Robust Beamforming for Multispeaker Audio Conferencing Under DOA Uncertainty

This paper presents a robust microphone array beamforming approach specifically designed for multispeaker audio conferencing, where the directions of arrival (DOAs) of the speakers can vary. First, we address the configuration of the array geometry. To achieve a consistent spatial response across all potential directions on the x-y plane, we propose a hybrid array geometry that combines a concentric circular array (CCA) on the x-y plane with a uniform linear array (ULA) along the z-axis. We discuss this geometry in detail and provide guidelines for selecting its parameters. Next, we focus on the beamforming strategy and introduce a method that directly controls the distortion level for signals originating from a specified near-end region (NER). In this context, we develop three high-directivity beamformers and evaluate their effectiveness based on white noise gain (WNG), directivity factor (DF), and beampattern characteristics. The proposed beamforming approach is experimentally assessed under various reverberant conditions, including both static and moving speech sources. Our results show that it outperforms existing methods in terms of perceptual evaluation of speech quality (PESQ) and short-time objective intelligibility (STOI) metrics. The improved performance is particularly pronounced in scenarios where the desired source direction differs from the nominal steering direction of the array, as well as when the reverberation level is mild.