Feature-based self-synchronous audio watermarking technology

In order to get the robust performance for audio watermarking, a new idea called synchronous valley approach for hiding location is proposed for public audio transmission. As the audio masking principle, the computation is addressed in the low frequency coefficients in discrete wavelet transforming. Five features for seven rules are drawn from the coefficients in the self-synchronous approach. The secrets are firstly coded with BCH technology and then hidden behind the synchronous positions by the algorithm of quantization discrete wavelet transforming. Experimental results show that the synchronous valley approach is very effective when locating the secrets in the watermarking embedding and extracting process. Meanwhile, the watermarking technology is robust to the common attacks, such as MP3 coding, re-sampling, re-quantization, echo delay, Gaussian noise and cropping.