Transform-domain and DSP based secure speech communication system

摘要

Secure speech communication has been of great significance in civil, commercial and particularly in military communication systems. Speech scramblers play a major role in these systems and transform-domain, specially, DCT (Discrete Cosine Transform) based encryption of speech has often been a preferable choice for researchers and engineers. In such systems, besides security, removal of redundant information and the execution speed of the algorithms have also been the main focus of research and exploration. In this paper, we concentrate on the investigation of these three parameters in general and security and speed in particular. In this research these attributes are envisaged by exploiting ITU-T G.723.1 speech codec, DCT based speech encryption algorithm and Texas Instruments' 32-bit floatingpoint DSP (Digital Signal Processor) TMS320C6711, respectively. This work, in fact is a continuation and an advancement of the work carried out by Jameel et al. [A. Jameel, M.Y. Siyal, N. Ikram, A Robust Secure Speech Communication System using G.723.1 and TMS320C6711 DSP, Microprocessors and Microsystems 30 (1) (2006) 26-32.] in which they presented a redundancies-free, secure and fast speech communication system using G.723.1 codec, time-domain speech scrambling techniques and Texas Instruments TMS320C6711 DSP. On the contrary, in this exploration the same system is improved significantly in terms of security and speed. In order to reduce residual intelligibility thereby to enhance security a new DCT based algorithm using sub-framing and lookup table is proposed, implemented and analyzed. In terms of speed, the reference code of G.723.1 is further optimized on advanced level. Additionally, the schemed is designed in such a way that it also diminishes talk spurt and intonation present in the original speech signal which, as a result, increases the level of security. Hence a redundancies-free, secure and high speed communication system has been proposed, implemented and analyzed using various types of speech segments. The results show that the proposed system is suitable for real-time civil, commercial and military applications.