measurements, make the DSI a multidimensional,
robust, and objective outcome measure for the assess-
ment of vocal quality. It provides the individual voice
therapist with an outcome measure for an individual
patient, without being biased by time, subjective evalu-
ation, or other factors that influence perceptual ratings.
It enables clinicians to place a voice in an absolute way,
so that therapy can be discussed and its effectiveness
evaluated. Its universal use can enable the comparison

806 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Figure 7. Relationship between the Voice Handicap Inventory and
the Dysphonia Severity Index for 40 consecutive patients visiting
the voice clinic. A low VHI score corresponds to a healthy voice.
of therapy techniques and surgical treatment, as well
as therapy efficiency by follow-up assessment. There-
fore, it could prove valuable to both research and daily
clinical practice.
Acknowledgments
We wish to thank Soetkin Debelder for the syntax check
as well as all subjects contributing to this study by lending
their voice. Part of this work was presented at the IALP
meeting in August 1998 in Amsterdam.
References
Andrews, M. L. (1995). Manual of voice treatment: Pediat-
rics through geriatrics. San Diego: Singular Publishing
Group, Inc.
Askenfeld, A. G., & Hammarberg, B. (1986). Speech
waveform perturbation analysis: A perceptual-acoustical
comparison of seven measures. Journal of Speech and
Hearing Research, 29, 50–64.
Baken, R. J. (1987). Clinical measurement of speech and
voice. London: Taylor & Francis Ltd.
Bassich, C. J., & Ludlow, C. (1986). The use of perceptual
methods by new clinicians for assessing voice quality.
Journal of Speech and Hearing Disorders, 51, 125–133.
Blaustein, S., & Bar, A. (1983). Reliability of perceptual
voice assessment. Journal of Communication Disorders,
16, 157–161.
Callan, D. E., Kent, R. D., Roy, N., & Tasko, S. M. (1999).
Self-organizing map for the classification of normal and
disordered female voices. Journal of Speech, Language,
and Hearing Research, 42, 355–366.
Colton, R. (1994). Physiology of phonation. In M. S.
Benninger, B. H. Jacobson, & A. F. Johnson (Eds.), Vocal
arts medicine: The care and prevention of professional voice
disorders (pp. 30–60). Stuttgart: George Thieme Verlag.
Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
Crevier-Buchman, L., Laccourreye , O., Wuyts, F. L.,
Monfrais-Pfauwadel, M., Pillot, C., & Brasnu, D.
(1998). Comparison and evolution of perceptual and
acoustic characteristics of voice after supracricoid partial
laryngectomy with cricohyoidoepiglottopexy. Acta
Otolaryngologica, 118, 594–599.
De Bodt, M. S. (1997). A framework for voice assessment:
The relation between subjective and objective parameters
in the judgment of normal and pathological voice. Doctoral
dissertation. University of Antwerp.
De Bodt, M. S., Wuyts, F. L., Raes, J. P. F., & Qiu, J.
(1996). Research work of the Belgian Study Group on
Voice Disorders 1996: Materials and methods. Acta Oto-
Rhino-Laryngologica Belgica, 50, 325–329.
De Bodt, M. S., Wuyts, F. L., Van de Heyning, P. H., &
Croux, C. (1997). Test-retest study of the GRBAS Scale:
The influence of experience and professional background
on perceptual rating of voice quality. Journal of Voice,
11(1), 74–80.
Dejonckere, P. H., & Lebacq, J. (1996). Acoustic, percep-
tual, aerodynamic and anatomical correlations in voice
pathology. Journal of Otorhinolaryngology and Its Related
Specialties, 58, 326–332.
Dejonckere, P. H., Remacle, M., Fresnel-Elbaz, M.,
Woisard, V., Crevier-Buchman, L., & Millet, B. (1996).
Differential perceptual evaluation of pathological voice
quality: Reliability and correlations with acoustic mea-
surements. Revue de Laryngologie Otologie Rhinologie,
117, 219–224.
Dejonckere, P. H., & Wieneke, G. H. (1992). G.R.B.A.S.-
scaling of pathological voices: Reliability, clinical rel-
evance, and differentiated correlation with acoustic
measurements, especially with cepstral measurements. In
E. Loebell (Ed.), Proceedings of the 22th World Congress
IALP, 1992. Hannover: IALP.
Eskenazi, L., Childers, D. G., & Hicks, D. M. (1990).
Acoustic correlates of vocal quality. Journal of Speech and
Hearing Research, 33, 298–306.
Fisher, L. D., & Van Belle, G. (1993). Biostatistics: A
methodology for the health sciences. New York: John Wiley
& Sons, Inc.
Gelfer, M. P. (1988). Perceptual attributes of voice: Develop-
ment and use of rating scales. Journal of Voice, 2, 320–326.
Hammarberg, B. (1992). Pathological voice qualities.
Perceptual and acoustic characteristics of a set of Swedish
“reference” voices. Bulletin d’Audiophonology, 8, 39–52.
Hammarberg, B., Fritzell, B., Gauffin, J., Sundberg,
J., & Wedin, L. (1980). Perceptual and acoustic corre-
lates of abnormal voice qualities. Acta Otolaryngologica,
90, 441–451.
Heylen, L., Wuyts, F. L., Mertens, F., De Bodt, M.,
Pattyn, J., Croux, C., & Van de Heyning, P. (1998).
Evaluation of the vocal performance of children using a
voice range profile index. Journal of Speech, Language,
and Hearing Research, 41, 232–238.
Hillenbrand, J., Cleveland, R. A., & Erickson, R. L.
(1994). Acoustic correlates of breathy vocal quality.
Journal of Speech and Hearing Research, 37, 769–778.
