A new multiple audio watermarking algorithm applying DS-CDMA

In this paper, a new multiple audio watermarking algorithm applying DS-CDMA is proposed. First, an orthogonal gold sequence set is generated via preferred-pair of m sequences, and then are used to spread spectrum of the two copyright watermarks. Before embedded into the DWT domain of the original audio in a fashion similar to DS-CDMA, the CDMA encoded watermark messages are mixed with a chaotic sequence for more security. This chaotic sequence is introduced as a secret embedding key so that the proposed scheme is robust to watermark counterfeiting. Blind recovery of each embedded watermark is achieved by using the fast fixed-point independent component analysis (FastICA) algorithm and the secret keys. Experiment results demonstrate that the watermarks with the proposed algorithm can be detected with low bit error rate (BER) and high security against common audio processing manipulations.