中文摘要：

在复杂的语音环境中,利用麦克风阵列语音增强技术能有效地拾取目标语音信号并消除噪声干扰,但传统的麦克风阵列波束形成的加权系数是固定的.为了能灵活地控制麦克风阵列波束的形成方向以及消除其波束旁瓣带来的残余噪声,提出了一种基于麦克风阵列的自适应语音增强技术.该技术将自适应滤波器和麦克风阵列相结合形成波束可控的广义旁瓣消除器,然后在广义旁瓣消除器后面续接一个改进的谱减法,并加入契比雪夫窗函数.仿真实验结果表明,所提出的语音增强方法能有效去除语音信号中的噪声干扰,相比于传统广义旁瓣消除器,信噪比大约提高了3.5 d B.

英文摘要：

In a complex speech environment, microphone array speech enhancement tech- niques can effectively extract target speech signals and eliminate noise. In a conventional microphone array, the beam forming weighting coefficients are fixed. To flexibly control the beam forming direction of the microphone array and eliminate residual noise due to side-lobes, this paper presents an adaptive enhancement technique for microphone array speech. A microphone array with an adaptive filter is used to form a controllable general- ized sidelobe canceller （GSC）, followed by improved spectral subtraction with a Chebyshev window function. Simulation results show that the proposed method can effectively remove noise, improving the performance by about 3.5 dB with respect to the ordinary GSC.