Screening speech disorders in progressive neurological diseases via long-term average spectrum

Area of Science:

• Speech Science
• Neurology
• Biomedical Engineering

Background:

• Dysarthria, a speech disorder, is common in various neurological diseases.
• Existing methods for dysarthria analysis may lack specificity across diverse conditions.
• Long-term averaged spectrum (LTAS) offers a potential universal approach to speech pattern analysis.

Purpose of the Study:

• To assess the sensitivity of LTAS descriptors for identifying dysarthria.
• To evaluate LTAS's effectiveness across a wide spectrum of neurological diseases and dysarthria types/severities.
• To explore LTAS's utility in the differential diagnosis of neurological speech disorders.

Main Methods:

• Collected reading passage data from 461 speakers (306 healthy, 155 neurological patients).
• Analyzed four LTAS spectral moments: mean, standard deviation, skewness, and kurtosis.
• Included patients with Parkinson's disease, multiple sclerosis, cerebellar ataxia, and other neurological conditions.

Main Results:

• Spectral mean differed significantly between controls and patients with Parkinson's disease and multiple sclerosis (lower) and cerebellar ataxia (higher).
• Significant LTAS feature changes were primarily noted in hypokinetic dysarthria and mixed dysarthrias with hypokinetic components.
• No clear progressive increase in LTAS feature changes with increasing dysarthria severity was observed.

Conclusions:

• LTAS-based speech analysis can provide valuable insights for differential diagnosis in neurological diseases with overlapping symptoms.
• LTAS is more informative when analyzing speech from specific neurological diseases rather than pooling diverse dysarthria types.
• LTAS analysis shows promise as a tool to aid in the diagnosis of neurological conditions affecting speech.