Audio watermarking robust against time-scale modification and MP3 compression

摘要

In continuation to earlier work where the problem of time-scale modification (TSM) has been studied [S. Xiang, J. Huang, R. Yang, Time-scale invariant audio watermarking based on the statistical features in time domain, in: Proceedings of the 8th Information Hiding Workshop (IH 2006), Lecture Notes in Computer Science, vol. 4437, Springer, Berlin, pp. 93-108] by modifying the shape of histogram extracted from the time domain, here we consider the additional ingredient of resisting common signal processing operations, such as MP3 compression. In other words, we study the problem of the watermark against both TSM and lossy compression. To this end, in this paper we transfer the shape of audio histogram in the time domain to the low-frequency subband by: (ⅰ) segmenting an audio signal into portions in reference to the bin width of the time-domain histogram, (ⅱ) concatenating the portions in each bin, and (ⅲ) DWT filtering the concatenation of the portions in each bin. The watermark is inserted by shaping the histogram after the DWT by reassigning population in each group of three consecutive bins. Extensive testing shows that in comparison with the time-domain scheme [S. Xiang, J. Huang, R. Yang, Time-scale invariant audio watermarking based on the statistical features in time domain, in: Proceedings of the 8th Information Hiding Workshop (IH 2006), Lecture Notes in Computer Science, vol. 4437, Springer, Berlin, pp. 93-108], the proposed DWT-based watermarking method is more robust to TSM, MP3 compression, etc.