Metraipolis › archive › dc_fc87d29de56e
Exploiting Speech LLM Representations for Multilingual and Cross-Lingual Parkinson's Disease Detection — arXiv:2609.14431
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.14431
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T02:58:53.564Z
- SHA-256 of the archived text
a688484b9ac353ba83bc8be0ebd2444b7fee83ff129523fbb1c16746a1c122ea- Stable URL of this page
https://metraipolis.com/documento/dc_fc87d29de56e
How to cite
Exploiting Speech LLM Representations for Multilingual and Cross-Lingual Parkinson's Disease Detection — arXiv:2609.14431 — arXiv (en), https://arxiv.org/abs/2609.14431 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fc87d29de56e (Metraipolis archive dc_fc87d29de56e, sha256 a688484b9ac353ba83bc8be0ebd2444b7fee83ff129523fbb1c16746a1c122ea)
BibTeX
@misc{dcfc87d29de56e,
title = {Exploiting Speech LLM Representations for Multilingual and Cross-Lingual Parkinson's Disease Detection — arXiv:2609.14431},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.14431},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fc87d29de56e; sha256 a688484b9ac353ba83bc8be0ebd2444b7fee83ff129523fbb1c16746a1c122ea},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.14431. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Speech Large Language Models (Speech LLMs) have shown strong performance across diverse tasks, yet their utility for pathological speech analysis remains underexplored. In this work, we investigate the effectiveness of internal representations from encoder and decoder components of Speech LLMs for Parkinson's Disease (PD) detection across multilingual and cross-lingual settings. Our findings reveal that encoder representations consistently outperform their decoder counterparts in most models and settings and that pathological cues may be progressively attenuated as audio representations are projected into the language model space. We further show that generative outputs are less reliable for clinical tasks compared to internal representations. To leverage information spread across multiple layers, we propose a Squeeze-and-Excitation (SE)-based dynamic layer aggregation framework, which surpasses best-layer selection in multiple experiments, suggesting that PD-relevant acoustic cues are distributed across transformer layers rather than concentrated in one. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in eess.AS). Categorie arXiv: Sound (cs.SD) ; Audio and Speech Processing (eess.AS). Pagina del lavoro: https://arxiv.org/abs/2609.14431 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T02:45:36.509Z dalla pagina https://arxiv.org/catchup/eess.AS/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).