Layer-Aware Early Fusion of Acoustic and Linguistic Embeddings for Cognitive Status Classification
Krystof Novotny, Laureano Moro-Velázquez, Jiri Mekyska · 2026
Read Layer-Aware Early Fusion of Acoustic and Linguistic Embeddings for Cognitive Status Classification by Krystof Novotny, Laureano Moro-Velázquez, Jiri Mekyska (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-23004.txt.
Description
Krystof Novotny, Laureano Moro-Velázquez, and Jiri Mekyska use speech to judge cognitive status. Speech carries both sound patterns and word patterns that reflect decline in thinking. Models that use only one of those sides miss part of the picture.
Their study fuses sound and transcript features early, and checks which encoder layers help most. Layer-aware fusion improved classification of cognitive status in their tests of the approach. A speech-based check is cheap and non-invasive, which suits early screening in clinics. Attention to encoder layer depth is the study's twist, because early and late layers carry different kinds of speech information.
Catalog record
| CREATOR | Krystof Novotny, Laureano Moro-Velázquez, Jiri Mekyska |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-23004