The Impact of Audio Watermarking on Audio Anti-Spoofing Countermeasures

This paper presents the first study on the impact of audio watermarking on spoofing countermeasures. While anti-spoofing systems are essential for securing speech-based applications, the influence of widely used audio watermarking, originally designed for copyright protection, remains largely unexplored. We construct watermark-augmented training and evaluation datasets, named the Watermark-Spoofing dataset, by applying diverse handcrafted and neural watermarking methods to existing anti-spoofing datasets. Experiments show that watermarking consistently degrades anti-spoofing performance, with higher watermark density correlating with higher Equal Error Rates (EERs). To mitigate this, we propose the Knowledge-Preserving Watermark Learning (KPWL) framework, enabling models to adapt to watermark-induced shifts while preserving their original-domain spoofing detection capability. These findings reveal audio watermarking as a previously overlooked domain shift and establish the first benchmark for developing watermark-resilient anti-spoofing systems. All related protocols are publicly available at https://github.com/Alphawarheads/Watermark_Spoofing.git

Key Contributions

• Watermark-Spoofing Dataset: first benchmark dataset pairing diverse audio watermarking methods (6 handcrafted + 3 DNN-based) with ASVspoof and In-the-Wild anti-spoofing benchmarks at multiple watermark densities
• Empirical finding that watermarking consistently and proportionally degrades state-of-the-art spoofing detection EER, identifying it as a novel domain shift
• Knowledge-Preserving Watermark Learning (KPWL): a two-phase teacher–student adaptation framework that recovers watermarked-domain performance while preserving clean-domain spoofing detection accuracy

🛡️ Threat Analysis

Details

Similar Papers