Sound source localization and tracking : real life application using UMA-16

Abstract

This thesis addresses two major challenges in acoustic signal processing: motion parameter estimation using the Doppler effect, and sound source localization with microphone arrays. The first part explores how the instantaneous frequency (IF) shift induced by the Doppler effect enables the estimation of a moving source’s velocity, altitude, and emission frequency using a single microphone, through a closed-form solution validated by simulation. The second part focuses on speaker localization and separation using spatial filtering methods (SRP-PHAT, MVDR, LCMV). A real-time system is implemented with a MiniDSP UMA-16 microphone array and a camera for visual projection.