Automatic method of pause measurement for normal and dysarthric speech

Area of Science:

• Neurology
• Speech Pathology
• Computational Linguistics

Background:

• Friedreich's Ataxia (FRDA) is a rare inherited disease that causes progressive nervous system damage.
• Speech impairments are a common symptom of FRDA, but objective measurement methods are needed.
• Understanding speech alterations in FRDA can aid in diagnosis and monitoring.

Purpose of the Study:

• To develop and validate an automatic method for detecting and classifying pauses in conversational speech.
• To assess the impact of Friedreich's Ataxia on speech pause characteristics.
• To explore the utility of pause distributional analysis as a biomarker for FRDA.

Main Methods:

• Recorded 3-minute speech samples from 13 FRDA patients and 18 healthy controls.
• Utilized an automatic method to detect pauses from speech intensity contours.
• Applied Expectation-Maximization algorithm to fit pause durations with bimodal lognormal distributions.
• Analyzed pause distribution parameters (means, modes) between groups.

Main Results:

• Significant differences observed in pause distribution means between FRDA speakers and controls.
• FRDA speakers exhibited larger mean pause durations in both distribution modes.
• A disproportionately lower number of pauses were associated with the first mode in FRDA speakers.
• Pause distributional analysis revealed distinct patterns in FRDA speech.

Conclusions:

• Distributional analysis of speech pause duration is a promising method for measuring FRDA's effects on functional speech.
• This approach may serve as a sensitive, objective biomarker for Friedreich's Ataxia.
• Further research is warranted to validate these findings in larger cohorts.