传声器阵列波束比判决语音增强方法

发布时间：2018-06-13 13:33

本文选题：波束形成器 + 语音增强　；参考：《声学学报》2017年04期

【摘要】：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。
[Abstract]:Aiming at the problem that it is difficult for single beamformer to suppress spatial coherent interference in depth, a speech enhancement method combining minimum variance distortionless response beamformer and symmetric subarray delay summation beamformer is proposed. A beam-output ratio factor is defined. According to the variation of the amplitude of the factor in the target sound region and the interference sound region, the method of adjusting the diagonal load of the sampling covariance matrix is given. Furthermore, the factor is used to filter the spatial interference in the post filter. In this paper, the real time updating method of upper and lower threshold in decision filtering is explained. The proposed algorithm can further suppress spatial interference and noise, and can meet the real-time needs. Experiments on a microphone array show that the proposed method is superior to the classical diagonal loading algorithm and the sampling covariance matrix scan reconstruction algorithm in output signal-to-noise ratio and speech quality.
【作者单位】：中国科学院声学研究所语言声学与内容理解重点实验室;中国科学院大学;
【基金】：国家自然科学基金项目(11461141004,61271426,11504406,11590770,11590771,11590772,11590773,11590774) 中国科学院战略性先导科技专项项目(面向感知中国的新一代信息技术研究:XDA06030100,XDA06030500,XDA06040603) 国家863计划(2015AA016306) 国家973计划项目(2013CB329302) 新疆维吾尔自治区科技重大专项项目(201230118-3)资助
【分类号】：TN912.35

【相似文献】