论文部分内容阅读
针对单一波束形成器难以深度抑制空间相干干扰的问题,提出了一种综合了最小方差无畸变响应波束形成器与对称子阵延时求和波束形成器的语音增强方法。定义了一种波束输出比因子,根据该因子在目标声区域和干扰声区域的幅值变化,给出了采样协方差矩阵对角加载量的调整方法,并进一步利用该因子在后滤波环节对空间干扰进行判决滤波。文中对判决滤波时的上限阈值和下限阈值的实时更新方法给出了说明。所提出的算法能进一步抑制空间干扰和噪声,且可满足实时需要。在传声器圆阵上的实验表明,该方法在输出信干噪比及语音质量上,均优于经典对角加载算法及采样协方差矩阵扫描重构算法。
Aiming at the problem that the single beamformer can not suppress the spatial coherent interference deeply, a speech enhancement method based on the minimum variance distortionless beamformer and the symmetric sub-array delay-sum beamformer is proposed. A beam output ratio factor is defined. Based on the change of the amplitude of the target sound region and the disturbing sound region, a method of adjusting the angular loading of the sampling covariance matrix is given. By using this factor in the post-filtering step Space interference decision filtering. The paper gives a description of the real-time updating method of the upper threshold and the lower threshold in the decision filtering. The proposed algorithm can further suppress the spatial interference and noise, and can meet real-time needs. Experiments on the microphone circular array show that this method is superior to classical diagonal loading algorithm and sampling covariance matrix scan reconstruction algorithm in terms of output SINR and speech quality.