Sound Source Localization and Tracking using Kalman Filtering

Abstract

Beamforming and Kalman ltering have vast applications in signal processing. It has been used widely in source location estimation algorithms. The rst part of this thesis deals with the estimation of the location of sound source using microphone array processing and beamforming. The algorithm based on power maximization gives raw position estimates of the source. An array is composed of a set of microphones that are placed in certain locations. Microphone array is to give high gain to signal from one desired direction and attenuate the other. This is implemented by maximizing the power of beamformer's output in one speci c direction. Following that, a source tracking system based on the Kalman lter algorithm is proposed. The microphone array approach is employed in the rst step to nding the sound source location. The Kalman ltering method is then used to re ne those estimations. For the motion of the target source, three system models are proposed: motion along with a single coordinate, constant velocity, and constant acceleration model. The outcome of combining these two techniques is a robust system to sound source localization and tracking.