From Black Box to Biomarker: Sparse Autoencoders for Interpreting Speech Models of Parkinson's Disease

Open in new window