Sparse Autoencoders Make Audio Foundation Models more Explainable

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Mariotte, Théo, Lebourdais, Martin, Almudévar, Antonio, Tahon, Marie, Ortega, Alfonso, Dugué, Nicolas
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914205000531968
author Mariotte, Théo
Lebourdais, Martin
Almudévar, Antonio
Tahon, Marie
Ortega, Alfonso
Dugué, Nicolas
author_facet Mariotte, Théo
Lebourdais, Martin
Almudévar, Antonio
Tahon, Marie
Ortega, Alfonso
Dugué, Nicolas
contents Audio pretrained models are widely employed to solve various tasks in speech processing, sound event detection, or music information retrieval. However, the representations learned by these models are unclear, and their analysis mainly restricts to linear probing of the hidden representations. In this work, we explore the use of Sparse Autoencoders (SAEs) to analyze the hidden representations of pretrained models, focusing on a case study in singing technique classification. We first demonstrate that SAEs retain both information about the original representations and class labels, enabling their internal structure to provide insights into self-supervised learning systems. Furthermore, we show that SAEs enhance the disentanglement of vocal attributes, establishing them as an effective tool for identifying the underlying factors encoded in the representations.
format Preprint
id arxiv_https___arxiv_org_abs_2509_24793
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sparse Autoencoders Make Audio Foundation Models more Explainable
Mariotte, Théo
Lebourdais, Martin
Almudévar, Antonio
Tahon, Marie
Ortega, Alfonso
Dugué, Nicolas
Sound
Artificial Intelligence
Machine Learning
Audio and Speech Processing
Audio pretrained models are widely employed to solve various tasks in speech processing, sound event detection, or music information retrieval. However, the representations learned by these models are unclear, and their analysis mainly restricts to linear probing of the hidden representations. In this work, we explore the use of Sparse Autoencoders (SAEs) to analyze the hidden representations of pretrained models, focusing on a case study in singing technique classification. We first demonstrate that SAEs retain both information about the original representations and class labels, enabling their internal structure to provide insights into self-supervised learning systems. Furthermore, we show that SAEs enhance the disentanglement of vocal attributes, establishing them as an effective tool for identifying the underlying factors encoded in the representations.
title Sparse Autoencoders Make Audio Foundation Models more Explainable
topic Sound
Artificial Intelligence
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2509.24793