CMSC 660 Project Solutions Optimization methods for Sound Source Localization using Microphone arrays

Microphone arrays are widely employed for applications like teleconferencing, high quality sound capture, speaker recognition/identification, acoustic surveillance, head aid devices, speech acquisition in automobile environments etc. For all these applications the benefits that a microphone array provides over a single microphone are two fold. First using a microphone array we can localize a sound source and track its position accurately. The second benefit is that once the source location is known the microphone array can be electronically steered to the source providing spatial filtering. So the essential requirement for all these applications is the ability of the microphone array to locate a speech or sound source accurately. Broadly three types of methods exist for localizing a sound source: Focalization using a steered beamformer, High resolution spectral-estimation methods and Time Difference of Arrival (TDOA) based methods. Most commonly used method in practice is the TDOA based method and we will explore the optimization methods and other scientific computing techniques that can be applied to this method. The TDOA method can be summarized as follows. In this assignment we assume that the sound source is a point source and the microphones we use have omni-directional receiving pattern(although realistic modelling can be done). So there is a delay between the sound wave received by a pair of microphones. So Sound source localization is a two step problem.