首页|声源定位算法反馈机制的引入

声源定位算法反馈机制的引入

扫码查看
近年来人们对声源定位实时性的需求不断提高,实时性要求在采集较短的音频情况下实现准确定位.通过麦克风阵列的几何分布,在时延估计定位算法引入一种普适的反馈机制以提高声源定位的准确性.利用Matlab编写仿真程序在二维平面内进行定位仿真,对比引入反馈机制前、后的定位效果.仿真结果表明,引入反馈机制后声源定位算法的抗噪性、抗混响性和稳定性均有不同程度的提高.搭建一套声源定位设备进行实际测试,实验结果表明,大多数声源坐标估计值和实际值的偏差在1 cm以内,实现了 cm级的定位.这为噪声场中声音信号的处理提供了更多的可能性,例如实现更高性能的降噪.
Introducing Feedback Mechanism to Sound Source Location
In recent years,the demand for real-time sound source localization(SSL)is increasing,and real-time SSL require accurate localization while collecting short audio.Therefore,a universal feedback mechanism is introduced into the time delay estimation algorithm to improve the accuracy of SSL based on the geometric distribution of the microphone array.Firstly,a simulation localization experiment is carried out in a two-dimensional plane using a simulation program written by Matlab,and the localization results before and after the introduction of the feedback mechanism are compared.The results show that after introducing the feedback mechanism,the noise resistance,reverberation resistance and stability of the SSL algorithm are all improved with different degrees.Then a set of SSL equipment is built for practical testing,and the deviation of most of the estimation values and actual value is within 1 cm.We accomplished centimeter-level SSL,which provides more processing in noisy fields,such as achieving higher performance in noise reduction.

sound source localizationtime delay evaluationfeedback mechanismgeneral cross correlationleast square method

周龙杰、赵思屹、马洪宇

展开 >

武汉大学物理科学与技术学院,武汉 430072

声源定位 时延估计 反馈机制 广义互相关 最小二乘法

武汉大学实验技术项目

WHU-2022-SYJS-01

2024

实验室研究与探索
上海交通大学

实验室研究与探索

CSTPCD北大核心
影响因子:1.69
ISSN:1006-7167
年,卷(期):2024.43(1)
  • 15