Based The Msc8122 Of Amr-wb Algorithm Optimization And Realization

The Adaptive Multi-rate Wideband (AMR-WB) speech codec was selected by 3GPP/ETSI for GSM and WCDMA system for providing high quality speech services. Then, it was also selected by ITU-T as recommendation G.722.2 for wideband speech coding around 16kbit/s. AMR-WB provides high speech quality,low average bit rates and higher intelligence,making it well suitable for many applications in wireless system and wireline system.DSP has played a dominant role in development of digital speech coding with compactness, high performance and low power dissipation.The MSC8122 device of Freescale Company is a highly integrated DSP that combines four StarCore140 cores, which delivers enhanced performance while maintaining low power dissipation and greatly reducing overall system cost.Enhancing system user capacity and reducing system power dissipation and cost is the principal consideration factor in course of real application of AMR-WB algorithm. It need chooses high performance and low power dissipation DSP and efficiently optimizes the AMR-WB algorithm to reduces its complexity.Firstly,this paper gives a systemic analysis of AMR-WB algorithm, which focus on analyses the principle of core algorithm of coder and decoder, such as linear prediction analysis and quantization, adaptive codebook search, algebraic codebook search and speech synthesis. It also details the fixed-point C source code recommended by 3GPP and briefly analyses the technology of variable rate.Secondly, the paper researches on the architecture and the key technology of MSC8122 which includes M1, M2 and TDM interface; it also details the flow of software design and the framework of system hardware.Then,the paper researches on the key assemble optimization technology of MSC8122 and details the ways of assemble optimization of core algorithm, as well as analyses the interfaces, input parameters and initialization of coder and decoder and gives the way of multi-channel real-time realization of AMR-WB algorithm based on SmartDSP OS and MSC8122/26 ADS board. Finally, the paper gives the performance test result of optimized AMR-WB algorithm and real-time realization.The result proved: after effective optimization, complexity of AMR-WB algorithm fall 59% and the time complexity of AMR-WB algorithm only 16.62 MCPS; each StarCore140 core can real-time run 16 channels'AMR-WB algorithm and the output speech has a high reconstructed speech quality.