Optimization-based method for the calibration of geometrical acoustic models

Uncertainty of the final results of room acoustic modelling is a result of uncertainties of many model input parameters, which have to be discovered and limited if possible. It can be done by model calibration. In this paper, a detailed procedure of room acoustic model calibration is proposed. It is based on detection and limitation of possible sources of discrepancies between measured and simulated selected room acoustic parameters defined in ISO 3382-1 standard. As independent parameters which describe the variability of the acoustic field in the room the most accurately, clarity C-80 and early decay time EDT are suggested. Based on uncertainty analysis, sound absorption coefficients of all materials used in the interior and selected sound scattering coefficients as well as positions of the receivers are chosen to be input parameters adjusted in the model calibration process. Correction of these parameters is done using optimization algorithms in order to accelerate the procedure. Methods for choosing key ray-tracing simulation parameters are presented, according to which it is possible to obtain repeatable results in the shortest possible time. All analyses are done for five different interiors (three philharmonics and two churches). Correctness of sound absorption/scattering coefficients calculated in the calibration process is validated in the comparison of simulation and measurement results, using a different sound source position. Obtained results are specific for analysed rooms, but proposed methodology is universal for all kinds of room acoustic models. (C) 2020 Published by Elsevier Ltd.