The features and porformances of audio watermarking techinques currently being used are presented. In order to make a comparison, an adaptive content-based watermark embedding and extracting scheme in andio signals is proposed. The watermark embedding scheme is highiy related to the signal contents. This method can obtain a good tradeoff between inaudibility and robustness in embodded signals. First the original audio signal is divided adaptively into different kinds of segments according to its characteristics in the time domain. Then the embedding scheme is designed for each segment according to the propoities in the frequency domain. Finally a multiple-echo hopping techiuque is used to embed and extract watermarks. This method is compared with other current techinques of audio waterrmarking and results of the evaluation are presented.
展开▼