A dual cepstrum-based watermarking scheme with self-synchronization

This study presents a supplementary cepstrum-based scheme to achieve blind audio watermarking via the mean-value manipulation. The proposed scheme can work with the conventional statistical mean manipulation to double the watermark capacity under the same framework. As these two cepstral schemes affect the spectrum in low frequencies in a superposition manner, the developed watermarking system employs the conventional scheme to initiate a start and the new one to actually embed watermark bits. During the embedding phase, the rapid variations between frames caused by the watermarking process are avoided by adding a transition space. In addition, the watermarking system maintains data synchronization through the insertion of a sinusoid-like signal which keeps synchronization with the frame rate. Experimental results confirm the robustness of the proposed scheme against various attacks, including time-scaling and time-shifting that many other watermarking techniques failed to survive through.

► A supplementary cepstrum-based scheme to achieve blind audio watermarking. ► Can work with the conventional statistical mean manipulation. ► Rapid variations between frames are avoided by adding a transition space. ► Watermarking system maintains data synchronization through the insertion of a sinusoid-like signal. ► Robustness against various attacks, including time-scaling and time-shifting.