AUDIO STEGANOGRAPHY BASED ON SIGNALMODULATION IN WAVELET DOMAIN

Abstract

In this paper, an audio steganography system is proposed to hide a sequence of
binary digits (bits) in digital audio data. The hiding method is based on applying
amplitude modulation technique on wavelet transform coefficients. The embedding
schema implies partitioning the audio signal (cover) into a number of non overlap
slices, then the average of AC energy of all slices are examined; and the slices have
the highest energy are considered as good hosts for secret bit. The shift in
coefficients values, caused by embedding process, may cause modification in the
energy status of the host slice; this problem is avoided by applying an algebraic
manipulation so as to ensure that the extraction detector can to correctly identify the
audio host slices. Bits insertion is done by transforming the selected host slices into
wavelet domain using the biorthogonal tap (9\7) wavelet filters. Then, some of the
highly valued coefficients of the high frequency subband are quantized to use them
to host the secret bits {0, 1} by adding or subtracting a modulation step. The test
results indicated that a perfect retrieval of secret bits could attained, while the hiding
rate is small and the quality of stego objects is high.