基于最大似然估计的自适应回声消除算法  

Block-Sparse Least Mean Square Algorithm Based on M-Estimator Against Impulsive Interference

作　　者：魏丹丹 周翊[2] 赵宇 WEI Dandan;ZHOU Yi;ZHAO Yu(School of Information Engineering,Zunyi Normal university,Zunyi Guizhou 563002,China;School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]遵义师范学院信息工程学院,贵州遵义563002 [2]重庆邮电大学通信与信息工程学院,重庆400065

出　　处：《西南大学学报（自然科学版）》2023年第4期210-218,共9页Journal of Southwest University(Natural Science Edition)

基　　金：贵州省自然科学基金项目([2018]1180);重庆市教委科学技术研究项目(KJQN201900605);遵义市科技计划项目(遵市科合HZ字(2021)212号).

摘　　要：针对智能语音终端设备在复杂声学环境中由于回声抑制不佳而影响用户听觉体验的问题,该文提出一种基于最大似然M-估计的鲁棒型自适应回声消除算法.该算法利用M-估计函数构造权矢量函数来降低声学环境中冲激噪声的影响,采取Hample三段函数得到新的代价方程来替代原有的低阶范数最小均方误差,然后再利用负梯度最陡下降方法得到新的滤波器权系数更新公式.计算机仿真实验结果表明,该算法在复杂声学环境中依然能够保证良好的回声抑制能力,并且能够及时跟踪回声信道的变化.相比同类算法,该算法在复杂声学环境中的稳态误差性能和收敛速度均有提高.The least mean square(LMS)aiming at the problem that intelligent voice terminal equipment affects the user s hearing experience due to poor echo suppression in complex acoustic environments.This paper proposes a robust adaptive acoustic echo cancellation algorithm based on maximum likelihood M-estimator.The algorithm constructs the weight vector function based on M-estimator to reduce the impact of impulse noise in the acoustic environment.The Hample three-stage function is used to obtain a new cost equation to replace the original low-order norm least mean square error.A new vector updating equation for filter coefficients was deduced by steepest-descent method.Computer simulation experiment results show that the proposed algorithm can still ensure good echo suppression ability in complex acoustic environment and track the changes of echo channel in time.Compared with similar algorithms,the new algorithm has improved steady-state error performance and convergence speed in complex acoustic environments.

关 键 词：回声消除 鲁棒性 低阶混合范数 M-估计 跟踪性能 

分 类 号：TN911.72[电子电信—通信与信息系统]