Research On Microphone Array Acoustic Localization And Speech Enhancement

As traditional sound picking up tool, single microphone is not competent for noise processing, speech source localization and tracking, sound extraction and separation, etc., affecting the qualities of sound communication seriously. If using multi-microphone as a microphone array, to do space-time signal processing, the signal-to-noise ratio should be increased greatly, and the" shortages mentioned above should be overcome. So, the research about microphone array is becoming a hotspot in the field of array sound signal processing now days.Microphone array can be employed for acoustic localization and speech enhancement, and used in videoconference, voice check, hearing aid, hand-free phone in car and so on. Classical theories about array signal processing, applied in the field of radar and sonar, are so perfect that they can be used for dealing with microphone array signal processing conveniently only after some revision.Take the first step from the application background, this paper discusses the source localization based on microphone array, and analyses three kinds of correlation algorithms. Then, this thesis is focused on one of them-an arithmetic so-called MUSIC, which is revised according to the characteristics of speech signal as the input of the microphone array. The revised arithmetic can be applied on location the speech source in the near-field. Meanwhile, a new array signal processing algorithm, which is based on the T-shaped array near-field speech source three-dimension localization, is introduced in details. It is able to raise the accuracy of location while reducing the computational complexity of the arithmetic in the same time.With the information provided by the speech source location, two algorithm of speech enhancement which called Delay and Sum Beam Forming (DSBF) and Generalized Sidelobe Canceller (GSC) are introduced. After the above work, we accomplished the design of the system.