Improving security of quantization-index-modulation steganography in low bit-rate speech streams

In this study, we mainly concentrate on quantization-index-modulation (QIM) steganography in low bit-rate speech streams, and contribute to improve its security. Exploiting the characteristics of codebook division diversity in the complementary neighbor vertices algorithm, we first design a key-based codebook division strategy, which follows Kerckhoff’s principle and provides a better security than the previous QIM approach. Further, to resist the state-of-the-art steganalysis, following a general belief that fewer and smaller cover changes are less detectable and more secure, we present an improved QIM steganography, which introduces random position selection to adjust the embedding rate dynamically, and employs matrix encoding strategy to enhance the embedding efficiency. The proposed approach is evaluated with ITU-T G.723.1 as the codec of cover speech and compared with the previous work. The experimental results demonstrate that the proposed approach outperforms the traditional QIM approach on both steganographic transparency and steganalysis resistance. Moreover, it is worth pointing out that our approach can effectively work in conjunction with not only G.723.1 codec but also all other parametric speech coders, and be successfully applied into Voice-over-Internet-Protocol systems.