Speex算法中AGC算法的分析与仿真  

Analysis and Simulation of AGC Algorithm in Speex

作　　者：李菁菁[1] 黄孝建[1] 李敬[2] 

机构地区：[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

出　　处：《通信技术》2010年第4期164-166,共3页Communications Technology

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,通过仿真说明该算法可以实现自动电平控制。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.

关 键 词：VAD AGC 统计模型 能量比较 

分 类 号：TN713[电子电信—电路与系统]