摘要
本文介绍了ITU-TG.723.1语音编解码器在朗讯科技(LucentTechnologies)公司新一代定点数字信号处理芯片DSP16k上的实时实现技术。本文编程实现了G.723.1标准的所有功能模块,包括在变码率下的静音检测(VAD)和舒适噪声产生(CNG)模块。本文的实现完全复合ITU-TG.723.1标准的定点算法,通过了ITU-T的所有测试矢量。同时木文比较了G.723.1在两种定点数字信号处理器DSP16xx和DSP16k上实现时内存占用空间和计算复杂度的差异。
This paper describes real-time implementation of the ITU-TG.723.1 speech vocoder on DSP16k digital signal processor of the Lucent Technologies Inc.. The G.723.1 speech vocoder is an ITU-T standard for a dual rate speech encoder and decoder. An option for variable rate operation is possible. The implementation in this paper is fully compliant with the ITU-TG.723.1 Standard for Fixed Point DSPs and it passes aIl ITU-T test vectors. The differences in memory and complekity between the implementations on DSP16xx and on DSP16k are presented.
出处
《应用声学》
CSCD
北大核心
2000年第5期15-19,共5页
Journal of Applied Acoustics
基金
贝尔实验室(中国)上海资助