aiqbench analysis
This research evaluates cross-lingual generalization of speech foundation models for Parkinson's disease severity assessment, highlighting that while transfer learning is feasible, performance remains sensitive to dataset properties and preprocessing strategies. The study underscores the critical need for robust feature extraction and explainable methods to handle inter-speaker variability in clinical settings. Ultimately, this work advances the application of AI in motor speech disorders but emphasizes that current models require further adaptation to be reliable across diverse linguistic contexts.
Cross-Lingual Parkinson's Disease Severity Assessment Using Pre-trained Speech Embeddings: A Multi-Class Evaluation
arXiv:2609.20875v1 Announce Type: cross Abstract: Parkinson's disease (PD) often manifests through speech impairments, facilitating accessible, non-invasive, and cost-effective severity assessment for early diagnosis and progression tracking. Despite advances in speech foundation models (SFMs), their cross-lingual generalization for PD severity multi-class classification remains underexplored due to limited labeled data, a lack of explainable methods and variability across languages and datasets. In this work, we evaluate pre-trained embeddings from four state-of-the-art open-source SFMs across three datasets in zero-shot and k-shot cross-lingual settings for multi-class PD severity assessment. Our results show that pre-trained speech embeddings enable meaningful cross-lingual transfer, although performance is sensitive to dataset properties, preprocessing, and adaptation strategy. Misclassifications under these conditions related to inter-speaker variability and atypical speech patterns highlight the need for more robust feature extraction and modeling for PD severity assessment while emphasizing the importance of explainability for reliable clinical insights.
Abstract:Parkinson's disease (PD) often manifests through speech impairments, facilitating accessible, non-invasive, and cost-effective severity assessment for early diagnosis and progression tracking. Despite advances in speech foundation models (SFMs), their cross-lingual generalization for PD severity multi-class classification remains underexplored due to limited labeled data, a lack of explainable methods and variability across languages and datasets. In this work, we evaluate pre-trained embeddings from four state-of-the-art open-source SFMs across three datasets in zero-shot and k-shot cross-lingual settings for multi-class PD severity assessment. Our results show that pre-trained speech embeddings enable meaningful cross-lingual transfer, although performance is sensitive to dataset properties, preprocessing, and adaptation strategy. Misclassifications under these conditions related to inter-speaker variability and atypical speech patterns highlight the need for more robust feature extraction and modeling for PD severity assessment while emphasizing the importance of explainability for reliable clinical insights.
Accepted and published at IEEE SLT 2026 - IEEE Spoken Language Technology 2026. OneVoice-MSD 2026: Multilingual Speech Technologies for Motor Speech Disorders. this https URL Please cite the conference version
Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD)
arXiv:2609.20875 [eess.AS]
(or arXiv:2609.20875v1 [eess.AS] for this version)
https://doi.org/10.48550/arXiv.2609.20875 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 194 words
Related coverage
Forbes Asia 100 To Watch 2026
forbes.com · Aug 24, 2026
[AINews] The Future of Latent Space
Latent Space · Sep 25, 2026
[AINews] Opus 5.5 is good at explainer videos
Latent Space · Sep 29, 2026
How to build a cancer vaccine, and whether they will work this time
LessWrong · Jun 8, 2026