Least-Distortion Maximum Gain Beamformer for Time-Domain Region-of-Interest Beamforming
Ariel J. Frank, Israel Cohen · IEEE Transactions on Audio Speech and Language Processing · 2025
Region-of-interest (ROI) beamformers are very useful in cases where precise information about the source position is unavailable, such as situations involving estimation errors or source movement. This paper presents an approach to ROI beamforming using convolutive filters in the time domain. The proposed beamformer can focus on a specific spatial region of interest while suppressing interference and noise from other directions. We formulate the signal model, considering a desired source signal propagating from a particular ROI and an array of sensors capturing a convolved version of the signal in some noise field. Appropriate performance measures are introduced to derive and analyze ROI-centric beamformers. The ROI beamformer is designed to maximize the gain in signal-to-noise ratio under a constraint on the signal distortion in the spatial region of interest. Additional parameters are introduced to balance the beamformer's gain, distortion, and robustness. Simulations demonstrate the effectiveness of the proposed method in ROI beamforming and interference suppression.