Adapting Self-Supervised Speech Representations for Cross-lingual Dysarthria Detection in Parkinson's Disease
arXiv:2603.22225v3 Announce Type: replace 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
The proliferation of self-supervised learning for speech and the increasing demand for accessible diagnostic tools for neurological conditions are converging to enable these advanced applications.
This development allows for more accurate and cross-lingual detection of diseases like Parkinson's, broadening the reach of early diagnosis and intervention, particularly in underserved linguistic communities.
The ability to adapt speech representations across languages significantly improves the viability of AI-driven diagnostic tools for conditions manifesting in speech, reducing dependency on vast language-specific datasets.
- · AI healthcare diagnostics
- · Patients with Parkinson's Disease
- · Speech technology developers
- · Elderly care sectors
- · Traditional, language-dependent diagnostic methods
Improved early detection rates for neurological diseases like Parkinson's globally.
Accelerated development and adoption of AI-powered diagnostic tools in diverse linguistic and geographic regions.
Potentially, the establishment of universal speech-based diagnostic platforms that transcend language barriers for a range of conditions.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL