Research Of Sound Source Location Techniques Based On Microphone Arrays

Sound source location based on microphone arrays has been widely applied in allkinds of communication circumstance. How to construct a small-sized sound sourcelocation system with low cost, simple structure and high estimation precision of directionof arrival (DOA) has been received a growing interest and become a very hot area.Multiple Signal Classification method has been widely used in the large scale system ofsound source location, but its application in the small scale system was restricted by itslarge amount of arithmetic. So the paper focuses on this method and make it can beimplemented on MCU by improving the Multiple Signal Classification method.Firstly, simply introduces three kinds of existing sound source localization methodsbased on microphone arrays, and analyzes their advantages and disadvantages.Secondly, detailed discusses the far-field models of microphone array receivedsignals with narrow band. Then introduces Multiple Signal Classification method andimproved svdfree method. Besides, verifies the validity of the svdfree method bysimulating.Thirdly, in view of the requirement of Multiple Signal Classification method thatmulti-channel signals must be sampled synchronously, makes a research of a method forcompensating the non-synchronous sampling error. Then, makes a comparison withre-modeling method.Then, based on STC5A16AD MCU experimental plate, a real-time single soundsource localization system is designed. The critical section of the software system are theimproved svdfree method and the before and back sampling compansation method.Finally, does a great amount of experiment test, the experimental results show thatthe localization algorithm of this system has simple implementation, little computationand high accuracy. So it can be used in real time positioning system.