Enregistré dans:
Détails bibliographiques
Auteurs principaux: Kopar, Serli, Rane, Roshan Prakash, Mychajliw, Christian, Federmann, Lydia, Eschweiler, Gerhard, Berg, Daniela, Gijsen, Sam, Perez-Toro, Paula Andrea, Ritter, Kerstin
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2605.27189
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916050485903360
author Kopar, Serli
Rane, Roshan Prakash
Mychajliw, Christian
Federmann, Lydia
Eschweiler, Gerhard
Berg, Daniela
Gijsen, Sam
Perez-Toro, Paula Andrea
Ritter, Kerstin
author_facet Kopar, Serli
Rane, Roshan Prakash
Mychajliw, Christian
Federmann, Lydia
Eschweiler, Gerhard
Berg, Daniela
Gijsen, Sam
Perez-Toro, Paula Andrea
Ritter, Kerstin
contents This study examines the relationship between speech representations and the hierarchical structure of cognitive assessment in mild cognitive impairment. Utilizing 5,754 German neuropsychological assessment recordings, we evaluate six cognitive tasks across three score levels: task, domain, and global levels. We compare hand-crafted acoustic features with self-supervised learning (SSL) embeddings. Results show that although SSL representations generally outperform hand-crafted features at lower levels, this trend reverses for MCI classification. Furthermore, task-specific constraints influence performance: tasks with greater response freedom exhibit performance dilution as hierarchical levels increase, suggesting ``specialist'' representations, whereas the performance of highly structured tasks increases toward higher levels, suggesting ``generalist'' representations. These findings show links between task constraints and assessment hierarchy in automated clinical speech analysis.
format Preprint
id arxiv_https___arxiv_org_abs_2605_27189
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Binary: Speech Representations Across the Cognitive Score Hierarchy
Kopar, Serli
Rane, Roshan Prakash
Mychajliw, Christian
Federmann, Lydia
Eschweiler, Gerhard
Berg, Daniela
Gijsen, Sam
Perez-Toro, Paula Andrea
Ritter, Kerstin
Computation and Language
Machine Learning
Sound
Audio and Speech Processing
Neurons and Cognition
This study examines the relationship between speech representations and the hierarchical structure of cognitive assessment in mild cognitive impairment. Utilizing 5,754 German neuropsychological assessment recordings, we evaluate six cognitive tasks across three score levels: task, domain, and global levels. We compare hand-crafted acoustic features with self-supervised learning (SSL) embeddings. Results show that although SSL representations generally outperform hand-crafted features at lower levels, this trend reverses for MCI classification. Furthermore, task-specific constraints influence performance: tasks with greater response freedom exhibit performance dilution as hierarchical levels increase, suggesting ``specialist'' representations, whereas the performance of highly structured tasks increases toward higher levels, suggesting ``generalist'' representations. These findings show links between task constraints and assessment hierarchy in automated clinical speech analysis.
title Beyond Binary: Speech Representations Across the Cognitive Score Hierarchy
topic Computation and Language
Machine Learning
Sound
Audio and Speech Processing
Neurons and Cognition
url https://arxiv.org/abs/2605.27189