摘要
随着通用DSP性能的不断提升,通过软件实现AMR-WB等各种计算密集型媒体应用成为可能。首先研究了AMR-WB语音编码中采用的基本原理;然后采用YHFT-DSP/700模拟器详细分析了各种编码模式下各模块的计算复杂度,并给出每个模块所需的输入输出缓冲和访存带宽大小,为AMR-WB在通用DSP上的优化提供了指导。
With the development in the performance of the general DSP, it becomes possible to implement compute-intensive media application such as AMR-WB with software. The principle of the AMR-WB speech coding algorithm is studied first. Then the computation complexity of the each module under all modes by using YHF＇r-DSPR00 emulator is analyzed in detail and the requirement of I/O buffer and memory bandwidth is presented. The research work provides guidance for optimization of the AMR-WB speech coding algorithm based on general DSP.
出处
《电声技术》
2009年第7期68-72,共5页
Audio Engineering
基金
863计划资助(2007AA01Z287)
