In situ estimation of the acoustic surface impedance using simulation-based inference

Jonas M. Schmid, Johannes D. Schmid, Martin Eser, Steffen Marburg · 2025

Read In situ estimation of the acoustic surface impedance using simulation-based inference by Jonas M. Schmid, Johannes D. Schmid, Martin Eser, Steffen Marburg (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-08873.txt.

Introduction

Jonas M. Schmid and 3 co-authors find the sound traits of room surfaces from sparse measures in a 2025 paper. Sound work in enclosed spaces needs the impedance of walls, a value for how much sound a surface takes in or sends back. Old ways rest on simple views of the sound field.

A Bayesian frame, which weighs odds for each guess, infers impedance from a few pressure measures. Neural nets map fake data straight to odds for the true values, with no slow sampling. Tests on a box room, a tube, and a car cabin all give sound results. Checks show the stated doubt is well matched to real error.

Catalog record

CREATORJonas M. Schmid, Johannes D. Schmid, Martin Eser, Steffen Marburg
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-08873

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

In situ estimation of the acoustic surface impedance using simulation-based inference Jonas M. Schmid, Johannes D. Schmid, Martin Eser, Steffen Marburg Abstract Accurate acoustic simulations of enclosed spaces require precise boundary conditions, typically expressed through surface impedances for wave-based methods. Conventional measurement techniques often rely on simplifying assumptions about the sound field and mounting conditions, limiting their validity for real-world scenarios. To overcome these limitations, this study introduces a Bayesian framework for the in situ estimation of frequency-dependent acoustic surface impedances from sparse interior sound pressure measurements. The approach employs simulation-based inference, which leverages the expressiveness of modern neural network architectures to directly map simulated data to posterior distributions of model parameters, bypassing conventional sampling-based Bayesian approaches and offering advantages for high-dimensional inference problems. Impedance behavior is modeled using a damped oscillator model extended with a fractional calculus term. The framework is verified on a finite element model of a cuboid room and further tested with impedance tube measurements used as reference, achieving robust and accurate estimation of all six individual impedances. Application to a numerical car cabin model further demonstrates reliable uncertainty quantification and high predictive accuracy even for complex-shaped geometries. Posterior predictive checks and coverage diagnostics confirm well-calibrated inference, highlighting the method’s potential for generalizable, efficient, and physically consistent characterization of acoustic boundary conditions in real-world interior environments. I Introduction With advances in computational power and numerical algorithms, wave-based simulation techniques such as the finite element method (FEM) or the boundary element method (BEM) have become essential tools for predicting sound fields in enclosed spaces Prinn.2023 . These methods support acoustic design already in the early development stages and are applied in a wide range of contexts, including music rehearsal and performance spaces, recording studios, lecture halls, office environments, as well as automotive and aircraft cabins Gurbuz.2023 . As emphasized by Aretz and Vorländer Aretz.2014b , the accuracy and robustness of such simulations critically depend on the precise characterization of all acoustically interacting surfaces Vorlander.2013 ; Fratoni.2025 , which is a major source of uncertainties Thydal.2021 . In wave-based models, boundary conditions are commonly described by the complex-valued surface impedance, or equivalently by its reciprocal, the surface admittance Marburg.2008 . These quantities inherently account for both amplitude and phase information, both being indispensable for physically accurate predictions Marburg.1999 . Traditionally, energy-based quantities such as the absorption coefficient have been employed in room acoustics to describe the fraction of acoustic energy absorbed by a boundary. While extensive datasets of random-incidence absorption coefficients are available in literature, these data provide only magnitude, but lack the corresponding phase information. This limitation is particularly critical in small spaces and the low frequency range, where room modes dominate CardosoSoares.2022 . Despite the increasing relevance of wave-based methods in room acoustics, comprehensive datasets of the surface impedance remain scarce Fratoni.2025 . Three different methods have emerged to measure the absorptive properties of a material. Reverberation room method: The reverberation room method, standardized in ISO 354 ISO354.2003 , determines the acoustic absorption coefficient of materials by measuring the change in reverberation time of a diffuse sound field when a test