Study On Hierarchical Steganography For Low Bit-rate Speech Streams

With the development and popularization of the Internet and communication technology, Voice over IP(Vo IP) is becoming increasingly popular. With the advantage of real-time characteristic, universality and huge volume of data hiding, the Vo IP speech streams have caused widespread concern of researchers, and information hiding based on Vo IP has also become a new hotspot. In this thesis, we pick up the low bit-rate speech streams that popularly exist in Vo IP scenarios as the carrier to hide messages, and mainly focus on unequal and hierarchical steganographic methods in accordance with the redundancies and non-equilibrium of adopted carriers. The main contributions of our study are as follows.(1) The available bits of low-rate speech frames for information hiding are analysed by determining the redundancies of low bit-rate speech streams. The least significant bits of speech frames are established based on comprehensive tests by inverting each bit respectively in speech frames. Moreover, the objective speech quality assessment standards, including perceptual speech quality evaluation criteria and the signal to noise ratio, are employed to assess the embedding impact of all bits in each speech frame. Taking ITU G.729 a that is widely used in Vo IP scenarios as an example, the redundancies of speech frames are analyzed, and the least significant bits are accordingly marked.(2) Taking advantage of the perspective of similarity theory, a steganographic method based on similarity matching is proposed. The core idea of this method is to adjust the change rate of the embedding operation by setting similarity thresholds for the available redundant bits with diverse steganographic performance. Experimental results demonstrate that the proposed method can provide good imperceptibility, and better speech quality compared with the conventional LSB replacement method.(3) The idea of hierarchical hidden bits for low bit-rate speech frames is presented according to the unequal contribution of different bits in each speech frame to the quality of reconstruction speech. Based on this idea, an adaptive hierarchical steganography is proposed to reduce the embedding distortion as far as possible. Experimental results demonstrate that the proposed method can render better imperceptibility than the traditional method while providing the identical embedding capacity.(4) In order to achieve an optimal balance of imperceptibility and hiding capacity, a hierarchical steganographic method using encoding strategy is presented. The main idea of the proposed method is to adopt different steganographic methods for diverse levels of the hidden bits displaying different hidden performance, to enhance the overall steganographic performance Experimental results demonstrate that the proposed method can achieve the optimal balance of embedding imperceptibility and hiding capacity, and provide better steganographic performance compared with the existing methods.