A robust watermarking method for MPEG-4 SLS audio

Abstract

Audio lossless compression technology has been widely used to provide customers with higher quality music works. The protection of these works becomes very important. However, there are few digital watermarking methods specifically for lossless compressed audio, and the features of the lossless codec have not been used to improve the performance of watermarking. In this research, we propose a robust audio watermarking method to protect the music encoded by MPEG-4 SLS (Scalable Lossless Coding) format and improve the watermark security and robustness in two aspects. We design an adaptive selection method for watermark embedding position to enhance the security. The significant states combinations of consecutive scale factor-bands (SFB) are constructed as the feature data to determine the embedding target SFB. In this way, the embedding SFB adaptively changes with the audio signal variation, so the security improves. Based on the feature of the MPEG-4 SLS codec, we design new watermark embedding and extraction methods to improve the robustness. The watermark data are embedded into the target SFB. Meanwhile, the embedding is achieved through changing the probability distributions of Bit-plane coding results, which is stable when the audio faces attacks. The watermark data are extracted by comparing the probability distributions before and after embedding. The probability distributions are stable under attacks, so the robustness improves. The experimental data show that the proposed method can resist conventional signal processing attacks while guaranteeing the imperceptibility.