子带MCRASC-MGSC微型麦克风阵语音增强算法  

Subband MCRASC-MGSC algorithm for speech enhancement with the mini-typed microphone array

作　　者：曾庆宁[1,2] 欧阳缮[1,2] 

机构地区：[1]西安电子科技大学通信工程学院,陕西西安710071 [2]桂林电子科技大学信息与通信工程学院,广西桂林541004

出　　处：《西安电子科技大学学报》2010年第6期1011-1016,共6页Journal of Xidian University

基　　金：国家自然科学基金资助项目(60272038);广西回国人员基金资助项目(0639028)

摘　　要：微型麦克风阵语音增强在基于微型设备的语音通信及语音识别中具有重要的应用价值.通过引进多路抗串扰自适应信号抵消(MCRASC)技术,对子带修正广义旁瓣抵消(MGSC)方法提供更为有效的信号阻塞方式,为微型阵列语音增强提出了一种更为可行的算法.理论分析和真实环境下的实验结果验证了所提算法的有效性和优越性.实验中,所提算法比子带MGSC算法所获增强语音的信噪比提高了12.5 dB.Speech enhancement with the mini-typed microphone array is of great value in speech communication and speech recognition with mini-typed devices. By employing the Multi-channel Crosstalk Resistant Adaptive Signal Cancellation（MCRASC） technique, a more efficient module for signal blocking is presented for the subband Modified Generalized Sidelobe Cancellation（MGSC） method, and thus a more feasible algorithm is proposed for mini-typed array based speech enhancement. Theoretic analysis and experimental results in real environment have verified the availability and advantages of the proposed algorithm. The signal to noise ratio has been improved by 12.5 dB for the enhanced speech by the proposed algorithm compared with the subband MGSC algorithm in the experiment.

关 键 词：麦克风阵 语音增强 广义旁瓣抵消 子带 

分 类 号：TN957.52[电子电信—信号与信息处理]