Eigenbeam-based acoustic source tracking in noisy reverberant environments

In this work, an adaptive acoustic source tracking algorithm is proposed. It is based on eigenbeams which perform spatial decomposition of the sound field, works in two dimensions for tracking of azimuth and elevation, has low computational complexity, and is robust to noise and room reverberation. The tracking is performed using an adaptive principal component analysis of the particle velocity vector, which points from the acoustic source to the sensor. The particle velocity vector is estimated using a spherical microphone array, and is formed by combining the first-order eigenbeams.