Research On Audio Watermarking Algorithms In Cepstrum Domain

The copyright protection of digital audios has been the focus of people since its extensive use.Digital watermarking is the crucial technology to solve copyright protection.Aiming at audio copyright protection problems,this thesis studies algorithms on digital audio watermarking.The main contents are as follows:(1)Aiming at scenes requiring high robustness such as copyright protection,cepstrum 2D mapping audio watermarking algorithm is proposed.The audio sequence is mapped to the two dimensional matrix.The watermark is embedded in cepstral coefficients of block matrix.Simulation results of various audios show the algorithm can resist re-sample,re-quantization,time domain amplification,cropping,low pass filtering,MP3 compression and Gaussian white noise attacks.The similarity of extracted watermarks is more than 99%on average under attacks.The proposed algorithm performs better than the wavelet and cepstrum domain combined method on robustness,and has better robustness in resample attack compared with the advanced norm space embedding method.The algorithm can be applied on speech,vocal and instrumental music.(2)Aiming at scenes requiring high imperceptibility,blind cepstrum time domain energy audio watermarking algorithm is proposed.Embedding segments is selected by computing time domain energies.The mean of cepstral coefficients of embedding segments is modulated to fall on different side of threshold,so as to embed the watermark.The algorithm can adaptively adjust the embedding threshold.A blind extraction is achieved.Simulation results show the proposed algorithm has good imperceptibility performance.Compared with other improved statistical mean manipulation method and wavelet domain imperceptibility watermarking algorithms,the proposed method has better imperceptibility.Except the insufficient robustness in re-quantization,the similarity of extracted watermark is more than 93%under the re-sample,time domain amplification,cropping,low pass filtering,MP3 compression and Gaussian white noise attacks.The algorithm can be applied on speech,vocal and instrumental music.