Indoor Sound Source Localization Based On Microphone Array Topology And Array Calibration

In recent years,with the continuous development of positioning technology,the outdoor positioning technology represented by the Global Navigation Satellite System(GNSS)has become increasingly mature.However,in indoor positioning,due to the interference of the complex indoor environment and the occlusion of the building wall,the GPS positioning accuracy is seriously reduced.Therefore,it is meaningful to study indoor positioning.Indoor sound source localization has become an important part of indoor localization research because it does not require specific signal transceiving equipment.The microphone array has become a research hotspot for indoor sound source localization because it can obtain not only the time domain information of the sound signal but also the space domain information of the sound signal,which can further improve the positioning accuracy.This master thesis analyzes the advantages and disadvantages of the Steer Response Power-Phase Transform(SRP-PHAT)localization algorithm for indoor sound source localization,which is usually only applicable to large aperture microphone arrays and near In the field sound source localization,and ignoring the problem of the influence of array topology and array calibration on indoor sound source localization accuracy,we propose a two-stage filter fusion indoor sound source localization system based on three-dimensional star array topology.Firstly,based on this system,we analyzed the influence of the microphone array topology of different dimensions on the positioning accuracy of the indoor sound source.Comprehensively comparing the positioning accuracy and algorithm calculation amount under different array topologies,we have found a set of small aperture microphone array types and parameter settings suitable for indoor sound source positioning.Secondly,we analyzed the frequency response consistency of the signal received by each microphone.In order to solve the problem that the nonlinear frequency response difference between each microphone will affect the positioning accuracy,we proposed a two-stage filter fusion mechanism based on multi-channel low-pass filtering and multi-channel adaptive filtering.And we analyzed the relationship between the difference of the frequency response curve of each microphone before and after filtering and the positioning accuracy through simulation experiments and actual measurement experiments.The actual test results show that when the distance between the sound source and the array is within 3m,the average positioning error of the system is about 12 cm,and the average positioning accuracy is significantly better than the traditional SRP-PHAT method and GCC-q?? method.When the distance between the sound source and the array reaches 6m,the positioning accuracy of this system is the same as that of the GCC-q?? method.Compared with the traditional SPR-PHAT method,the positioning error is reduced by 48%.