Study On Distributed Microphone Array Localization Method

The microphone array consists of multiple microphones arranged in a certain distribution. When the microphones collect and process sound signals at the same time, not only the time domain and frequency domain information, but also the azimuth information of the sound can be obtained. Sound source localization techniques based on microphone array have been widely used in military detection, security monitoring, human-computer interaction, and many other fields. In recent years, with the development of network communications and mobile computing technology, the microphone array has developed from rule structure into distributed architecture. At present, how to estimate the location of the sound source based on distributed microphone array has become an emerging research topic, which also has practical significance. This thesis studies distributed microphone array sound source localization and array calibration method, and the major work has been done as follows:(1) By combining ranging methods based on time-delay estimation and based on acoustic energy decay model, designing and placing the sound sources ingeniously, the location of each element in the distributed microphone array can be determined. This method can be used for real-time calibration of the microphone array.(2) The thesis takes fingerprint localization method into the distributed microphone array, and gives out the localization fingerprint based sound source localization program for distributed microphone array.(3) The thesis implements fingerprint based sound source localization using deterministic methods, probabilistic methods, and neural networks. On the basis of it, the nearest neighbor and weighted K-nearest neighbor switching method, and weighted K nearest neighbor with dynamic K are proposed to improve the localization accuracy. Besides, the weight used in probability weighted method is modified, and three kinds of weight are discussed.(4) Gaussian mixture regression is also taken into fingerprint based localization method, and it is made use of to set up a non-linear mapping relationship between the energy ratio and sound source location. The simulation results proved the effectiveness of this method.