Acoustic-prosodic markers in older adults with and without Parkinson's disease

To describe and compare acoustic-prosodic measures derived from reading aloud and spontaneous speech, analysed jointly, in older adults with and without Parkinson's disease (PD), using speech data collected through mobile phone recordings in both in-person and remote conditions.

The study sample comprised 39 older adults (19 males and 20 females), aged 60-93 years, including individuals with PD at different clinical stages and neurologically healthy controls. Twenty-four acoustic-prosodic features related to fundamental frequency (f_o), intensity, voice quality, and pause were semi-automatically extracted using Praat's Prosody Descriptor Extractor from 821 reading and spontaneous speech samples collected between 2020 and 2024. Effect sizes guided sex-based group comparisons.

Key differences between controls and individuals with PD were found in speech rate (large effect), and in articulation rate, coefficient of variation for intensity (cvint), spectral emphasis (emph; an acoustic correlate of vocal effort), shimmer, and standard deviation of f_o (f_osd) (moderate effects). These showed consistent, sex-specific PD-related changes: slower rates, reduced f_osd, increased cvint, and altered shimmer.

Among the twenty-four acoustic-prosodic parameters analysed, temporal measures, particularly speech and articulation rates, showed the largest effect sizes and most consistently differentiated older adults with and without PD. Cvint, shimmer, and f_osd contributed at a secondary level, with relative relevance varying according to PD severity. Overall, alterations in speech timing constitute the primary acoustic dimension distinguishing PD-related speech changes from age-related effects in this dataset, supporting the descriptive relevance of speech-based acoustic-prosodic analyses obtained from mobile phone recordings in non-invasive research settings.