Audio steganalysis based on auto-correlation property of sdteganographic tools
Fei Qing Wu · Journal of Communications · 2006
After investigating the changes of audio statistics features, auto-correlation property in steganographic soft-ware tools was found when different random message was embed in same audio with same steganographic tool. Based on this property, a novel steganalysis scheme was proposed which was implemented by using support vector machine (SVM) to classify stego-audio from clear audio with the audio and its linear prediction statistics features. This scheme can not only detect the existence of hidden messages, but also identify the tools used to hide them. As shown by the experimental results, the proposed method can be very effectively used to detect hidden messages embedded by Hide4PGP, Stegowav and Steghide.