中文摘要：

为保证DCT域水印的鲁棒性，研究水印信号的嵌入位置非常必要．对音频载体，尽管有许多文献提出将水印嵌在中频系数或对听觉效果重要的系数上，但始终缺乏一个统一的方法．本文在DCT噪声信号模型研究的基础上，提出了一种新的音频水印嵌入位置选择策略和一个应用该策略的自适应音频水印算法．通过定义DCT系数的噪声敏感度，建立了水印嵌入位置和嵌入水印后的音频信号的听觉感知性之间的关系，根据音频水印的不可听性的要求选择最优的嵌入位置，然后调节水印强度来满足鲁棒性的要求，从而最大程度地保持音频水印的不可听性和鲁棒性，为解决音频水印嵌入过程中不可听性和鲁棒性之间的矛盾提供了一种策略．应用该策略，本文提出了一个自适应音频水印算法．实验证实了算法的鲁棒性．

英文摘要：

Where should watermarks be embedded in DCT domain in order for the audio watermark to be in inaudible and robust？ Though many works in the literature agree that watermark should be embedded in the middle-frequency coefficients or perceptually significant components, how to determine the position embedded has no universal definition. A new embedding strategy and an algorithm applying the strategy are proposed for audio watermarking in DCT domain based on research of DCT noise modal. By defining a noise sensitivity vector,a math relationship between the watermark embeddable position and the audible quality of the audio file with watermark is set up. First the optimal position to be embedded based on the inaudible requirement is selected. Second the power coefficient of watermark is adjusted to satisfy the requirement of robustness. So the utmost need of inaudible and robust of audio watermarking scheme was filled. It is a good approach to settle the trade-off between the two requirements of audio watermarking ：inaudible and robustness. Experimental results demonstrate that the embedding strategy is effective and the watermarks thus generated are very robust.