An Audio Steganography based on Twi-DWT and Audio-extremum Features

Audio steganography, hiding the secret audio into a host audio, is one of the research difficulties in Computer Network Information Security. It is employed to hide the secret spreading, protect security and prevent illegal tampering. In the paper, a novel steganography based on twice discrete wavelet transform (Twi-DWT) and extremum features is proposed. Firstly, in order to improve the embedding capacity, we designed a second-order difference to extract the extremum sequence of the secret signal as the covert information. Next, the host signal was decomposed twice by DWT to extract a security sequence, which achieves a balance between transparency and robustness. Finally, the security signal is replaced by the covert information. The content and waveform of the stego audio are similar to that of the original host audio, which indicates our method has a good transparency. Through the simulations, it is verified that our algorithm is superior to classical ways on robustness and embedding capacity.