传声器阵列语音增强新方法与仿真实验  

A New Method and Simulation Testing for Speech Enhancement Based on Microphone Arrays

作　　者：樊伟[1] 刘庆华[1] 

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004

出　　处：《电声技术》2009年第10期61-65,共5页Audio Engineering

基　　金：广西自然科学基金(0832007Z)

摘　　要：结合宽带LCMV语音增强和噪声自适应对消原理,提出了一种基于传声器阵列的多级自适应语音增强方法。在真实声学环境下构建实验模型并对语音信号进行采集,通过MATLAB仿真验证了算法。分析表明,该方法计算复杂度低,对语音信号损伤小,信号SNR提高显著。A novel approach about multistage adaptive algorithm for speech enhancement is proposed, based on broadband LCMV（Linearly Constrained Minimum Variance） theory and the ANC（Adaptive Noise Cancelation） theory. The signals are collected in the real acoustic environment. Then the algorithm is validated using the MATLAB simulation tools. The results show that the complexity of the new method is simple and efficient, and there is little distortion to speech signals. Moreover, it is marketable for the improvement of the SNR.

关 键 词：语音增强 LCMV滤波 自适应抵消 传声器阵列 NLMS算法 

分 类 号：TN912[电子电信—通信与信息系统]