Cepstral analysis of hypokinetic and ataxic voices: correlations with perceptual and other acoustic measures.

OBJECTIVES To investigate the validity of cepstral analyses against other conventional acoustic measures of voice quality in determining the perceptual impression in different motor speech disorders-hypokinetic and ataxic dysarthria, and speech tasks-prolonged vowels and connected speech. METHODS Prolonged vowel productions and connected speech samples (reading passages and monologues) from 43 participants with Parkinson disease and 10 speakers with ataxia were analyzed perceptually by a trained listener using GRBAS. In addition, acoustic measures of cepstral peak prominence (CPP), smoothed CPP (CPPs), harmonics-to-noise ratio (HNR), shimmer %, shimmer dB, amplitude perturbation quotient (APQ), relative average perturbation (RAP), jitter, and pitch perturbation quotient (PPQ) were performed. Statistical analysis involved correlations between perceptual and acoustic measures, as well as determination of differences across speaker groups and elicitation tasks. RESULTS CPP and CPPs results showed greater levels of correlation with overall dysphonia, breathiness, and asthenia ratings than the other acoustic measures, except in the case of roughness. Sustained vowel production produced a higher number of significant correlations across all parameters other than connected speech, but task choice did not affect CPP and CPPs results. There were no significant differences in any parameters across the two speaker groups. CONCLUSIONS The results of this study are consistent with the results of other studies investigating the same measures in speakers with nonmotor-related voice pathologies. In addition, there was an indication that they performed better in relation to asthenia, which might be particularly relevant for the current speaker group. The results support the clinical and research use of CPP and CPPs as a quantitative measure of voice quality in populations with motor speech disorder.