Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kheir, Yassine El, Samih, Youness, Maharjan, Suraj, Polzehl, Tim, Möller, Sebastian
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909482107273216
author Kheir, Yassine El
Samih, Youness
Maharjan, Suraj
Polzehl, Tim
Möller, Sebastian
author_facet Kheir, Yassine El
Samih, Youness
Maharjan, Suraj
Polzehl, Tim
Möller, Sebastian
contents This paper conducts a comprehensive layer-wise analysis of self-supervised learning (SSL) models for audio deepfake detection across diverse contexts, including multilingual datasets (English, Chinese, Spanish), partial, song, and scene-based deepfake scenarios. By systematically evaluating the contributions of different transformer layers, we uncover critical insights into model behavior and performance. Our findings reveal that lower layers consistently provide the most discriminative features, while higher layers capture less relevant information. Notably, all models achieve competitive equal error rate (EER) scores even when employing a reduced number of layers. This indicates that we can reduce computational costs and increase the inference speed of detecting deepfakes by utilizing only a few lower layers. This work enhances our understanding of SSL models in deepfake detection, offering valuable insights applicable across varied linguistic and contextual settings. Our trained models and code are publicly available: https://github.com/Yaselley/SSL_Layerwise_Deepfake.
format Preprint
id arxiv_https___arxiv_org_abs_2502_03559
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detection
Kheir, Yassine El
Samih, Youness
Maharjan, Suraj
Polzehl, Tim
Möller, Sebastian
Audio and Speech Processing
Sound
This paper conducts a comprehensive layer-wise analysis of self-supervised learning (SSL) models for audio deepfake detection across diverse contexts, including multilingual datasets (English, Chinese, Spanish), partial, song, and scene-based deepfake scenarios. By systematically evaluating the contributions of different transformer layers, we uncover critical insights into model behavior and performance. Our findings reveal that lower layers consistently provide the most discriminative features, while higher layers capture less relevant information. Notably, all models achieve competitive equal error rate (EER) scores even when employing a reduced number of layers. This indicates that we can reduce computational costs and increase the inference speed of detecting deepfakes by utilizing only a few lower layers. This work enhances our understanding of SSL models in deepfake detection, offering valuable insights applicable across varied linguistic and contextual settings. Our trained models and code are publicly available: https://github.com/Yaselley/SSL_Layerwise_Deepfake.
title Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detection
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2502.03559