用于语音识别的鲁棒自适应麦克风阵列算法  

Robust adaptive microphone array algorithm for speech recognition

作　　者：赵贤宇[1] 王作英[1] 

机构地区：[1]清华大学电子工程系,北京100084

出　　处：《清华大学学报（自然科学版）》2004年第10期1433-1436,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目(863-306-ZD03-01-2)

摘　　要：对现实环境中存在的混响以及非平稳干扰语音信源等因素导致的算法性能下降,提出了一种用于语音识别的鲁棒旁瓣对消算法。讨论了旁瓣对消算法在自适应麦克风阵列中的应用,分析了算法在不同的混响条件下、不同的干扰源的噪声抑制能力。该算法通过分帧处理将输入信号划分为一系列短时平稳的信号片段。根据当前帧的信噪比决定自适应滤波器的权系数更新方式。采用一定的范数约束来限制自适应滤波器权系数的误调整。实验结果表明该麦克风阵列在混响的现实环境中能够有效抑制平稳噪声源和交叠谈话背景干扰,提高了语音识别器的抗噪性能。The generalized sidelobe cancellor method was used with an adaptive microphone array to improve speech recognition. Its noise suppression capability in different environments with reverberation and different kinds of noise sources is related to its speech recognition rate. The algorithm seeks to improve performance degradation when reverberation is present and with non-stationary interference. The algorithm frames input signalsinto short-time stationary segments. The weight adaptation mode of the interference suppression filter is controlled by the signal-to-noise ratio of the current frame. Norm constraints are imposed to restrict false adaptation of the filter weights. Test results show that the algorithm effectively suppresses stationary interference and cocktail party non-stationary interference in realistic reverberating environments and significantlyimproves thespeech recognition performancein adverse noise conditions.

关 键 词：语音识别 麦克风阵列 自适应信号处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]