Based On Audio Information Hiding Algorithm

In the field of information security research, information hiding has drawn much attention. Internationally, the research on the information hiding is very active and many multinational corporations,colleges as well as research institutions specialise in it. How to work out the embedded algorithm, which has large steganographic capacity and high security, has become a common issue which many scholars concern on.In true-life, steganography can be applied in secret communication by military units or intelligence agencies, and digital watermarking can be applied in copyright protection of digital products. Therefore, the research on information hiding has significance both in thoretics and in practice.This thesis studies information hiding based on audio frequency.Based on the research of information hiding and HAS, this thesis mainly analyzed different kinds of embedded algorithm, particularly in classic LSB,DCT and DWT , and completed emulation.Finally, according to the demand of large steganographic capacity and high security, two audio hiding algorithms were brought forward.One of the algorithms is LSB algorithm,which combines adaptive energy and sliding window . The conventional LSB is very popular for its large capacity and easy application,but it has a shortage for it is detectable. The improved LSB is proposed in this thesis. Firstly, in improved algorithm, the short-time energy used to endpoint detection of audio recognition was calculated for fixed-size audio frame.Vocal or soundless, surd or sonant, can be distinguished by short-time energy. So, embedding data in the high energy frame can decrease the distortion.On the other hand, using sliding window when embedding data in frame have the effect just like spread spectrum.What's more ,breaking the equalization of the nonrandom property of LSB, this algorithm can resist statical steganalysis. Pilot study shows that the proposed algorithm is feasible, and has large capacity and high security.Another algorithm based on short-time energy is also proposed, which combines DCT and DWT. Firstly, select high energy frame by calculate short-time energy.Secondly, make two level wavelet decomposition to high energy frame. Finally, embed data to DCT coefficient by taking Discrete Cosine Transform to approximation components. IDCT and IDWT can reconstruct singals. Besides the advantage of introducing short-time energy above-mentioned, the approximate components of DWT has nice capability of resisting disturbance and the algorithm applying twice transform increases the complexity of detecting. Pilot study shows that this algorithm has fine performance in information hiding.