语音增强系统中量测缺失问题的研究  

Study on losses of measurement in speech enhancement

作　　者：高宝成[1] 孙晶晶[1] 

机构地区：[1]北京邮电大学自动化学院,北京100876

出　　处：《电子测量技术》2011年第12期62-65,共4页Electronic Measurement Technology

摘　　要：在使用多个无线传感器对目标语音信号进行检测的语音增强系统中,由于网络环境带来的信息丢失,会对其滤波稳定性造成影响。以卡尔曼滤波为背景,探讨三传感器语音增强系统中量测缺失对卡尔曼滤波稳定性带来的影响。得出一些重要结论。在期望有界准则下,每个通道的量测成功传递概率存在一个关键值,该关键值的变化范围存在上界和下界。当量测成功传递概率大于关键值上界时,滤波一定稳定;当量测成功传递概率小于关键值下界时,滤波一定发散。以及多传感器对滤波稳定性的影响并非各个传感器的简单叠加。最后通过仿真实验证明了上述结论的正确性。In the speech enhancement system based on wireless sensor network,the effect of communication delays and loss of information in the control loop cannot be neglected.In this paper,we address this problem from the stability of Kalman filtering over unreliable network based on three sensors.We study the statistical convergence properties of the estimation error covariance,showing the existence of a critical probability in the successful transmission probability for each measurement.If the successful transmission probability surpasses the upper bound of the probability,the filtering is stable while if the successful transmission probability does not reach the lower bound of the probability,the filtering is unstable.Moreover,it can be seen that the affect of different measurements on the stability is not so simple as to be their linear sum-up.All these conclusions above will be verified by the simulation.

关 键 词：无线传感器 语音增强 卡尔曼滤波 量测缺失 滤波稳定性 

分 类 号：TN912.35[电子电信—通信与信息系统]