Study On Characteristic-Based Digital Audio Watermarking

As the development of the Internet and digital technology, the transmission and storage of multimedia information become more and more convenient. Due to the openness of the Internet, people can get and distribute multimedia data easily. As a result, the advantages of the copyright owners are damaged seriously by the increasing problem of piracy. And as the development of software for digital signal treatment, the content of multimedia information can be deleted, modificated or replaced easily. So the copyright protection and content authentication of digital works are becoming more and more important.Based on the feature of audio, such as zero crossing rate and mean, two audio watermarking algorithms are proposed. The achieved results are listed as follow.(1) ZCR (Zero Crossing Rate) can be changed by audio signal process. Based on this feature, a scheme is proposed for content authentication. First, the audio signal is divided into frames and the ZCR of each frame is calculated. Second, the Hash operation is performed on the ZCR of each frame. The results are separated into a lot of groups. The XOR operation is performed between the groups, and the watermarking bits are gotten. Before embedding them into the audio signal using the quantization of mean, the watermarking bits should be encrypted by choas sequence. The temper attacks can be detected by compareing the extracted watermarking bits with the reconstructed bits. The results of experiment show that this scheme is sensitive to common signal processing and malicious attacks. And the tempered regions can be locatized effectively. The fasle alarm and fasle rejection probalility of the proposed algorithm are low.(2) The mean of audio is stable against the time scale modification. An adaptive robust watermark scheme is proposed based on this feature. The signs between the mean of audio and the DC coefficient of DCT (Discrete Cosine Transform) are same. So the watermarking bits are generated by the chaotic system of Logistic, and embedded into the audio by changing the sign of DC coefficient. The step of each frame is generated by the mean of the absolute value of samples and the amplitude range. Experimental results show that this scheme can resist the common signal processing, time scale modification and many attacks on Stirmark platform.