Research Of Algorithm Based On Audio Digital Watermarking

As an effective way of protecting the copyrights of digital products, Digital Watermark is the leading subject in information security fields internationally. As the popularization of computers and internet, digital audio products have been used more and more widely which dramatically improved the efficiency and precision of the transmission of information. Digital Watermark can effectively prevent individual or organization from copying, tampering or selling products without the permit of the authors, which significantly resolved the copyright problem of digital products.The use of Digital Audio Watermark is very similar to the use of Digital Watermark in other media, it mainly includes copyright protection, pirate trace and certification. Compared with Digital Image and Video Watermark, Digital Audio Watermark faces more challenges. On one hand, the Human Auditory System(HAS) is very sensitive to random noise,so the embeddable watermark is limited, and on the other hand, there are so many audio edit tools on the internet to change the structure of digital audios, which threatens the survival of watermark.This paper introduces the interrelated knowledge of Digital Watermark, analyzes the principles of some present audio watermark technologies. Using intuitive binary image as watermark information and audio data as the embedded object, based on detailed analysis of the features of audio watermark technologies and requirements, I analyzed some of the present algorithms and proposed directions of improvement.The self-adapting Digital Watermark algorithm in DCT domain and Digital Watermark in DWT domain based on modular arithmetic are also discussed in this paper. The former algorithm works like this: first reduce the dimension of the two-dimension binary image watermark, and do spread spectrum modulation with an m sequence, then cut the digital audio signals into sections and choose sections according to HAS to do the Discrete Cosine Transform (DCT), in DCT domain, quantify the modulated watermark signals and then embed them in DCT domain or amplitude to finish the imbed of the self-adapting watermark information. The latter algorithm compares the energy of the first two delicate components of each frame, according to the watermark bit "1" or "0", by unchanging or reducing the delicate components to embed watermark. When extracting watermark, the audio signals will be decomposed with relevant wavelet, and extract {0,1} sequences from low frequency factors to restore watermark signals. Experiments prove that neither of the two algorithms changed the perceptual quality of the original signals and both of them had the advantage of simplicity, small amount of calculation and good robustness and transparency. Both the algorithms can apply blind detection.