Research On The Information Gathering And Preprocessing Of Sound Source Localization System

The Sound Source Localization(SSL) is one of the most important techniques which can be widely used in military field in snipersystem, sentinel system, smart bombs and so on, for information collecting and probing. In the civil area, SSL is applied to video conference, intelligent robot, speech enhancement and so on. The research of SSL in a real sound field is a challenging topic and interests has been shown in this area.This dissertation mainly focuses on one of the key technology in SSL, namely, information collection and preprocessing which include the design of signal acquisition circuit and study of Time Delay Estimation(TDE) algorithm as well. Firstly, we designed the signal acquisition circuit. Based on the analysis of characteristics of sound source signal, the performance index of common microphones and signal regulate circuit, we employed CHZ-402, which is a capacitor type electret microphone, as acoustic sensor. In addtion, we also designed a corresponding pre-amplifier circuit, a filter circuit and a main-amplifier circuit. Secondly, several important TDE algorithms were introduced, including TDE algorithm based on correlation analysis; TDE algorithm based on phase data analysis; TDE algorithm based on adapted filtering scheme and TDE algorithm based on higher order spectrum analysis. Then, we discussed the relationship among them based on both theoretical analysis and computer simulation. Thirdly, we analysed the generalized cross-correlation method(GCC) and proposed several useful interpolation methods based on the results of Matlab simulation. After the analysis of adaptive TDE algorithm, we discussed the algorithm influence of sliding window. Finally, a comparesion of the applicable conditions of several TDE algorithm was presented and GCC was chosen as a kind of algorithm which can be applied to experiments.Based on the study of TDE algorithm simulation in Matlab, we proposed an experiment to estimate time delay of real sound sources and analysed factors which can effect estimate accuracy. The experimental results show that GCC method is a feasible one for TDE.