Study of a generalized spherical array beamformer with adjustable binaural reproduction

Microphone array processing and beamforming methods are widely used to enhance speech signals in noisy and reverberant environments. Recently, a generalized spherical array beamformer (GSB) was introduced incorporating beamforming with binaural sound reproduction. This GSB improves both the spatial realism and the speech intelligibility in the reproduced acoustic scene. In this paper the limitations and the trade-offs that result from incorporating both beamforming and binaural reproduction in a single array are investigated both theoretically and experimentally. Results show a clear trade-off between binaural reproduction and spatial selectivity performance, such that improving one is only possible at the expense of the other. These results lead to the proposal of the design of an adjustable generalized beamformer, which is suggested as a future research direction with the aim of enabling the GSB to adapt to changing environments.