Abstract: In this paper，the encoding method and the frame structure of MP3 are expounded in detail，a watermark algorithm is proposed based on the scale factor in MP3 frame.By introducing the step of frame altering analysis，the algorithm can realize that watermarks is embedded adaptively.By modifying threshold [SNRlim]（the ratio of signal to noise），the algorithm can meet different requirements of some system，that is what enable the algorithm to have a bigger flexibility.Based on expense of some embedding positions，the watermark can be embedded with self-synchronization.The experiment results show this algorithm can get a good audio effect with various types of audio.

Key words: audio watermark, compress domain, MP3, scale factor

摘要： 针对MP3编码方式和帧结构进行分析，利用帧中的比例因子提出了一种压缩域水印算法。算法通过引入帧变化分析步骤，很好地实现了水印的自适应嵌入。对于不同的系统要求，还可以通过修改门限信噪比的值[SNRlim]按需进行水印嵌入，使算法具有更大灵活性。同时在牺牲一定的嵌入空间前提下，实现了水印信息的自同步。实验表明，对于不同风格的音乐，该水印算法具有很好的不可听性。

关键词: 音频水印, 压缩域, MP3, 比例因子