AI speech model classifies dysarthria severity in real-world recordings
A proof-of-concept AI system classified dysarthria severity in real-world recordings with 79% per-speaker F1. It showed strong agreement with two speech therapists.
Read the original →