Abstract: In order to protect the copyright of the digital audio effectively, combined with the important features of human auditory system and singular value decomposition, a digital audio zero-watermarking algorithm in the wavelet domain is presented. The algorithm uses chaotic sequence to encrypt the watermarking image, then chooses most suitably constructing the audio segment of zero-watermarking according to the local features of audio signals in time domain. The selected audio segments are carried on discrete wavelet transform, and low frequency components of the wavelet domain are extracted for singular value decomposition. The zero-watermarking is constructed using the largest singular value of the low frequency coefficients, and the digital audio copyright protection is realized. The experimental results show that the safety and the imperceptibility of the watermarking are very good, and the algorithm has good robustness for different styles of audio signals. It can effectively resist the attack of Gauss noise, low-pass filtering, resampling, requantization, shear and compression.

Key words: audio watermarking, zero-watermarking, singular value decomposition, discrete wavelet transform, adaptive

摘要： 为了对数字音频的版权进行有效的保护，结合人类听觉系统和奇异值分解的重要特性，提出了一种小波域数字音频零水印算法。用混沌序列对水印图像进行加密，根据音频信号的时域局部特征选择最适合于构造零水印的音频段，对选取的音频段进行离散小波变换，提取小波域的低频分量作奇异值分解，利用低频系数的最大奇异值构造零水印，实现数字音频的版权保护。实验结果表明，水印的安全性和不可感知性很好；对于不同风格的音频信号，算法均具有良好的鲁棒性，能够有效抵抗高斯噪声、低通滤波、重采样、重量化、剪切以及压缩等攻击。

关键词: 音频水印, 零水印, 奇异值分解, 离散小波变换, 自适应