Study On Tracking Algorithm Of Distributed Microphone Arrays

With the development of the computer technology and multi-sensor networks, distributed systems started to become a research hotspot in recent years. In speech processing, because of the presence of noise and reverberation, a single pickup tool of speech greatly reduces the quality of voice communications. So microphone array began to rise, and in the past three decades array technology have been developed significantly. The microphone array combined with distributed system, this paper made the following areas:(1)To study the status and methods of positioning and tracking technology, introduce briefly Bayesian resampling technique and theory, detail the Kalman filter and particle filter theory.(2)Details of the system synchronization, node localization and data fusion of the distributed system, and introduce the development status of these problems. Details federated filter and distributed particle filter which can be used in distributed system.(3)Details of how to use particle filter in tracking of speaker, including which model can be used, what localization function to use and how to calculate the likelihood function.(4)In the experiment, using GCC-PHAT method to get a time delay. And gives the tracking results of using standard particle filter and the tracking result of using distributed particle filter, compare results and conclusions are given.