摘要
为了满足短波窄带数字保密通信的需要,本文提出并实现了一种速率为600b/s的甚低速语音编码算法。该算法基于先进的混合激励低速语音编码技术,通过对模型结构及参数编解码等方面的改进,在600b/s的速率上获得了比较满意的合成语音。客观音质测试和非正式主观试听结果表明,算法合成语音的可懂度和清晰度与传统的2.4kb/s线性预测（LPC）语音编码算法相当。该算法已经在单片DSP芯片TMS320VC5416上实时实现。
In order to meet the need of short-wave narrow-band digital security communication, a new 600b/s speech algorithm is proposed and implemented in this paper. Based on the advanced mixed- excitation low bit-rate speech coding techniques, the algorithm can synthesize fairly satisfactory speech at 600b/s by improving the model structure and parameter coding methods. The results of objective quality test and informal subjective listening test show that the intelligibility and articulation of the algorithm are comparative to that of the traditional 2.4kb/s LPC algorithm. The 600b/s speech algorithm has been successfully implemented on single DSP chip TMS320VC5416.
出处
《信号处理》
CSCD
2002年第5期403-409,共7页
Journal of Signal Processing
基金
本文得到国家高等学校骨干教师资助计划资助
