Microphone Array for Headset with Spatial Noise Suppressor

This paper presents a hybrid microphone array architecture used for processing the signals from a small microphone array that is used in a headset. The processing chain consists of fixed end-fire beamforming, adaptive spatial noise reduction and stationary noise suppression. The far-field design algorithm used for the fixed beamformer is adapted to the specifics of the headset by compensating for the directivity of the mouth and the sound diffraction around the head. The spatial noise reduction generalizes the suppression rule for optimal MMSE power noise reduction to multiple dimensions. The algorithm was tested with a headset that used a three element microphone array. It is computationally effective and provides up to 18 dB of ambient noise suppression.