Saved in:
Bibliographic Details
Main Authors: Chiu, Aemon Yat Fei, Fung, Kei Ching, Li, Roger Tsz Yeung, Li, Jingyu, Lee, Tan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.05310
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917315483795456
author Chiu, Aemon Yat Fei
Fung, Kei Ching
Li, Roger Tsz Yeung
Li, Jingyu
Lee, Tan
author_facet Chiu, Aemon Yat Fei
Fung, Kei Ching
Li, Roger Tsz Yeung
Li, Jingyu
Lee, Tan
contents Enhancing explainability in speech self-supervised learning (SSL) is important for developing reliable SSL-based speech processing systems. This study probes how speech SSL models encode speaker-specific information via a large-scale probing analysis of 11 models, decomposing identity into acoustic, prosodic, and paralinguistic attributes. The results confirm a general hierarchy wherein initial layers encode fundamental acoustics and middle layers synthesise abstract traits. Crucially, the consensus that final layers purely abstract linguistic content is challenged. It is discovered that larger models unexpectedly recover speaker identity in their deep layers. Furthermore, the intermediate representations of speech SSL models are found to capture dynamic prosody better than specialised speaker embeddings. These insights decode the complex internal mechanics of SSL models, providing guidelines for selecting interpretable and task-optimal representations.
format Preprint
id arxiv_https___arxiv_org_abs_2501_05310
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Large-Scale Probing Analysis of Speaker-Specific Attributes in Self-Supervised Speech Representations
Chiu, Aemon Yat Fei
Fung, Kei Ching
Li, Roger Tsz Yeung
Li, Jingyu
Lee, Tan
Audio and Speech Processing
Sound
Enhancing explainability in speech self-supervised learning (SSL) is important for developing reliable SSL-based speech processing systems. This study probes how speech SSL models encode speaker-specific information via a large-scale probing analysis of 11 models, decomposing identity into acoustic, prosodic, and paralinguistic attributes. The results confirm a general hierarchy wherein initial layers encode fundamental acoustics and middle layers synthesise abstract traits. Crucially, the consensus that final layers purely abstract linguistic content is challenged. It is discovered that larger models unexpectedly recover speaker identity in their deep layers. Furthermore, the intermediate representations of speech SSL models are found to capture dynamic prosody better than specialised speaker embeddings. These insights decode the complex internal mechanics of SSL models, providing guidelines for selecting interpretable and task-optimal representations.
title A Large-Scale Probing Analysis of Speaker-Specific Attributes in Self-Supervised Speech Representations
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2501.05310