Speech Articulation Dynamics Reveal New Depression Biomarkers
A study published on arXiv identifies novel depression biomarkers by analyzing the dynamical properties of tract variables—geometric features of speech articulator configuration. The approach quantifies predictability, complexity, and randomness in articulation using the Largest Lyapunov Exponent, Correlation Dimension, and Sample Entropy. Experiments on the Androids Corpus, which includes 64 speakers with clinician-diagnosed depression and 54 control speakers, show high Cliff's delta values, indicating effective discrimination between groups. This method explores previously unexamined aspects of articulatory processes in depression.
Key facts
- Study identifies depression biomarkers from tract variable dynamics.
- Tract variables describe geometric features of speech articulators.
- Quantifies predictability, complexity, and randomness in articulation.
- Uses Largest Lyapunov Exponent, Correlation Dimension, Sample Entropy.
- Experiments conducted on Androids Corpus.
- Androids Corpus includes 64 depressed and 54 control speakers.
- High Cliff's delta values indicate effective discrimination.
- Approach explores previously unexamined articulatory aspects in depression.
Entities
Institutions
- arXiv