Saved in:
Bibliographic Details
Main Authors: Herron, Felix, Hjuler, Maja, Rossato, Solange, Allauzen, Alexandre, Portet, François
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.18249
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910148793991168
author Herron, Felix
Hjuler, Maja
Rossato, Solange
Allauzen, Alexandre
Portet, François
author_facet Herron, Felix
Hjuler, Maja
Rossato, Solange
Allauzen, Alexandre
Portet, François
contents Speech encoder models are known to model members of some speaker groups (SGs) better than others. However, there has been little work in establishing why this occurs on a technological level. To our knowledge, we present the first layerwise fairness analysis of pretrained self-supervised speech encoder models (S3Ms), probing each embedding layer for speaker identification (SID) automatic speech recognition (ASR). We find S3Ms produce embeddings biased against certain SGs for both tasks, starting at the very first latent layers. Furthermore, we find opposite patterns of layerwise bias for SID vs ASR for all models in our study: SID bias is minimized in layers that minimize overall SID error; on the other hand, ASR bias is maximized in layers that minimize overall ASR error. The inverse bias/error relationship for ASR is unaffected when probing S3Ms that are finetuned for ASR, suggesting SG-level bias is established during pretraining and is difficult to remove.
format Preprint
id arxiv_https___arxiv_org_abs_2604_18249
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Where Do Self-Supervised Speech Models Become Unfair?
Herron, Felix
Hjuler, Maja
Rossato, Solange
Allauzen, Alexandre
Portet, François
Computation and Language
Speech encoder models are known to model members of some speaker groups (SGs) better than others. However, there has been little work in establishing why this occurs on a technological level. To our knowledge, we present the first layerwise fairness analysis of pretrained self-supervised speech encoder models (S3Ms), probing each embedding layer for speaker identification (SID) automatic speech recognition (ASR). We find S3Ms produce embeddings biased against certain SGs for both tasks, starting at the very first latent layers. Furthermore, we find opposite patterns of layerwise bias for SID vs ASR for all models in our study: SID bias is minimized in layers that minimize overall SID error; on the other hand, ASR bias is maximized in layers that minimize overall ASR error. The inverse bias/error relationship for ASR is unaffected when probing S3Ms that are finetuned for ASR, suggesting SG-level bias is established during pretraining and is difficult to remove.
title Where Do Self-Supervised Speech Models Become Unfair?
topic Computation and Language
url https://arxiv.org/abs/2604.18249