Gradient flow source localization in noisy and reverberant environments

Gradient flow is a technique for localization of an acoustic source using miniature microphone arrays by relating temporal and spatial gradients of the impinging source signal. The performance of the gradient flow in noisy and reverberant environment is examined and quantified through simulations that incorporate additive measurement noise, directional interference signal and room acoustic model. The algorithm demonstrates robust performance for additive signal-to-noise ratio down to 5 dB and at the signal-to-interference ratio of 10 dB. In the echoic room, localization performance starts to deteriorate under moderate reverberation conditions. The experimental results from a miniature microphone array recordings in a conference-room environment verify the presented simulations.