Improving The Performance of Spread Spectrum Audio Watermarking by Modified Correlation Analysis

Digital Watermark is a visible or invisible mark inserted into the digital multimedia data so that it can be detected in the later stage of evidence of rightful ownership. Most watermarking schemes focus on image and video watermarking, and a few are concerned with audio. Audio watermarking is the process of embedding a watermark signal into an audio signal. This process is difficult because of the high sensitivity of Human Auditory System (HAS). Spread Spectrum watermarking techniques are widely used to watermark audio data due to its robustness and watermark imperceptiveness. As the spread spectrum watermark detection techniques are based on correlation analysis, compared to some threshold value, there is always a probability of non detection of a watermark, even if it is present, termed as false negative probability. Analogous to it is the probability of detection of a watermark, in a host signal which is actually non-watermarked, termed as the false positive probability. In this work, a novel method of correlation analysis is proposed to improve the watermark embedding and its detection at the receiver.