Compressive beamforming for moving sound source auralization

Conventional beamforming (CBF) is currently widely used for sound source localization. However, the quality of the results obtained with CBF is compromised by the fact that the CBF's output does not represent the source only, but the convolution of the source and the array's point spread function. Furthermore, when it comes to source signal reconstruction, CBF delivers only spectral amplitude information, i.e., there is no phase information, which is essential in the process of auralization.Compressive beamforming (CB) has been recently proposed to improve the resolution of array measurements. CB has the advantage that it not only may improve source localization ability, but also allows the reconstruction of the source signal. Different from CBF, CB performs well on very few data snapshots and even on single-snapshot data, a characteristic which is beneficial when dealing with moving sound sources. This paper utilizes compressive beamforming to first localize multiple moving sound sources, and then reconstruct the source signals. To conclude we compare results obtained with CBF and CB.