BACKGROUND: Delineation of Parkinson's disease (PD) from multiple system atrophy (MSA) can be challenging in early disease stages. Speech characteristics have been studied as digital biomarkers in PD and ataxias. Currently, data on speech in MSA is limited.

OBJECTIVES: To determine whether speech characteristics can serve as a digital biomarker to differentiate between MSA and PD.

METHODS: Twenty-one MSA patients and 23 PD patients underwent a battery of speech assessments: text reading, sustained phonation and diadochokinetic tasks. Speech characteristics were extracted using the software, Praat.

RESULTS: MSA and PD speech can be described by three meaningful factors. MSA speech exhibited more reading pauses, higher pitch variability, prolonged syllables, and a more irregular speech rhythm, allowing differentiation from PD with a ROC-AUC of 0.89. Speech characteristics were correlated with motor impairment and disease severity.

CONCLUSION: MSA can be differentiated from PD with good accuracy using speech analysis.