BUT Systems for WildSpoof Challenge: SASV in the Wild

TL;DR

This paper introduces a robust SASV framework for the WildSpoof Challenge, combining diverse self-supervised audio models, a novel feature augmentation strategy, and fusion techniques to improve spoofing detection in varied environments.

Contribution

We propose a novel SASV system integrating multiple self-supervised front-ends, a distribution uncertainty-based augmentation, and fusion methods to enhance spoofing robustness.

Findings

Achieved lower a-DCFs and EERs compared to baseline systems.

Demonstrated effectiveness of distribution uncertainty augmentation.

Improved robustness against unseen neural vocoders and recording conditions.

Abstract

This paper presents the BUT submission to the WildSpoof Challenge, focusing on the Spoofing-robust Automatic Speaker Verification (SASV) track. We propose a SASV framework designed to bridge the gap between general audio understanding and specialized speech analysis. Our subsystem integrates diverse Self-Supervised Learning front-ends ranging from general audio models (e.g., Dasheng) to speech-specific encoders (e.g., WavLM). These representations are aggregated via a lightweight Multi-Head Factorized Attention back-end for corresponding subtasks. Furthermore, we introduce a feature domain augmentation strategy based on Distribution Uncertainty to explicitly model and mitigate the domain shift caused by unseen neural vocoders and recording environments. By fusing these robust CM scores with state-of-the-art ASV systems, our approach achieves superior minimization of the a-DCFs and EERs.