Detection of hidden messages that are embedded in the audio signals using s-tools

Abstract

There are a lot of variants of the covert communication channels formation during transmitting an usual, non- suspicious data that are possible with using current steganography methods. But these forms of communications are the problem for the state security, as well as for various organizations because they could be used for an illegal activities. Accessibility, spreading and improvement of steganography software and technologies led to a significant increase of interest to the steganographic detection methods in a cover media, in particular, in an audio signals. In this paper was investigated the effectiveness of the audio steganalysis method that based on the "negative resonance" phenomenon for detecting the stego signals, formed by program S-Tools. The important relations of this method were found: the dependence of steganalysis accuracy on the amount of elements in the SVM training set; on the method of training set forming; on steganographic capacity of the stego signals. Also in this paper were found distinguishing statistics for the S-Tools and the roles of the stegokey and of each elements of the signals characteristic vectors have been evaluated. This research allows the improvement of the estimates of accuracy of the method and allows to find their effectiveness in the different steganalysis conditions.