Experiments on Multiple-point Room Equalization Applied to Medium-sized Enclosed Spaces

Several solutions for multiple-point acoustic magnitude equalization of medium-sized enclosed spaces are analyzed comparatively. The main design steps are discussed: derivation of the room acoustic prototype, an average of the transfer functions obtained for individual listening points; synthesis of the equalizing filter; characterization of the resulting equalized acoustic chain. First, two approaches for designing the equalizer are presented: a classical synthesis method, which employs the Levinson–Durbin algorithm (LD), and a newer one, based on a genetic algorithm (GA) tailored for the synthesis of FIR filters. Frequency warping able to “stretch” the narrow low-frequency octaves is also considered. A novel experimental setup is proposed to ensure that the equalized acoustic chain’s characterization is performed at the same signal level as the un-equalized chain. A new Figure-of-Merit is introduced that allows for a simple yet intuitive comparison of various equalizing solutions considering both the frequency characteristic and the spatial spread between multiple audition points. Finally, experimental data are presented and analyzed: four equalization solutions (LD- or GA-based, with or without frequency warping) were derived for each of three medium-sized enclosed spaces, with audition areas between 36 and 105 m 2 , and reverberation times ranging from 0.7 to 3.8 s.