A Robust, Digital-Audio Watermarking Method

Most of the previous audio-watermarking schemes are robust to common signal-processing attacks, but show severe problems when faced with desynchronization attacks. To solve the problems associated with these approaches, we propose an audio-watermarking scheme based on support-vector-machine (SVM) theory to protect against desynchronization attacks by using audio statistics characteristics and a synchronization code technique. Experimental results with SVM show that our proposed scheme is inaudible, robust against common signal processing, and robust against desynchronization attacks.