根据MP3文件的结构特点,提出一种无损的数字音频水印算法。该算法在MP3码流中嵌入水印,并不修改MP3文件的音频数据,仅利用文件中帧头的私有位作为标记,标示该帧主数据中的特定位置是否为水印信息。该方法不损伤音频质量,具有优良的不可感知性。接收端不需要无水印的原始音频信号,可顺序进行盲提取,而且计算量小,适合于流媒体应用。
This paper presents a digital watermarking scheme for MP3 files. Secret information is hidden in the MP3 stream without actually changing the audio data based on a detailed analysis of the MP3 file structure. The private bit in each frame head is used to indicate the presence of secret data so that blind extraction of the watermark is straightforward. Since the audio data are unchanged, the watermark does not introduce any distortion to the audio signal. Both embedding and extraction of the watermark are performed sequentially with low computational complexity, and therefore suitable for streaming media applications.