Hirano, M. (1981). Clinical examination of voice. New York:
Springer Verlag.
Wuyts et al.: Dysphonia Severity Index 807
 Hirano, M. (1990). Clinical applications of voice tests. In
Assessment of speech and voice production (pp. 196–203).
Rockville, MD: NIDCD.
Hirano, M., Hibi, S., Terasawa, R., & Fujiu, M. (1986).
Relationship between aerodynamic, vibratory, acoustic
and psychoacoustic correlates in dysphonia. Journal of
Phonetics, 14, 445–456.
Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic
regression. New York: Wiley.
Jacobson, B., Johnson, A., Grywalski, C., Silbergleit,
A., Jacobson, G., Benninger, M. S., & Newman, C. W.
(1997). The Voice Handicap Index (VHI): Development and
validation. American Journal of Speech-Language
Pathology, 6, 66–70.
Kent, R. D. (1996). Hearing and believing: Some limits to
the auditory-perceptual assessment of speech and voice
disorders. American Journal of Speech-Language Pathol-
ogy, 5(3), 7–23.
Kreiman, J., Gerratt, B. R., & Berke, G. S. (1994). The
multidimensional nature of pathologic vocal quality.
Journal of the Acoustical Society of America, 96, 1291–1301.
Kreiman, J., Gerratt, B. R., Kempster, G. B., Erman,
A., & Berke, G. S. (1993). Perceptual evaluation of voice
quality: Review, tutorial, and a framework for future
research. Journal of Speech and Hearing Research, 36,
21–40.
Laver, J. (1980). The phonetic description of voice quality.
London: Cambridge University Press.
Lee, P. A. (1980). Normal ages of pubertal events among
American males and females. Journal of Adolescent
Health Care, 1(1), 26–29.
Piccirillo, J. F., Painter, C., Fuller, D., & Fredrickson,
J. M. (1998). Multivariate analysis of objective vocal
function. Annals of Otology Rhinology Laryngology, 107(2),
107–112.
Piccirillo, J. F., Painter, C., Fuller, D., Haiduk, A., &
Fredrickson, J. M. (1998). Assessment of two objective
voice function indices. Annals of Otology Rhinology
Laryngology, 107(5, Pt 1), 396–400.
Rabinov, C. R., Kreiman, J., Gerratt, B. R., &
Bielamowics, S. (1995). Comparing reliability of percep-
tual ratings of roughness and acoustic measures of jitter.
Journal of Speech and Hearing Research, 38, 26–32.
Revis, J., Giovanni, A., Wuyts, F., & Triglia, J. M. (1997).
Comparison of different phonetic materials for perceptive
analysis of dysphonia. Revue de Laryngologie Otologie
Rhinologie, 118(4), 247–252.
Revis, J., Giovanni, A., Wuyts, F. L., & Triglia, J. M.
(1999). Comparison of different voice samples for
perceptual analysis. Folia Phoniatrica et Logopedica, 51,
108–116.
Stewart, A. L., & Ware, J. E. (1992). Measuring functioning
808 Journal of Speech, Language, and Hearing Research • Vol. 43 • 796–809 • June 2000
Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
and well-being: The medical outcomes study approach.
Durham, NC: Duke University Press.
Thurman, L., & Klitzke, C. A. (1994). Voice education and
health care for young voices. In M. S. Benninger, B.
Jacobson, & A. Johnson (Eds.), Vocal arts medicine: The
care and prevention of professional voice disorders (pp.
226–268). New York: Thieme Medical Publishers, Inc.
Titze, I. R. (1995). Workshop on acoustic voice analysis.
Summary statement. Denver: National Center for Voice
and Speech.
Van de Heyning, P. H., Remacle, M., Van Cauwenberge,
P., De Bodt, M. S., Wuyts, F. L., Moerman, M., Van
Lierde, K., Raes, J. P. F., Heylen, L., Sasserath, M.,
Henri de Hassonville, A., Clement, P. A. R., Qiu, J.,
Molenberghs, G., Bruckers, L., Mertens, F., Pattyn,
J., & Millet, B. (1996). Research work of the Belgian
Study Group on Voice Disorders 1996. Acta Oto-Rhino-
Laryngologica Belgica, 50, 321–386.
Welsh, A. H. (1996). Aspects of statistical inference. New
York: Wiley.
Wendler, J., Rauhut, A., & Krüger, H. (1986). Classifica-
tion of voice qualities. Journal of Phonetics, 14, 483–488.
Wilson, D. K. (1987). Voice problems of children (3rd ed.).
Baltimore: Williams & Wilkins.
Wirz, S., & Mackenzie Beck, J. (1995). Assessment of
voice quality: The vocal profiles analysis scheme. In S.
Wirz (Ed.), Perceptual aproaches to communication
disorders (pp. 39–55). London: Whurr Publishers.
Wolfe, V., Fitch, J., & Cornell, R. (1995). Acoustic predic-
tion of severity on commonly occuring voice problems.
Journal of Speech and Hearing Research, 38, 273–279.
Wuyts, F. L., De Bodt, M. S., Bruckers, L., &
Molenberghs, G. (1996). Research work of the Belgian
Study Group on Voice Disorders 1996: Results. Acta Oto-
Rhino-Laryngologica Belgica, 50, 331–341.
Wuyts, F. L., De Bodt, M. S., & Van de Heyning, P. H.
(1999). Is the reliability of a visual analog scale higher
than that of an ordinal scale? An experiment with the
GRBAS scale for the perceptual evaluation of dysphonia.
Journal of Voice, 13, 508–517.
Wuyts, F. L., De Bodt, M. S., Van de Heyning, P. H., &
Van Hoof, R. (1998). Perceptual evaluation of voice
quality with and without clincal patient information
[Abstract]. IALP Conference, Amsterdam.
Received May 13, 1999
Accepted November 22, 1999
Contact author: Floris L. Wuyts, PhD, University Hospital of
Antwerp, Department of Otorhinolaryngology and Head
and Neck Surgery, Wilrijkstraat 10, B-2650 Edegem,
Belgium. Email: wuyts@uia.ua.ac.be
 Appendix. Clinical recording protocol.

