Enhancing Multi-Corpus Training in SSL-Based Anti-Spoofing Models: Domain-Invariant Feature Extraction

TL;DR

This paper introduces the IDFE framework that uses multi-task learning and gradient reversal to extract domain-invariant features, improving the robustness of speech spoofing detection across multiple datasets.

Contribution

It proposes a novel domain-invariant feature extraction method for multi-corpus training in SSL-based anti-spoofing models, addressing dataset bias issues.

Findings

Reduces equal error rate by 20% across four datasets

Addresses dataset-specific bias in multi-corpus training

Improves robustness of spoofing detection models

Abstract

The performance of speech spoofing detection often varies across different training and evaluation corpora. Leveraging multiple corpora typically enhances robustness and performance in fields like speaker recognition and speech recognition. However, our spoofing detection experiments show that multi-corpus training does not consistently improve performance and may even degrade it. We hypothesize that dataset-specific biases impair generalization, leading to performance instability. To address this, we propose an Invariant Domain Feature Extraction (IDFE) framework, employing multi-task learning and a gradient reversal layer to minimize corpus-specific information in learned embeddings. The IDFE framework reduces the average equal error rate by 20% compared to the baseline, assessed across four varied datasets.