Steganalysis of Compressed Speech Based on Histogram Features

A steganalysis method for analysis-by-synthesis compressed speech based on the histogram features of the pulse position parameter is proposed in this paper. Speech compressed by G.729 coder is taken as an example to illuminate the algorithm. On the basis of analysis to the pulse position histogram characters of cover and stego speech, a new feature, histogram flatness is proposed. Furthermore, we use histogram flatness, the center of mass of histogram characteristic function, and histogram variance as classifying features, and employ SVM classifiers to classify speech into cover and stego data. Experiment results show the proposed algorithm has good performance for steganalysis of speech compressed by G.729 coders. This method can be also extended to compressed speech coded by other AbS-based coders.