ESSV Konferenz Elektronische Sprachsignalverarbeitung

Title: Implementation of a Superdirective Beamformer based on an Extended Modal Subspace Decomposition

Authors: Martin Eichler, Arild Lacroix


Microphone arrays in combination with a suitable beamforming algorithm provide means to spatially filter signals from a given sound field. One very general beamforming approach is the method of Modal Subspace Decomposition (MSD) which regards the array and the beamformer as an operator applied to a coefficient vector. Based on this operator, an eigenvalue problem is considered the solution of which yields a set of eigen-beampatterns which serve as a basis for the series expansion of the desired beampattern (directivity). From the series expansion coefficients, the optimum filter coefficients for the beamformer filters are calculated. In an earlier work, an extension to MSD was presented which uses a certain class of IIR filters instead of FIR filters, thus making the algorithm superdirective. Hence, even with very compact arrays of arbitrary geometry, it is possible to achieve directivity at low frequencies where the wavelength is large in comparison to the array size. In this contribution, the underlying theory of this approach is outlined and measurement results obtained from a real-time implementation in a real acoustic environment are discussed.

Year: 2009
In session: Signalverarbeitung
Pages: 174 to 181