Study On Content-Based Watremarking For Audio Authentication

In recent years, with the rapid development of science and technology, multimedia has been widely used by people. At the same time, the multimedia information security has become the focus of attention. Digital watermarking as a very effective means to guarantee the multimedia security, has been gradually applied in image, audio and video for the copyright protection, content authentication and integrity protection.Recently, the digital speech is widely used in telephone communication, court probe and so on, because its characteristics are easy to edit, copy, store, and transmit. However, how to ensure authenticity and integrality of speech is a practical and important subject. Digital watermarking has been widely used in the speech signal, that means the digital audio watermarking technology is relatively mature, but there are still some problems, Based on independent component analysis technology, this paper mainly focuses on designing speech watermarking algorithm for content authentication and tamper recovery, the main work is given as follows:(1) In order to overcome the weakness of tamper localization cannot resist to synchronization attacks and cannot recover the tampered area, a recovery watermarking algorithm based on ICA was designed for speech content integrality and recovery. As for the target to recover the tampered speech, a feature-based watermark is generated according to its content and mixed with the original speech. On the other hand, a watermarking which is generated based on the characteristics of the energy and the index of the frame is hidden in the mixed audio frames. Experimental results show that the modified signal not only can be effectively detected, but also can be successfully recovered with acceptable quality.(2) Many previous fragile watermarking schemes adopt binary images as watermarks, which reduce the security of watermarking systems. On the other hand, the content-based or feature-based second generation digital watermarking has limited applicability, and its partial feature point may be damaged by watermarking operations and various common signal processing. To overcome these problems, a novel watermarking algorithm which is based on robust feature points for speech content authentication is proposed. In the proposed scheme, first, the robust feature points of each audio frame are detected. Then the location of the robust feature points is used to generate watermarking. After that, the each audio frame is performed with discrete wavelet transform (DWT) and discrete cosine transform (DCT), and finally the watermarking bits are embedded into the hybrid domain. Experimental results show that the proposed scheme is inaudible. Furthermore, the tamper detection ability and tolerance robustness to common signal processing operations are excellent.