The Realization And Optimization Of G.729B On DM642

Low bit rates coding of speech is especially important for the restriction of communication channels and networks to bit rates transmitted, so many international organizations published their own standards of data code, including the G.729s speech coding recommendations issued by ITU-T. G.729 is one of the most complicated algorithms of speech compression, which is based on conjugate-structure algebraic-code-excited linear-prediction (CS-ACELP), coding of speech at 8kbit/s. Annex B to ITU-T recommendation G.729 defines the Voice Activity Detection (WAD), discontinuous Transmission (DTX) and Comfort Noise Generator (CNG) algorithms. These algorithms are used during silence periods of speech to reduce ratio of transmission and improve ratio of use in signal channel. It's a best silence compression scheme for G.729.The TMS320DM642 Digital Media Processor is considered to work well for people who use DSP at present. It is based on the C64x CPU, which is part of the C6000 DSP family. The DM642 integrates a number of peripherals to address the development of audio and video applications, such as multiple audio serial port, EMAC and so on. DM642 can not only make it easier for users to process the complicated algorithms of audio and video, but also can make the seamless connection between the coder and PCFs data bus. So it has become an effective approach using DM642 to realize the silence compression algorithm G.729B.The paper analyzes the theory of G.729 speech codec and key technologies of G.729 Annex B in details and introduces the hardware platform especially the structure characteristics of TMS320DM642 and TLV320AIC23B. Under the condition of guaranteeing favorable speech quality, it is necessary to reduce the code quantity of recommendation G.729B. Then the amended code is transplanted to the platform of DSP TMS320DM642 to be optimized by way of C language level and assembly level. Finally the algorithm of G.729 Annex B is implemented in real time on the fixed DSP. The results of experiment show that the decoded speech signals whose definition and nature characteristics are all good by the method of subjective audition, and its reconstructed wave shape and frequency spectrum are similar to the original ones. This research has been done through emulator and simulator. In this paper I not only have done optimizations to the G.729B algorithm, but also have realized it in the DM642 application system, which lays a foundation for G.729B to be productionized, such as the IP telephone and H.323 multimedia communication system etc.In the research process, the following work has been done:(1)The architecture of the hardware platform SEED-VPM642 especially the DM642 chip has been learned, then the functions and applications of McASP ( Multichannel Audio Serial Port) and audio codec TLV320AIC23B have also been studied.(2) Have been familiar with the software platform of system especially the code developing environment that is CCS.(3)The knowledge about speech coding and the theory of algorithm G.729/G729B have been researched.(4) The work to test the source code of G.729B and transplant it to DSP have been done.(5) The source code has been optimized by way of emulator and simulator and finally been implemented in real time on the fixed DSP.