Research On The ITU G.729A Speech Coding Algorithm And Its DSP Implementation

The 8kb/s speech coding algorithm G. 729 Annex A was standardized by Study Group 15 (SG15) of the International Telecommunication Union-Telecommunication Standardization Sector (ITU-T) in June 1996, based on conjugate-structure algebraic code excited linear prediction (CS-ACELP). G. 729A is the reduced-complexity version of 8kb/s speech coding algorithm G. 729 standardized by SG15 of ITU-T in November 1995. The algorithm structure of G. 729A is the same as that of G. 729 and the bitstream of G. 729 A is interoperable with that of G. 729. The speech quality produced by G. 729A is equivalent to that of 32 kb/s ADPCM for most operating conditions, achieving the request of long-distance telephone. These conditions include channel errors, multiple encodings. The delay is short, the algorithmic complexity is not high and G. 729A is suitably implemented on DSP chip. G. 729A can be used for many applications, including DSVD, digital satellite communications, visual telephone, IP telephone and mobile audio services. The compression theory of G. 729A is deeply researched in the dissertation. Then the dissertation presents some optimized methods of G. 729A and implements G. 729A on TMS320VC5402 fixed-point DSP of Texas Instruments. Furthermore, a system scheme of the digital leave word telephone adopted G. 729A standard is given.In the first chapter, introduces the classification method and evaluation standard of the digital audio coding algorithm, digital audio coding standards in existence, the developing tendency of the digital audio coding algorithm. The low bit rate speech coding standards G. 729 and G. 723. 1 are particularly introduced. Describes the actuality of DSP, TMS320C54X DSP family of Texas Instruments. Then analyses the significance of studying G. 729A and optimized achieving G. 729A on DSP chip. Discusses the reason of implementing G. 729A on TMS320VC5402 fixed-point DSP chip of Texas Instruments. Finally the research background and the content of the dissertation are given.In the second chapter, researches CELP that is the theory base of G. 729A standard. CELP uses speech mathematical model, linear prediction and vector quantization.In the third chapter, the compression theory of G. 729A is deeply researched. Then describes the special methods used to reduce the computational complexity and improve the reconstructed speech quality in G. 729A standard.In the fourth chapter, describes the development support platforms used for TMS320C54X DSP family. Analyses how to achieve the fixed-point operation in DSP and how to achieve the integrated calculation with C and assemble language in TMS320C54X DSP family. Then points out that it is necessary that optimize G. 729A standard. Puts forward some optimized methods. The computational complexity of the optimized G. 729 A algorithm has been reduced to 1/8 that of the original algorithm. The experiments show that the output speech has the high reconstructed speech quality with the optimized G. 729A algorithm.In the fifth chapter, designs a digital leave word telephone adopted G. 729A standard. Analyses the research background, functions and key techniques of the system.Furthermore, its hardware and software are described.In the sixth chapter, the main conclusions of the dissertation are summarized and the future research work is pointed out.