Multi-Dataset Co-Training with Sharpness-Aware Optimization for Audio Anti-spoofing
This addresses generalization issues in audio anti-spoofing for automatic speaker verification, offering a resource-efficient alternative to large models, though it is incremental as it builds on existing techniques.
The paper tackled the problem of audio anti-spoofing models lacking generalization across datasets by developing a compact model using multi-dataset co-training and sharpness-aware optimization, achieving competitive results with 4,000 times fewer parameters than large pre-trained models.
Audio anti-spoofing for automatic speaker verification aims to safeguard users' identities from spoofing attacks. Although state-of-the-art spoofing countermeasure(CM) models perform well on specific datasets, they lack generalization when evaluated with different datasets. To address this limitation, previous studies have explored large pre-trained models, which require significant resources and time. We aim to develop a compact but well-generalizing CM model that can compete with large pre-trained models. Our approach involves multi-dataset co-training and sharpness-aware minimization, which has not been investigated in this domain. Extensive experiments reveal that proposed method yield competitive results across various datasets while utilizing 4,000 times less parameters than the large pre-trained models.