Quality assessment of room acoustic simulation tools by comparing binaural measurements and simulations in an optimized test scenario

Latest room acoustic (RA) simulation tools use the powerful combination of geometrical acoustics (GA) for higher frequencies with the finite-element-method (FE) for the lower end. This hybrid approach has the potential to provide a highly accurate agreement between measured and simulated results, even for small spaces where wave effects play an important role at lower frequencies. However, inevitable uncertainties in the characterization of the complex acoustic behavior of real-world sources and room materials make a perfect perceptual match between simulated and measured auralizations almost unachievable in an ordinary room. In order to still benchmark the quality of a room acoustic simulation tool based on a comparison with measured results, a suitable test room is therefore required. In this study we therefore use a reverberation chamber room with controlled acoustic boundary and source conditions to compare measured and simulated monaural and binaural room impulse responses (RIR), both objectively and by subjective listening test. All simulations are conducted using state-of-the-art FE and GA simulation tools developed at ITA of RWTH Aachen University. The binaural measurements were done using the ITA artificial head. The best possible elimination of any inaccuracies caused by geometry simplifications, source or material uncertainties or the use of different HRTFs, enables an unbiased quality assessment of the simulation algorithms themselves. Thus the suggested benchmark test provides a measure of how close a simulation can converge to the actual measurement in an ideal case.