Acoustic sensing from a multi-rotor drone

Abstract

We propose a time-frequency processing method that localizes and enhances a target sound by exploiting spectral and spatial characteristics of the ego-noise captured by a microphone array mounted on a multi-rotor micro aerial vehicle. We first exploit the time-frequency sparsity of the acoustic signal to estimate at each individual time-frequency bin the local direction of arrival (DOA) of the sound and formulate spatial filters pointing at a set of candidate directions. Then, we combine a kurtosis measure based on the spatial filtering outputs and a histogram measure based on the local DOA estimation to calculate a spatial likelihood function for source localization. Finally, we enhance the target sound by formulating a time-frequency spatial filter pointing at the estimated direction. As the ego-noise generally originates from specific directions, we propose a DOA-weighted spatial likelihood function that improves source localization performance by identifying noiseless sectors in the DOA circle. The DOA weighting scheme localizes the target sound even in extremely low signal-to-noise conditions when the target sound comes from a noiseless sector. We experimentally validate the performance of the proposed method with two array placements.