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

2026-03-23

AI Summary

Cross-lingual detection of Parkinson’s disease–related dysarthria is hindered by the language dependency of speech representations and the scarcity of labeled pathological data in target languages. This work proposes a Language Shift (LS) approach that disentangles linguistic identity information at the level of self-supervised speech representations and aligns the distributions between source and target languages using centroid vectors estimated from healthy control speech. By mitigating the confounding effect of linguistic variation, the method significantly improves sensitivity and F1 scores in cross-lingual settings across Czech, German, and Spanish datasets. Furthermore, it demonstrates consistent performance gains in multilingual scenarios, highlighting its robustness and generalizability for cross-lingual pathological speech analysis.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Computer Vision: Language and Vision
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

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.

Problem

• cross-lingual
• dysarthria detection
• speech representations
• Parkinson's disease
• language-dependent structure

Innovation

• language shift
• self-supervised speech representations
• cross-lingual dysarthria detection
• representation adaptation
• Parkinson's disease