A data-driven two-microphone method for in-situ sound absorption measurements

TL;DR

The study tackles in situ estimation of angle-dependent sound absorption for finite porous samples by augmenting the classic two-microphone method with a data-driven model. A 1D residual neural network maps the complex transfer function between two microphones and the source elevation to the absorption spectrum α(f) of an effectively infinite slab, trained exclusively on large-scale Boundary Element Method simulations based on the Delany-Bazley-Miki model. Numerical validation shows substantially reduced prediction error compared with the traditional method, and experimental validation across eight setups confirms the approach can reproduce or closely approximate impedance-tube and DBM references while mitigating edge-diffraction artifacts. The work enables practical, in situ estimation of acoustic absorption in realistic conditions and suggests avenues for fine-tuning with experimental data to further enhance robustness and generalization.

Abstract

This work presents a data-driven approach to estimating the sound absorption coefficient of an infinite porous slab using a neural network and a two-microphone measurement on a finite porous sample. A 1D-convolutional network predicts the sound absorption coefficient from the complex-valued transfer function between the sound pressure measured at the two microphone positions. The network is trained and validated with numerical data generated by a boundary element model using the Delany-Bazley-Miki model, demonstrating accurate predictions for various numerical samples. The method is experimentally validated with baffled rectangular samples of a fibrous material, where sample size and source height are varied. The results show that the neural network offers the possibility to reliably predict the in-situ sound absorption of a porous material using the traditional two-microphone method as if the sample were infinite. The normal-incidence sound absorption coefficient obtained by the network compares well with that obtained theoretically and in an impedance tube. The proposed method has promising perspectives for estimating the sound absorption coefficient of acoustic materials after installation and in realistic operational conditions.

Figures (8)

• Figure 1: Schematic of the two-microphone method above a baffled porous layer, flush-mounted with a rigid backing.
• Figure 2: Experimental setup of the two-microphone method in the anechoic chamber at the Marcus Wallenberg Laboratory for Sound and Vibration in Stockholm.
• Figure 3: Schematic of the network architecture. The total number of trainable parameters is 406300.
• Figure 4: Training and validation loss over $125$ training epochs.
• Figure 5: Histograms of the recorded error per sample for the two-microphone method and for the neural network predictions using the numerical test set.