Adaptive Microphone Beamforming

Frequency-invariant beamforming (FIB) plays a crucial role in distortion-free acquisition and processing of speech signals using microphone arrays. While classical FIB design methods typically operate under far-field plane wave assumptions, this paper introduces a novel approach based on the near-field spherical wave model. We integrate the spatial response variation (SRV) function into near-field Linear Constrained Minimum Variance (LCMV) broadband beamforming, formulating the spatial response as an optimization constraint and deriving a closed-form solution via Lagrange multipliers. The implementation involves calculating frequency-invariant weight vectors through matrix inversion operations, with key computational steps including covariance matrix estimation and constraint matrix construction. This method enhances robustness against environmental interference compared to conventional approaches.