Mixing-attack-proof Randomized Embedding Audio Watermarking System

Audio mixing greatly degrades the watermarking security. Consequently, it is of importance to introduce mixing-attack-proof audio watermarking algorithms. This article investigates audio mixing attack. A randomized embedding spread spectrum audio watermarking algorithm is accordingly proposed to improve the watermark immunity to mixing attacks. First, balanced modulation is introduced to get rid of the host signal interference and improve the correlation score stability. With more stable correlation scores, a randomized pseudo-noise embedding strategy is proposed to decrease the interference among different mixed components and bit error rate. To further improve the synchronization performances, a multiple synchronization strategy which simultaneously synchronizes different mixed components is also proposed. The experimental results indicate that the proposed algorithm shows a strong level immunity to mixing attacks and it can resist the mixing attack with up to 5 mixed components.