The Design And Implementation Of A Sampling And Processing Speech DSP System Based On ITU-T G.729A

In recent years, the application fields of digital signal processors (DSP) have expanded rapidly. The architecture of fixed-point DSP is very suitable for implementing fixed-point algorithm. Moreover, TMS320C5000 DSP is power-efficient and equipped with Integrated Development Environment . So using DSP system to implement speech processing algorithm can not only meet the requirement of real time signal processing but also reduce the cost of development.With the development of information technology, more and more data need to be stored and transmitted with limited bandwidth. So the technology of data compression becomes more important. Under this environment, many international organizations published their own standards of data code.G.729A is the speech coding recommendation issued by ITU-T.It is used for coding speech at 8kbit/s based on conjugate structure algebraic-code-excited linear-prediction (CS—ACELP) scheme. G.729A which is a very efficient digital coding method can be applied to telecommunication.Because real-time speech processing needs complex computation, fixed-point DSP TMS320VC5409 is used as the core processor in this experimental system. TMS320VC5409 has a high performance of signal processing, and it also has abundant on-chip memory and flexible on-chip peripherals. Two multi-channel buffered serial port (McBSP) on it can be used for data input and output ports. Analog interface chip TLC320AD50 can provide 16-bit signal conversion from digital to analog (D/A) and from analog to digital (A/D), and the maximum conversion rate is 22.05 kHz. The sampling rate used in G.729A recommendation is 8K, so it is enough to select TLC320AD50 as analog interface chip in this experimental system. Two TLC320AD50 are connected to TMS320VC5409 through two McBSP on it, respectively.This paper introduces the development of DSP and speech code technique, give particular discussion about G.729A compressing algorithm , gives a way about how design a DSP board. The kernel of this paper occurs in the realization of TMS320VC5409 DSP board and realization of G729A speech signal coder in previously board.TMS320VC5409's software development tool provides high-level language compiler, the C language compiler, which has the difficulty in real-time signal processing in our board. The G. 729A speech-compressing algorithm, written in C language, produces a huge amount of program codes, which cannot be stored in the chip-memory of DSP because of its 32K storing space. Therefore, G.729A speech-coding algorithm written in C language is faced with the problems of low processing speed, and synthesis. The main goal of this paper is to accomplish G.729A speech signal compressing algorithm programmed in C54X general assembly language, and to overcome the problems above. After about one year insisting and hardworking, this goal has become true. Combined with the characteristics of TMS320VC5409 and the specialty of fixed-point arithmetic, G.729A algorithm optimization is made G.729A speech codec has been successfully implemented on this experimental system. From the results obtained, the reconstructed speech is satisfying.