Whenever room acoustical simulations are compared to room acoustical measurements, significant differences are obvious. Results of previous studies in this field indicated that errors may particularly result from neglecting wave effects, namely of surface scattering, diffraction, or the attenuation of sound at grazing incidence over the audience or seating. Differences can also be related to uncertainties with respect to the boundary conditions of the simulation, i.e. the absorption and scattering coefficients of the walls.
In order to reliably identify shortcomings in sound propagation algorithms for acoustical environments, a Ground Truth database of Room Acoustical Simulation (GRAS) with knowledge of the room geometry, the boundary conditions, and with single-channel, and binaural sound field recordings is given as a reference for their evaluation. The database includes simple scenes such as ‘single reflection’, or ‘diffraction’ that isolate single acoustic phenomena, as well as complex rooms of three different sizes ranging from 200 to 8000 cubic meters volume.
Similar to the first round robins on room acoustical simulation, CAD models, and wall properties will be provided to the developers of room simulation software, along with defined source and receiver positions and data for the source directivity. Developers were asked to submit single-channel and binaural simulations using the original data provided (geometry, surface properties). The analysis of the different simulation results included a physical evaluation based on the impulse responses, and extracted room acoustical parameters, as well as a perceptual evaluation using the items from the Spatial Audio Quality Inventory (SAQI) that were designed for the evaluation of virtual and real acoustical environments.