Acoustic biomarkers of Parkinson’s Disease: a pilot study on the Italian Parkinson’s Voice and Speech dataset

Abstract

Beyond the most well-known motor symptoms, Parkinson's disease can also affect patients' speech and language abilities. This article presents an acoustic analysis of the Italian Parkinson's Voice and Speech dataset, comprising recordings of 65 speakers from southern Italy (28 patients, 22 elderly healthy controls, and 15 young healthy controls). The results reveal significant differences between pathological and control subjects across several metrics, including jitter, shimmer, HNR, and the Higuchi fractal dimension. Significant alterations were observed in vowel metrics, with a marked decrease in tVSA and VAI, and an increase in FCR. Consistent with existing literature, the study also confirms the documented reduction in pitch variability among patients. To complete the analysis, two additional variables (UPDRS scores and biological sex) were taken into account; notably, opposing trends between men and women emerged on duration measures (distribution of pauses, sound segments, and measures related to speech rate).