G.729语音编码算法优化研究及ARM实现

The Optimization and Implementation with ARM of Speech Encoding Algorithm Based on G.729

基于ARM（advanced RISC machine）体系架构和嵌入式Linux操作系统构建了算法平台,并在该平台上移植了G.729语音编解码算法。为了降低G.729算法的复杂度,提出了一种基于部分失真搜索（partial distortion search,PDS）和范数排序搜索（norm order search,NOS）的LSP（line spectrum pair）矢量量化算法及基于ARM指令集的算法优化策略。实验结果表明：改进后的G.729算法和ARM实现方案大大降低了语音处理时间,极好地满足了点对多点实时语音通信的要求。

A platform based on advanced RISC machine（ARM） architecture and Linux operating system is designed, in which the algorithm of G. 729 is implanted. To lower the complexity of G. 729, a LSP vector quantification method based on the improved NOS and PDS is put forward and ARM instructions are also modified. Finally experiment results demonstrate the modified G. 729 and ARM instructions lower heavily the time-consuming of speech processing, satisfy capitally the requirements of a point to multiple point real-time speech communication.