基于信号子空间逼近的多麦克风语音信号增强方法  

A Method Based on Multi-Microphone for Speech Enhancement by Signal Subspace Approach

作　　者：曹若臻[1] 于凤芹[1] 

机构地区：[1]江南大学通信与控制工程学院,江苏无锡214122

出　　处：《江南大学学报（自然科学版）》2006年第2期204-206,210,共4页Joural of Jiangnan University (Natural Science Edition)　

基　　金：江苏省自然科学基金项目(BK2002068)

摘　　要：对现有的基于自动波束形成的麦克风阵列语音信号增强算法提出了改进.将各个麦克风采集到的信号利用ABF(自适应波束形成)进行延时补偿并求和,消除信号中弱相干和不相干噪声;再利用信号子空间逼近的方法进一步去除残留噪声.仿真试验结果表明:把自适应波束形成技术和信号子空间逼近的方法结合起来,能够得到良好的去噪效果.A method using automatic beamforming （ABF） technology and signal subspace approach for speech enhancement is proposed. The signals are received by a microphone array consisting of 3 microphones. The method has two steps. Firstly, a automatic beamforming technology is used to combine signals from microphone array into one. Secondly, we take signal subspace approach method to cancel noises from the result of first step. The computer simulations show that this method can bring good effects,

关 键 词：麦克风阵列 语音信号增强 自适应波束形成 信号子空间逼近 

分 类 号：TP912.3[自动化与计算机技术]