MELP算法的研究及其嵌入式平台的实现
首发时间:2017-01-04
摘要:本文在标准2.4Kb/s MELP语音编码算法的基础上,设计和实现了0.6Kb/s MELP语音编码算法,主要采用多帧联合量化和帧间参数内插的技术来实现,并通过合成语音波形图和合成语音质量对该算法进行评估。由于算法的复杂度较高,为了满足实时语音通信的要求,对算法进行优化。优化前一个超级帧的延时为77.56ms,而优化后则为28.81ms,测试结果表明优化后的算法可以满足实时语音通信的要求。最终,在Linux操作系统下用C语言实现了一个基于MELP算法的实时语音的通信与信息系统,并将该系统成功移植到香橙派平板,即ARM开发板上。
For information in English, please click here
Research On MELP Algorithm and Realization of Embedded Platform
Abstract:Based onthe standard MELP speech coding algorithm, the 0.6Kb/s MELP speech coding algorithm is designed and implemented by structuring super frame and interpolation of intra-frame parameters. Synthetic speech waveforms and synthetic speech quality are adopted to evaluate this algorithm. In order to meet the requirement of real-time speech communication, the algorithm is optimized because of high complexity. The delay of a super frame before optimization is 77.56ms, after optimization becomes 28.81ms. The result shows that the optimized algorithm can meet the requrement of real-time speech communication. Finally, a real-time speech communication and information system is realized based on MELP algorithm via C language on Linux system, and this system has been successfully transplanted into the Orange Pi PC, which is an ARM development board.
Keywords: Communication and Information System Speech Coding MELP Real-time Speech
基金:
论文图表:
引用
No.4715729117600414****
同行评议
共计0人参与
勘误表
MELP算法的研究及其嵌入式平台的实现
评论
全部评论0/1000