Acoustic speech markers for tracking changes in hypokinetic dysarthria associated with Parkinson's Disease

Acoustic analysis has been successfully used for differential diagnosis of hypokinetic dysarthria associated with Parkinson’s disease. However, acoustic parameters that are robust for differential diagnosis may not be sensitive to tracking speech changes over time. Previous longitudinal studies implementing acoustic analysis have had limited sample sizes or variable lengths between data collection. This study focused on using acoustic correlates of perceptual features to identify acoustic markers able to track speech changes in people with Parkinson’s Disease (PwPD) over time.

Speech data was collected from 63 PwPD and 47 control speakers using an online podcast software at two time points, six months apart (T1 and T2). Recordings of a standard reading passage, minimal pairs, sustained phonation, and spontaneous speech were collected. Perceptual severity ratings were given by two speech and language therapists for T1 and T2, and acoustic parameters of voice, articulation and prosody were investigated. Two analyses were conducted to i) identify which acoustic parameters can track perceptual speech changes and ii) identify which acoustic parameters can track changes in speech intelligibility. Additionally, both analyses investigated if the acoustic parameters showed group differences for differential diagnosis between PwPD and control speakers at T1 and T2. Linear mixed model statistical analysis was conducted to answer the research questions.

Results showed acoustic parameters in articulation could detect significant group and speech change differences across T1 and T2. Other acoustic parameters in voice quality, articulation and prosody could either only differentiate between PwPD and controls or detect speech changes between T1 and T2. Results show promise for using acoustic parameters as markers to track speech changes in PwPD. Studies with longer data collection periods will indicate whether a trajectory of change can be observed. Further investigation could facilitate to the development of speech monitoring systems used by PwPD and speech and language therapists.