Embedding visually recognizable watermarks into digital audio signals

This paper presents a novel watermarking technique for digital audio signals. A visually recognizable binary image served as a watermark is embedded into an audio signal in discrete cosine transform domain. In order to enhance robustness and make good use of masking characteristics of the audio signal, the techniques of pseudo-random permutation and energy-based permutation are used during watermarking. The experimental results show that the proposed watermarking technique is robust to several signal manipulations, such as filtering, compression, and so on.