Probabilistic determination of frequency-dependent model parameters in acoustic material characterization (en)
* Presenting author
Abstract:
The sound absorption properties of a room’s boundaries are a significant source of uncertainty in room acoustics. In essence, acoustic material characterization deals with the determination of these sound absorption properties. However, all characterization approaches are also subject to numerous sources of uncertainty. This paper summarizes the application of an inverse probabilistic material characterization approach to assess the uncertain sound absorption properties of porous materials. The approach is based on a Bayesian formulation and uses a sequential transfer of information between adjacent discrete frequencies to enhance the inference’s performance. The inverse material characterization procedure is demonstrated for three scenarios: First, the uncertain and frequency-dependent sound propagation characteristics of a porous material are deduced from impedance tube measurements. Second, the frequency-dependent surface impedance of a locally reacting porous layer is estimated from free-field impedance measurements under normal incidence. Third, the frequency-dependent sound propagation characteristics of a non-locally reacting porous layer are estimated based on a multi-observation setup of free-field impedance measurements accounting for different angles of incidence. In all cases, the frequency-dependent parameters and the corresponding uncertainty are correctly estimated. The framework can readily be transferred to determine uncertain and frequency-dependent model parameters beyond the field of acoustic material characterization.