Acoustic characteristics associated with speech naturalness and listening effort of dysarthric speech

The aim of the current study was to determine the acoustic characteristics that contribute to speech naturalness and listening effort of dysarthric speech. Listeners heard words produced by speakers with dysarthria and rated either their speech naturalness or listening effort. They also identified the productions, which provided response time as a behavioral measure of listening effort. We examined the associations between the perceptual domains and the following vowel metrics for each production: formants 1 and 2 (F1 and F2), vowel duration, mean fundamental frequency (F0), and cepstral peak prominence (CPPS). Preliminary analyses were conducted using linear mixed-effects models to determine the effects of each acoustic variable on speech naturalness ratings, perceived listening effort, and response time. Listeners and speakers were included as random intercepts. The results indicated that only vowel duration was significantly associated with speech naturalness ratings. In contrast, vowel duration, F1, F2, and CPPS were significantly associated with both perceived listening effort and response time. Only mean F0 was not associated with listening effort. These results enhance our understanding of how dysarthria affects speech production and perception by using multiple productions from each speaker, allowing us to investigate within-speaker acoustic characteristics and their impact on perception.