1. Perceptual assessment by means of the GRBAS system.

2. Determination of Maximum Phonation Time (in s) on the
sustained vowel /a/ at habitual pitch and loudness. The
best of three trials is selected for calculation of the DSI.
3. Determination of highest frequency (Hz) and lowest
intensity (dB SPL) by means of, for example, an automatic
Voice Range Profile recording system (VRP Kay
Elemetrics). The patient is instructed to produce an /a/
and is encouraged by the feedback on the VRP-display.
The patient is allowed to use glides or scales to achieve the
highest frequency. The best value of three efforts is
selected as well for highest frequency as for lowest
intensity.
4. Determination of the jitter (%) by means of the Multi-
Dimensional Voice Program (Kay Elemetrics, Model 3405)
on a sustained vowel /a/ with a duration of approxi-
mately 3 s.
Calculation of the DSI using the equation:
DSI = 0.13 × MPT (s) + 0.0053 × F0-High (Hz) – 0.26 ×
I-Low (dB) – 1.18 × Jitter (%) + 12.4
This can be achieved by using, for example, an Excel
spreadsheet. The DSI is immediately available. The
spreadsheet can be obtained from the first author
(wuyts@uia.ua.ac.be).

Wuyts et al.: Dysphonia Severity Index 809

Downloaded From: http://jslhr.pubs.asha.org/ by a Univ of York-England User on 02/18/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx