| The sound source localization technology based on microphone array collects the target sound source signal through the microphone array,performs array processing on the signal to obtain each road sound feature,and combines the sound source localization algorithm to obtain the position information of the sound source.This location technology is a location technology based on the principle of passive target positioning.Compared with radar and sonar localization,this technology has better concealment and security.At present,the technology has been widely used in the field of military and life,such as sniper positioning,drone positioning,video conferencing and intelligent audio and so on.In this paper,we focus on the low-level and slow-moving targets such as drones and air blasts,and study several commonly used sound source localization algorithms,analyzing their differences and their respective advantages and disadvantages.Finally,we set up a set of sound localization system based on embedded system,and use PHAT generalized cross-correlation algorithm and improved Multiple Signal Classification(MUSIC)algorithm to realize real-time location of burst point and moving sound source.The main research contents include the following aspects:(1)Analyze the advantages and disadvantages of the commonly used three microphone array sound source localization algorithms and their respective application scenarios,and select the algorithm that is suitable for the target sound source characteristics studied in this paper.(2)Analyze the time-frequency characteristics of the target sound source,array signal processing model under different sound fields and the effects of the number of microphones,array element spacing,multipath and Doppler effects on the sound source localization accuracy.(3)In terms of positioning algorithm based on Time Difference of Arrival(TDOA),several commonly used time delay estimation algorithms are studied,such as basic cross-correlation algorithm,generalized cross-correlation algorithm and cross-power spectrum phase method,and so on.Each algorithm is verified and analyzed by MATLAB,and their time delay estimation accuracy under different noises is compared.Finally,the geometric positioning algorithm of five-element linear array and five-element pyramid array is analyzed and their positioning formulas are deduced.(4)In terms of positioning algorithm based on high-resolution spectrum estimation,the steps to implement the sound source azimuth estimation using the classical MUSIC algorithm are analyzed.And an improved MUSIC algorithm is introduced for far-field single sound sources.Finally,under the different experimental conditions,the classic MUSIC algorithm and the improved MUSIC algorithm are verified and analyzed by MATLAB.(5)The overall system design and implementation,including the establishment of the hardware platform and MATLAB program design.A set of sound source positioning system is built by the X86 embedded single board,microphone array and external USB sound card.(6)Combined with the PHAT generalized cross-correlation algorithm,the five-element linear array and a five-element pyramid array are used to locate the azimuth and the three-dimensional coordinates of the far-field fried spots respectively.Combined with the improved MUSIC algorithm,the azimuth estimation of the single-frequency sinusoidal signal moving in the far field is performed by using a five-element linear array.Finally,through the experimental platform built,many experiments were performed in the real environment.The experimental results show that the PHAT generalized cross-correlation algorithm and the improved MUSIC algorithm can achieve better positioning effect on the target sound source in the environment with less noise. |
PDF Full Text Request