In this paper, a novel localized audio watermarking scheme based on signal to noise ratio （SNR） to determine a scaling parameter a is proposed. The basic idea is to embed watermark in selected high inflexion regions...In this paper, a novel localized audio watermarking scheme based on signal to noise ratio （SNR） to determine a scaling parameter a is proposed. The basic idea is to embed watermark in selected high inflexion regions, and the intensity of embedded watermarks are modified by adaptively adjusting a. As these high inflexion local regions usually correspond to music edges like sound of percussion instruments, explosion or transition of mixed music, which represent the music rhythm or tempo and are very important to human auditory perception, the embedded watermark is especially expected to escape the distortions caused by time domain synchronization attacks. Taking advantage of localization and SNR, the method shows strong robustness against common problems in audio signal processing, random cropping, time scale modification, etc.展开更多
基金Project supported by the National Natural Science Foundation of China (No. 60272031), the National Basic Research Program (973)of China (No. 2002CB312101), Zhejiang Provincial Natural ScienceFoundation (No. ZD0212), Doctorate Research Foundation of theState Education Commission of China (No. 20010335049), and theState Education Commission Foundation for Scholars Returned fromOverseas, China
文摘In this paper, a novel localized audio watermarking scheme based on signal to noise ratio （SNR） to determine a scaling parameter a is proposed. The basic idea is to embed watermark in selected high inflexion regions, and the intensity of embedded watermarks are modified by adaptively adjusting a. As these high inflexion local regions usually correspond to music edges like sound of percussion instruments, explosion or transition of mixed music, which represent the music rhythm or tempo and are very important to human auditory perception, the embedded watermark is especially expected to escape the distortions caused by time domain synchronization attacks. Taking advantage of localization and SNR, the method shows strong robustness against common problems in audio signal processing, random cropping, time scale modification, etc.
