摘要
在研究ITU-T的G.729 CS-ACELP语音编码算法基础上,本文介绍了使用TI公司的TMS320C6x系列DSP实现该算法的一些问题以及一些应用中的具体优化技术,这在一定程度上降低了该算法的计算复杂度,而输出语音仍然保持了很高的合成品质。然后,本文讨论了如何在一个电信级的应用内核上对该语音编码算法进行多声道扩展的问题,该多通道编码系统基本能达到实用要求。最后,在实时MCPS、内存要求、处理延时和实时性能等方面对该DSP实现的编码器做了一个评价,可以作为其他语音编码器开发的参考。
Based on studying the G.729 speech coding algorithm, this paper describes the details of the ITU-T G.729 speech coder implementation using Texas Instruments' high-performance programmable TMS320C6x DSP family and the various techniques used in optimization. It can considerably reduce the calculation burden and the out speech still keeps a high reconstructed quality. This paper presents the simple techniques to extend the coder for multiple channels using a telecom application specific kernel, the multiple channel coding system can basically reach the demand of the application. Finally we give an evaluation of the real-time MCPS, memory requirements, processing delays and associate trade-offs of the coder using this DSP, and it is helpful for constructing other speech coders.
出处
《计算机工程与科学》
CSCD
2004年第4期51-54,共4页
Computer Engineering & Science