Audio Watermarking Pre-process Algorithm

Huan Li,Zheng Qin,Liping Shao
DOI: https://doi.org/10.1109/icebe.2009.30
2009-01-01
Abstract:The conventional audio watermarking methods have the following deficiencies: 1) the embedded watermarking signal bits are too less; 2) the embedded image watermarking is directly reduced to one-dimensional space and then embedded in audio signals, so the watermarking pre-process is too simple which reduce the security. To address these problems, a novel Audio watermark pre-process Algorithm is proposed. The algorithm is implemented in audio watermarking position space, which changes the dimension of coordinates and uses arbitrary matrices to scramble audio watermarking. The algorithm also can pre-process the host audio signals, which increase the security. Experiments show that the proposed algorithm has a wonderful pre-process performance, high imperceptibility, and robustness in enduring common attacks.
What problem does this paper attempt to address?