REAL-TIME STEGANALYSIS OF LSB-REPLACEMENT IN DIGITAL AUDIO STREAMS

摘要

Data hiding in the LSB of audio signals is an appealing steganographic method. This is due to the large volume of real-time production and transmission of audio data which makes it difficult to store and analyze these signals. Hence, steganalysis of audio signals requires online oper-ations. Most of the existing steganalysis methods work on stored media files. In this paper, we present a steganalysis technique that can detect the existence of embedded data in the least significant bits of natural audio samples. The algorithm is designed to be simple, accurate, and to be hardware implementable. Hence, hardware implementation is presented for the proposed algorithm. The proposed hardware analyzes the histogram of an incoming stream of audio signals by using a sliding window strategy without needing the storage of the signals. The algorithm is mathematically modeled to show its capability to accurately predict the amount of embedding in an incoming stream of audio signals. Audio files with different amounts of embedded data were used to test the algorithm and its hardware implementation. The exper-imental results prove the functionality and high accuracy of the proposed method.