Time-domain dimensionality reduction design of polynomial wideband beamformer for circular microphone array

The existing frequency-domain polynomial wideband beamformer (PWB) design for circular microphone arrays (CMAs) degrades significantly for short filter tap length due to the effect of the fitting error. To deal with the problem, this paper proposes a highly efficient time-domain dimensionality reduction design approach. Firstly, a symmetry property of the weight coefficients of the PWB for CMAs is theoretically derived. Then a reduced-dimensional form for the time-domain PWB design is deduced by deleting nearly half of the redundant beamformer weight coefficients. Next, an efficient approach is proposed to solve the dimensionality reduction design optimization using the alternating direction method of multipliers. The proposed design outperforms its existing frequency-domain counterpart for short filter tap length, and is also superior for improving the design efficacy.