基于小波变换的双同步音频水印算法

针对同步攻击的问题，提出了一种改进的基于小波变换的双同步音频水印算法，该算法同时借助外同步和自同步2种思想，采用m序列作为外同步信号，谱熵作为自同步信号，将m序列嵌入音频信号的时域来定位水印嵌入的起始位置，并把紧接着嵌入m序列后的一段音频数据作小波变换，提取低频系数分段求其谱熵，利用谱熵的最大值来定位水印嵌入的具体位置，最后把水印嵌入到最大谱熵值对应的小波系数中。仿真实验表明，此方法在能抵抗一般信号处理攻击的基础上，也能抵抗像延时、随机剪切、抖动和时间伸缩 (time-scale modification,TSM)的同步攻击，具有较好的鲁棒性。

Double synchronous audio watermark algorithm based on DWT

A double synchronous audio watermark algorithm based on DWT is proposed in this paper, which uses external synchronization and intra-stream synchronization. We use m-sequence as the external synchronization signal and spectral entropy as the intra-stream synchronization signal, which can locate the original position of the watermark being embedded by the m-sequence, then we apply DWT to the a section of audio data embedded with m-sequence, extract the subsection of the low frequency coefficient and compute its spectral signal. After that we get the peak value of the spectral entropy, with which we can locate the specific spot of the embedded watermark. Finally we embed the watermark into the wavelet coefficient corresponding to the peak value of the spectral entropy. Simulations show that our algorithm is of good robustness and can not only resist normal signal processing attacks but can also combat image delay, cropping resistant, jitter and synchronous attack of TSM.