S. Bielamowicz et al., COMPARISON OF VOICE ANALYSIS SYSTEMS FOR PERTURBATION MEASUREMENT, Journal of speech and hearing research, 39(1), 1996, pp. 126-134
Dysphonic voices are often analyzed using automated voice analysis sof
tware. However, the reliability of acoustic measures obtained from the
se programs remains unknown, particularly when they are applied to pat
hological voices. This study compared perturbation measures from CSpee
ch, Computerized Speech Laboratory, SoundScope, and a hand marking voi
ce analysis system. Sustained vowels from 29 male and 21 female speake
rs with mild to severe dysphonia were digitized, and fundamental frequ
ency (F-0), jitter, shimmer, and harmonics- or signal-to-noise ratios
were computed. Commercially available acoustical analysis programs agr
eed well, but not perfectly, in their measures of F-0. Measures of per
turbation in the various analysis packages use different algorithms, p
rovide results in different units, and often yield values for voices t
hat violate the assumption of quasi-periodicity. As a result, poor ran
k order correlations between programs using similar measures of pertur
bation were noted. Because measures of aperiodicity apparently cannot
be reliably applied to voices that are even mildly aperiodic, we quest
ion their utility in quantifying vocal quality, especially in patholog
ical voices.