一种新的用于音频认证数字水印算法

摘要

为确保数字音频数据感知意义上的完整性,提出了一种新的音频认证水印算法.该算法把原始音频分帧,每帧分为两节,在第一节时域中嵌入同步码,将第二节进行DCT变换,计算邻帧关系向量,并将其作为水印信息嵌入其他帧第二节的DCT中频系数.认证时,计算邻帧关系误差,据事先设定的阈值来判断音频是否被窜改,若遭受窜改,可进行窜改定位.仿真实验表明,该算法在无原始水印参与的情况下,不仅可以实现音频媒体的完整性认证,而且窜改定位准确.%This paper proposed a new digital watermarking algorithm for audio authentication to protect the audio integrity in the perceptual sense. Firstly, it framed an original digital audio and then divided each frame into two parts. Secondly, with the spatial watermarking technique, embedded synchronization code into the first part. And then performed the DCT and calculated the AFRV( adjacent-frame relationship vector) used as the watermark information on the second part. Finally embedded the watermark bit into the DCT intermediate frequency coefficients of other frame' s second part. When authenticated, the algorithm could determine whether the audio had been malicious tampering by computing AFRE ( adjacent-frame relationship error) and according to the pre-set threshold without the help from the origin watermark. If the data were subjected to tampering, it could find out the data error localization. Simulation results show that the proposed algorithm can not only realize the audio media integrity authentication, but also accurately identify the tamper location.