Channel Correlation Based Robust Audio Watermarking Mechanism for Stereo Signals

摘要

The music industry, especially the Internet-based platforms, is confronting the challenge of protecting the copyright. The traditional watermarking algorithms have limited robustness against the desynchro-nization attacks, such as time-invariant-pitch-scaling (TIPS) and pitch-invariant-time-scaling (PITS). Furthermore, only a very small part of the existing methods is designed for stereo signals, which is one of the most widely used formats in web-based music platforms. In this paper, we propose a robust watermarking mechanism for stereo signals for web-based applications. We first transform both of the left channel signal and the right channel signal into the frequency domain. Then the watermark bits are embedded by adjusting the Pearson correlation coefficient (PCC) between the frequency coefficients of the left channel and the right channel in the selected frequency band. Since TIPS and PITS can be approximated as stretching the signal in the frequency domain, the proposed channel correlation based watermarking mechanism is resilient to these attacks. Besides, the proposed method is also robust to other common attacks. Theoretical analysis and simulation results show the validity of the proposed method.