specimen is introduced into a reverberation chamber. This method is widely used for characterizing materials under diffuse-field conditions. However, reverberation chambers are not always available in laboratory settings, and the diffuse-field assumption becomes invalid at low frequencies, leading to significant uncertainties Nolan.2018 . Impedance tube method: The impedance tube method, standardized in ISO 10534-2 ISO105342.1998 , is a well-established method for characterizing the normal-incidence surface impedance. It relies on evaluating the transfer function between two microphones positioned inside a rigid tube, with the test specimen mounted at its termination. Despite its practical utility, the method is inherently restricted to normal incidence of plane waves, which contrasts with real-world acoustic conditions where surfaces are exposed to complex sound fields and waves impinge from multiple directions Eser.2025 . Moreover, the use of small specimen sizes and idealized mounting conditions in the tube often fail to reflect realistic installation scenarios. As demonstrated by Brandão et al. Brandao.2013 , materials that appear to perform adequately in impedance tube measurements may display markedly different acoustic behavior under practical operating conditions. In situ methods: The limitations of the two aforementioned methods highlight the need for in situ methods that allow for non-destructive characterization of materials under realistic mounting conditions, without relying on laboratory setups or assumptions about wave incidence Brandao.2015 . To this end, several approaches have been developed, most of which are formulated as inverse problems. In these methods, sound pressure is measured in the vicinity of the sample, and the surface impedance or absorption coefficient is subsequently estimated by solving an inverse problem based on physical models of wave propagation Tamura.1990 . The sound field can be decomposed into elementary wave functions including plane waves Nolan.2020 and spherical waves Richard.2017 . Microphone arrays enable simultaneous sound pressure measurements at multiple positions and are commonly implemented in different configurations such as double-layered planar arrays Hald.2019 or spherical arrays Richard.2019b . These measurements are typically performed in (semi-)anechoic chambers under free-field conditions to ensure well-defined incidence and to minimize undesired wall reflections Eser.2023 ; Brandao.2011 . More recently, neural network–based approaches have been introduced to compensate for edge diffraction effects caused by the finite size of test samples Emmerich.2025 ; MullerGiebeler.2024 ; Zea.2023 . Alternatively, pressure–particle velocity probes can be employed to directly capture sound pressure and particle velocity in close proximity to the sample surface, with the measured data subsequently combined with the aforementioned techniques to estimate boundary properties BrieredelaHosserayeBaltazar.2022 ; MullerTrapet.2013 . Generally, the accuracy of methods based on wave propagation models strongly depends on the complexity of the selected model and inherent simplifying assumptions, with these limitations becoming particularly pronounced at low frequencies and in geometrically complex environments Luo.2020 . To overcome these limitations, discretization-based methods have been employed in which numerical methods such as the FEM or BEM serve as forward models to compute the sound pressure field in an acoustic domain for given boundary conditions. Such approaches are particularly advantageous for complex geometries and in cases where the domain is well known or a detailed geometric model already exists, for example in digital twin applications or model updating. Dutilleux et al. Dutilleux.2002 have combined a global evolutionary optimization strategy with the finite difference method and FEM to estimate boundary impedances of rectangular acoustic domains in the low-frequency range. Anderssohn and Marburg Anderssohn.2007 have employed a FEM model together with a two-step nonlinear optimization scheme to recover spatially non-uniform boundary admittances from interior sound pressure measurements. Similarly, Luo et al. Luo.2020 have used BEM in conjunction with iterative optimization procedures. Prinn et al. Prinn.2021 have introduced an eigenvalue approximation method combined with optimization to estimate normal incidence surface impedances at modal frequencies for the case of spatially uniform surfaces. More recently, neural network–based approaches have gained attention for estimating acoustic boundary conditions and predicting sound radiation Schmid.2025 . Schmid et al. Schmid.2024 and Xia et al. Xia.2024 have proposed physics-informed neural networks to infer acoustic boundary properties directly from pressure measurements, demonstrating promising accuracy. Nevertheless, these approaches remain limited to simple geometries. Inferring surface impedances from acoustic pressure measurements constitutes an ill-posed inverse problem that is highly

Read the complete text