Performance bounds for source localization using tetrahedral microphones

Abstract

The Cramér-Rao Bound (CRB) quantifies the minimum variance on a given unbiased estimation parameter. It can be used to derive a lower bound on source localization error for arrays, of which there are several probabilistic models in the literature. However, existing work only considers source localization using arrays of omnidirectional sensors. This presentation describes a model of the CRB as a minimum variance estimation of localization using multiple tetrahedral microphones which have cardioid microphones. Tetrahedral microphones are commonly used to record soundfields for virtual audio reproduction. The proposed method uses the cross spectral density of signals in order to calculate the CRB. This allows for more signal parameters to be included in the model. The model of the CRB was validated with an experiment in an anechoic room with a single sound source. Experimental results show the CRB forms a lower bound at low signal-to-noise ratios.