DeMark: A Query-Free Black-Box Attack for Quality-Preserving Audio Watermark Removal
DeMark removes learning-based audio watermarks query-free, reaching 0.92–0.96 success while preserving speech quality.
DeMark is a query-free black-box attack that removes learning-based audio watermarks without access to the encoder, decoder, or detector and without clean-watermarked pairs. It suppresses subtle non-speech time-frequency artifacts through Diverse Artifact Learning and Adaptive Artifact Scaling under quality-preserving constraints. Across two speech datasets and four state-of-the-art watermarking methods, average attack success rates were 0.92 and 0.96, with higher perceptual quality than existing adaptive attacks. The authors say the results show a practical vulnerability and call for more robust watermark designs.
- No queries to the encoder, decoder, or detector are required.
- Two stages learn and scale non-speech time-frequency artifacts.
- Average success rates were 0.92 and 0.96 on two datasets.
- Perceptual quality stayed above compared adaptive attacks.
From the original article
1 min read at arxiv.orgAudio watermarking protects digital speech by embedding imperceptible signals for ownership verification and misuse tracing. However, the security of learning-based watermarking remains insufficiently understood under realistic adversarial removal, where attackers cannot access or query the watermark encoder, decoder, or detector.
By Weikang Ding · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.