Go To Program 30 A Parametric Least-Squares Approximation for Multichannel Equalization of Room Acoustics Dominic Schmid 1 Gerald Enzner 1 1 Institute of Communication Acoustics, Ruhr-Universität Bochum, Bochum, Germany Algorithms for multichannel equalization of room acoustics are often derived using compact matrix notation. The corresponding matrix computations are, however, numerically demanding and not feasible on platforms with limited resources. We therefore suggest to transform matrix-based solutions for room equalization back into light-weight digital filters with less resource consumption and guaranteed numerical stability. In particular, we show that broadband least-squares (LS) equalization in matrix form can be efficiently approximated by a matched filter array (MFA) and a subsequent single-channel equalization stage, where the latter utilizes parametric all-pole modeling of the room acoustics. The resulting algorithm, which we term parametric least-squares equalization (PLSE), is highly flexible with respect to the desired equalization accuracy and the use of the available channel information. We finally evaluate the equalization of simulated and real acoustic systems and compare our results against a non-parametric minimum mean-square error (MMSE) bound. View pdf