Related Experiment Video
Updated: Feb 20, 2026
12:43
A Protocol for Comprehensive Assessment of Bulbar Dysfunction in Amyotrophic Lateral Sclerosis ALS
Published on: February 21, 2011
36.1K
Predicting Intelligibility Gains in Dysarthria Through Automated Speech Feature Analysis
Annalise R Fletcher1, Alan A Wisler2, Megan J McAuliffe1
1Department of Communication Disorders, University of Canterbury, Christchurch, New Zealand.
Journal of Speech, Language, and Hearing Research : JSLHR
|October 28, 2017
Summary
Automated acoustic analysis effectively predicts speech intelligibility gains in individuals with dysarthria, offering a faster way to assess treatment effectiveness.
Area of Science:
- Speech-Language Pathology
- Acoustic Analysis
- Dysarthria Research
Background:
- Behavioral speech modifications yield variable results for dysarthria intelligibility.
- Previous research identified relationships between baseline speech and intelligibility gains with specific cues.
Purpose of the Study:
- To determine if automated acoustic assessments can predict intelligibility gains in speakers with dysarthria.
- To reexamine features related to intelligibility gains following speech modification cues.
Main Methods:
- Fifty speakers (43 with dysarthria) read a passage in habitual, loud, and slow modes.
- Automated acoustic measurements (long-term average spectra, envelope modulation spectra, Mel-frequency cepstral coefficients) were extracted from baseline speech.
- Intelligibility gains were statistically modeled and predicted using cross-validation.
Main Results:
- Statistical models accurately predicted intelligibility gains for unseen speakers.
- Automated acoustic features outperformed manual measures in predicting gains in the loud speech condition.
Conclusions:
- Automated acoustic analyses show promise for rapid assessment of speech treatment options.
- These measures may improve participant selection and outcomes in treatment studies for dysarthria.
