Adapting Self-Supervised Speech Representations for Cross-lingual Dysarthria Detection in Parkinson's Disease

TL;DR

This paper introduces a novel representation-level language shift method to improve cross-lingual dysarthria detection in Parkinson's disease by reducing language-dependent biases in speech representations.

Contribution

It proposes a centroid-based language shift technique to align self-supervised speech representations across languages, enhancing dysarthria detection accuracy in cross-lingual scenarios.

Findings

LS improves sensitivity and F1 scores in cross-lingual detection

LS reduces language identity in speech embeddings

Method yields consistent gains in multilingual settings

Abstract

The limited availability of dysarthric speech data makes cross-lingual detection an important but challenging problem. A key difficulty is that speech representations often encode language-dependent structure that can confound dysarthria detection. We propose a representation-level language shift (LS) that aligns source-language self-supervised speech representations with the target-language distribution using centroid-based vector adaptation estimated from healthy-control speech. We evaluate the approach on oral DDK recordings from Parkinson's disease speech datasets in Czech, German, and Spanish under both cross-lingual and multilingual settings. LS substantially improves sensitivity and F1 in cross-lingual settings, while yielding smaller but consistent gains in multilingual settings. Representation analysis further shows that LS reduces language identity in the embedding space, supporting the interpretation that LS removes language-dependent structure.