This paper describes methods for processing signals recorded at a microphone array so as to estimate the signals that would have appeared at the elements of a different, collocated microphone array, i.e., “translating” measurements made at one microphone array to those hypothetically appearing at another array. Two approaches are proposed; a non-parametric method in which a fixed, low-sidelobe beamformer applied to the “source” array drives virtual sources rendered on the “target” array, and a parametric technique in which constrained beamformers are used to estimate source directions, with the sources extracted and rendered to the estimated directions. Finally, a hybrid method is proposed, which combines both approaches so that the extracted point sources and residual can be separately rendered. Experimental results using an array of 2mm-diameter microphones and human HRTFs are reported as a simple example.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.