基于可量化性能分级的自适应IP语音隐写方法

Adaptive Voice-over-IP Steganography Based on Quantitative Performance Ranking

论文以客观语音质量评价和信噪比为量化手段,分析了参数编码中语音帧的每个比特位对重构语音质量影响的不均衡性,并提出了一种载体可隐藏位的分级方案,以达到充分利用各载体位的目的.以此为基础,进一步提出了一种基于载体位置分级的自适应IP语音隐写方法.该方法可自适应地选择最佳的隐写位置,以提高隐蔽通信的不可感知性.以大量的语音样本为载体,对提出的方法进行了实验验证.结果表明,本文的分级方案是正确可行的,且提出的分级隐写方法较之传统方法具有更好的隐写性能.

In this paper, we first analyze the unequal impacts of various cover bits on the quality of decoded speech for parameter codecs, and further design a ranking scheme for available cover bits to utilize the cover fully. Moreover, we present an adaptive VoIP steganography based upon the performance ranking of cover bits, which can adaptively choose the optimal steganographic positions to enhance the imperceptibility of covert communications. The proposed methods are evalu- ated and tested with a large number of speech samples as the cover. The experimental results show that the proposed ranking scheme is sound and feasible,and the presented steganogaphic method also outperforms the existing ones.