A multipole model-based boundary admittance estimation for wave-based room-acoustic simulations using Bayesian inference

Abstract

Room-acoustic simulations for sound wave propagation require significant knowledge of the boundary conditions for the room boundary surface. This work presents the application of model-based Bayesian inference on the surface admittance estimation problem. First, this work deals with selecting the multipole admittance model through the higher level of inference, Bayesian model selection. And the parameters of the chosen multipole admittance model are estimated through the first (lower) level of inference, Bayesian parameter estimation. This work focuses on the approximation of the surface admittance as a function of frequency given a set of acoustic admittance. This frequency-dependent admittance is deduced either from the experimental measurements of a porous material or numerical predictions in order to demonstrate wave-based boundary conditions to incorporate arbitrary admittance functions with frequency ranges under consideration. According to the analytical results and numerical verifications conducted in this work, Bayesian inference based on the multipole model is well-suited for the estimation of the frequency-dependent boundary conditions within a wave-based simulation framework.