Study Of Sound Source Localization Based On Microphone Array Technology

At present, the sound source localization technology usually applied the method ofTDOA to estimate the position of the sound, but its’ localization accuracy is not high.However, on some occasion, such as the huge meeting hall, and there are many peoplewhose distance is very near talk at the same time, the high localization accuracy to thosepeople is very important to ensure their sound obtain very clearly. Therefore, the highlocalization of speech is a valuable study.In this paper, based on the large and medium-sized conference hall, indoor formultiple close people to high accuracy positioning as the research background, it startsfrom a research for the purpose of improving the precision of the sound sourcelocalization.The main work and innovation points summarized as the following:1. The simulation analysis of speech signal preprocessing, pre filtering, the purposeis to non stationary and broadband voice signal into a smooth and narrow-band signal,to facilitate the follow-up positioning algorithm to deal with. And analysis summarizesthe far and near field sound source signal model, The space incident signal model isestablished for subsequent sound source signal.2. Summarizing the microphone array, several kinds of basic topology structure ofthe space information: uniform line array, uniform circular array, cross array, arbitraryarray. Theoretically influence space of the positioning parameters is presented.Comprehensive easy to implement and locating effect two aspects factors, uniformlinear array and the uniform circular array is chosen as the study of basic topologystructure. At the same time, the qualitative and quantitative analysis of the error arrayelement spacing and the external environment impact on sound source localizationaccuracy, provide a basis for analysis in later chapters.3. In order to satisfy the large conference room, indoor the speaker more accuratelocation, at the same time, also considered positioning position distortion caused byfuzzy. This paper proposes has two different array elements spacing of the microphonearray positioning method. The method respectively using a linear combination of the different array element spacing, extract receives a frame after optimization of the soundsource signal, twice for the same sound source localization. The localization isdependent on the previous results and set parameters, using sound MUSIC algorithm forprecise positioning. Through the simulation analysis, the method not only makes thelocation accuracy has been effectively improved, at the same time, within easy reach ofthe space two sound source can achieve very accurate resolution. In addition, in order toeasy to follow-up studies, this paper set up the method of simulation test system, andfurther verifies the practical value of this method.4. Mike topology structure based on the above, the error of secondary sound sourcelocalization algorithm in environment, the positioning accuracy at the same time maycause inconsistent positioning results and actual position of the probability increases,decreases the reliability of the positioning results. Based on microphone arraycalibration theory, therefore, this article finally proposed above secondary localizationalgorithm to do the further correction, namely the ideal case, the localization of spatialstructure multiplied by a linear error correction function, which makes the receivedsignal is more close to the actual microphone array receiving sound source signal, so asto improve the reliability of positioning results. At the end of the paper after correctionis verified by the simulation can achieve better location performance.