Objective: This study aims to extract acoustic features from speech data obtained from patients undergoing the Assessment of Motor Speech for Dysarthria (AMSD), a clinical method used to assess dysarthria patterns  across disease groups.

Background: Basal ganglia disorders are associated with hypokinetic dysarthria, characterized by difficulty initiating speech and switching articulation. On the other hand, cerebellar disorders lead to ataxic dysarthria, characterized by prolonged speech duration and irregular speech patterns due to impaired motor planning. However, few objective methods can clearly distinguish between these two dysarthria patterns.

Method: We performed AMSD in patients diagnosed with Parkinson’s disease (PD), spinocerebellar degeneration (SCD), and multiple system atrophy (MSA-C and MSA-P). Results were compared with age-matched healthy controls (HC). We analyzed audio data of an AMSD task consisting of rapid repetition of the syllables /pa/, /ta/, and /ka/. Acoustic features for duration, pitch, intensity, and formant frequencies were analyzed. Formant frequencies reflect tongue position and contribute to aspects of timbre. We also analyzed maximum phonation time (MPT), from which we extracted acoustic features for jitter and shimmer, representing frequency and amplitude perturbations in vocal fold vibrations.

Results: We conducted AMSD on 30 HC, 15 PD, 8 MSA-C, 4 SCD, and 2 MSA-P. Compared with HC, standard deviations of pitch, intensity, Formants 1 and 2 were significantly smaller in PD (p < 0.05), but larger in MSA-C (p < 0.05). SCD showed trends similar to MSA-C. MPT was shorter in all disease groups. Jitter and shimmer tended to be larger in MSA-C and SCD than in HC (jitter: p < 0.05), whereas shimmer was smaller in PD (p < 0.05).

These findings suggest that cerebellar dysarthria (MSA-C and SCD) showed irregular acoustic profiles with prolonged duration and instability, whereas basal ganglia disorders such as PD present less irregularity with reduced initiation and acoustic modulation.

Conclusion: By extracting acoustic features from AMSD data, we were able to quantify the characteristics of dysarthria and more precisely evaluate differences in speech features among various diseases associated with parkinsonism and ataxia. Combining neurological examinations with acoustic analyses is expected to allow for even more detailed evaluations of dysarthria.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/quantitative-evaluation-of-speech-characteristics-in-assessment-of-motor-speech-for-dysarthria-amsd-using-acoustic-analysis/