摘要： 现有音频隐写软件大多基于最低比特位(LSB)隐写，对LSB隐写的检测算法研究具有重要意义。借鉴图像隐写检测中相关位平面和相邻像素相关性的分析思想，结合16 bit wav音频LSB的隐写特性，通过对音频数据相邻向量对的奇偶和大小分情况讨论，得出隐写会使相邻向量对的统计值增大的结论。通过经验阈值的设定，实现对音频LSB隐写的有效统计检测。理论和实验证明了该算法的正确性。

关键词: 音频隐写, 最低比特位, 相邻像素

Abstract: Most tools are based on Least Significant Bit(LSB), which makes the steganography on LSB a wide significance. Based on steganography ideas of bit planes and neighbor pixels correlation in image, with some inherent features of 16 bit coded wav audio and statistical changes by audio LSB, a conclusion of the statistical result in algorithm increasing is made by the neighbor vectors value and odd-even discussion. Theory and experimental results with a setting threshold value prove it is feasibility.

Key words: audio steganography, Least Significant Bit(LSB), neighbor pixels

中图分类号: 

• TP391