Speex算法中AGC算法的分析与仿真.pdfVIP

2010 年第04 期，第43 卷 通 信 技 术 Vol.43，No.04,2010 总第220 期 Communications Technology No.220,Totally Speex 法中AGC 法的分析与仿真 ① ① ② 李菁菁 ， 黄孝建 ， 敬 （①北京邮电大学信息与通信工程学院，北京 100876；②清华大学 电子工程系，北京 100084） 【摘 要】语音在采集和传输过程中，由于语音源的差异、信道的衰减、噪声干扰以及远近效应，导致信号的幅度相差 较大，利用自动增益控制(AGC)可以优化信号电平，提高通信质量。而传统的AGC 缺少对话音和非话音信号的判别，会导致对 非话音信号的放大，造成干扰。深入分析了Speex 中基于语音活动性检测(VAD)的AGC 算法，通过仿真说明 算法可以实现自 动电平控制。 【关键词】VAD；AGC；统计模型；能量比较 【中图分类号】TN713 【文献标识码】A 【文章编号】1002-0802(2010)04-0164-03 Analysis and Simulation of AGC Algorithm in Speex ① ① ② LI Jing-jing ， HUANG Xiao-jian ， LI Jing (①School of Information and Telecommunication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China; ②Department of Electronic Engineering, Tsinghua University, Beijing 100084, China) Abstract】During the collection and transmission of speech signals, the amplitudes of signal have lager difference due to the difference of speech sources, the decay of channels, noises and near-far effect. Through Automatic Gain Control, signal levels could be optimized to improve the quality of communication. However, the lack of discrimination between speech and non-speech sounds of the conventional AGC would lead to a large amplification of non-speech sounds. In this paper, the AGC algorithm based on voice activity detection in speex is analyzed in depth. The experimental simulation indicates that this algorithm could realize automatic level control. Key words】VAD; AGC; statistical model; energy comparison 0 引言