Research On Digital Audio Watermarking Algorithm Based On Lifting-wavelet Transform And Chaotic Keys

Digital audio watermarking is considered as an effective way to resolve the copyright protection of digital audio products. It embeds the logo information into the original audio at the premise of not affecting the use of audio signal. Copyright and integrity of audio file can be confirmed through watermark detection.This paper focuses on an audio watermarking algorithm research based on lifting wavelet transform and chaotic encryption. Main work includes as follows:Giving some fundamental knowledge related to digital audio watermark and analyzing the status quo of digital audio watermark;Some of the existing audio watermarking algorithms did not take enough account of human auditory system, in allusion to this; we decided the embedding strength in terms of masking threshold values adaptively which can ensure that the embedded information can not be perceived by human ears;Audio watermarking algorithm using classical wavelet transform must be calculated by convolution, which needs big and complex computations and takes a lot of time. To make our algorithm more effective we introduce wavelet-lifting method;As we know that the correlation function of m sequence can be periodic. This can reduce the safety of digital audio watermark. Therefore we employ modified logistic chaos mapping to encrypt the sequences to be hidden which can feature better security performance;Most detection of present algorithms was dependent on the exact alignment of signal of origin and to be detected. Once the position was lost by random cropping, then the right watermark cannot be detected simply. Before embedding watermark, we add a synchronization code which makes the watermark self-synchronized and embed watermark into the statistical expectation of sub-band coefficients in lifting-wavelet domain which can make our algorithm more robust against synchronization attacks;Simulation experiments were under 5 different types of music, comparisons with traditional schemes based on DFT, DCT, DWT were made as well. Experimental results show that our method has better transparency and security performance, and it is more robust against various common signal processing attacks;At last, the disadvantage and unresolved questions of the proposed algorithm were given and we discussed the direction of further study.