Implementation Of Speech Compress Coding Based On DSP

With the rapid development of DSP technology, it become a powerful tool in the speech signal processing, which makes it possible to implement all kinds of complex algorithms in real time. Based on different applications, many standards of speech compress coding have been established by different international standardization orgnizations. The 8kbit/s speech compress coding G729A was recommended by ITU-T in June 1996 based on Conjugate Structure-Algebraic Code Excited Linear Prediction (CS-ACELP). With short algorithm delay, good speech quality and acceptable algorithm complexity, G.729A has been widely used in the field of IP telephone, mobile communications, digital satellite communications, and so on. The Adaptive Multi-rate Wideband (AMR-WB) speech compress coding G.722.2 was recommended by ITU-T in January 2002 based on ACELP. Compared with the speech quality as produced by G.729A, that by G.722.2 is much more natural and intelligible. However, G.722.2 is of much higher complexity than G.729A. The foreseen applications for G.722.2 standard are the following: Mobile Communications, PSTN applications, ISDN wideband telephone, ISDN video-telephony and video-conferencing, etc.The paper is composed of following parts: Firstly, the present speech compress coding technologies and standards are described in the paper. And the speech compress coding standards constituted by ITU-T are introduced. Secondly, the algorithm principles of G.729A and G.722.2 arestudied, and the implementation process of G.729A algorithm is analyzed deeply. Structures and properties of two kinds of TI fix-point DSP are analyzed. Finally, the following two vocoders are implemented.(1) The G.729A vocdoer is implemented on TMS320C5416 in real time in pure assembly language. In the implementation process, several issues are analyzed and optimized, which include assignment of memories, selection of instructions, optimization of loops, expansion of functions, selection of address modes and use of registers, pipeline conflicts, etc. The test results show that the vocoder has low delay and can be applied in the realms requiring low delay such as multi-channel telecommunications.(2) The G.722.2 vocoder is implemented on TMS320C6416 in real time by optimized C code. In the process of implementation, the C code optimization technologies involve issues of intrinsic function, loop expansion and software pipeline, and so on. The test results show that the vocoder is fit for practical applications.In addition, for the above two vocoders, the components of the hardware platform and the design of the system software are analyzed and introduced, and the test results are given.