基于结构化时延估计算法的声源定位  

Sound Source Localization Based on Structured Time Delay Estimation Algorithm

作　　者：王莉[1] 赵艳阳[1] 刘雪峰 牛群峰[1] 

机构地区：[1]河南工业大学电气工程学院,郑州450001 [2]郑州大学信息工程学院,郑州4500012

出　　处：《科学技术与工程》2018年第2期123-129,共7页Science Technology and Engineering

基　　金：国家"十二五"科技支撑计划(2012BAF12B13)资助

摘　　要：针对声源定位研究中无法实现三维空间内高的定位问题、俯仰角以及方位角无法达到360°空间体内系统精度一致性的问题,提出了一种基于结构化延时估计算法的声源定位研究。首先对采集的波形进行去混响滤波处理,提高采集声音信号的信噪比以后,依据空间内拾音器的安装位置以及单元划分,引入细胞单元的结构化模型,该算法首先要根据细胞单元元素位置坐标,在空间体内建立虚拟坐标参考系,依据参考系中方位角、俯仰角、半径,推算出声源发声体在空间中的位置,该算法经过数据采集分析验证具有精度高、定位灵活等特点,适用于生活环境中的大部分场合。To solve the problem that the positioning obstacle in the three-dimensional space couldn＇t be achieved in the study of the sound source location,the pitch angle and the azimuth angle couldn＇t reach the accuracy of the system in 360° space,a method of sound source localization based on structured delay estimation algorithm was proposed.Firstly,the de-reverberation filtering of the acquired waveform was carried out to improve the signal-to-noise ratio of the collected sound signal,according to the installation location and the unit division of the pickup in the space,the structural model of the cell unit was introduced,in which the algorithm first established the virtual coordinate reference system in the space according to the coordinates of the cell element position,according to the azimuth,pitch angle and radius of the reference system,deducing the position of the sound source in the space.The algorithm is verified by data acquisition and analysis with high precision,flexible positioning and other characteristics,and is suitable for most of the living environment.

关 键 词：声源定位 结构化模型 去混响 滤波 延时估计算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]