基于CELP的语音编解码系统及其DSP实现  

CELP based speech coding/decoding system and its DSP realization

作　　者：覃爱娜[1] 杨勇[1] 陈明义[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《中南工业大学学报》2003年第4期416-419,共4页Journal of Central South University of Technology(Natural Science)

摘　　要：为了解决传统的语音编码使用波形编码或者参数编码方法时存在的编码速率较高,语音合成的自然度不高等不足,研究了以码激励线性预测(CELP)为基础的混合编码方法;探讨了CELP的算法原理及其关键技术,特别是对其中的码本搜索算法进行了改进,减小了算法的复杂度,提高了运行速度,并在TI公司的TMS320VC5410开发套件上得到了实时实现.实验结果表明,该算法在编码速率为8kbit/s时有较好的语音回放效果.There are several disadvantages in the traditional waveform or parameter speech coding method, such as the relatively high coding speed, or bad naturalness of the constructed speech .This article discusses the mixed coding method based on Code Excited Linear Prediction(CELP),presents CELP Coding/Decoding theory and its essential technology ,with which code book searching algorithm has particularly improved. Based on the TMS320VC5410 EVM version of TI Co, the realtime system can ensure good voicereplay quality at the 8 kbit/s coding speed.

关 键 词：语音 编码 码激励线性预测 数字信号 

分 类 号：TN912.3[电子电信—通信与信息系统]