Towards Superdirective Beamforming with Loudspeaker Arrays

ABSTRACT In this paper, the superdirective beamforming principle as applied to loudspeaker arrays is investigated. Novel superdirectional beamformer designs (e.g. by Parra, [1]) offer significant spatial selectivity at low frequencies and allow for a small array aperture and some variability in loudspeaker positioning. The theoretical gains of superdirectional beamforming are generally difficult to obtain in practice due to the extreme sensitivity of these beamformers to small random errors and the amplification of spatially white noise, such as transducer self-noise. Passive loudspeakers, unlike microphones, do not suffer from self-noise, but their directivity patterns are usually not well defined, so that deviations between different speakers are common, which act similar to random errors. The results of a succinct sensitivity analysis of an exemplary design [1] are shown in this paper. The use of matched loudspeaker arrays promises an improvement in the performance of superdirectional loudspeaker arrays.