A digital audio watermarking scheme based on blind source separation

An effective digital audio watermarking scheme based on blind source separation (BSS) for blind watermark detection is proposed in this paper. In the embedding procedure, the original watermark is modulated by chaotic sequence to generate an embedding watermark. Then watermarked audio signal is combined by two signals which are mixed by original audio signal and embedding watermark according to BSS mixing model. In the detecting procedure, watermarked audio signal is split into two signals first, and then BSS technique is applied to them to obtain the separated embedding watermark. At last, we can determine the watermark presence by correlation values between the separated embedding watermark and embedding watermark. Experimental results show the effectiveness of the proposed method.