Physics-Informed Machine Learning For Sound Field Estimation

Shoichi Koyama,Juliano G. C. Ribeiro,Tomohiko Nakamura,Natsuki Ueno,Mirco Pezzoli

2024-08-27

Abstract:The area of study concerning the estimation of spatial sound, i.e., the distribution of a physical quantity of sound such as acoustic pressure, is called sound field estimation, which is the basis for various applied technologies related to spatial audio processing. The sound field estimation problem is formulated as a function interpolation problem in machine learning in a simplified scenario. However, high estimation performance cannot be expected by simply applying general interpolation techniques that rely only on data. The physical properties of sound fields are useful a priori information, and it is considered extremely important to incorporate them into the estimation. In this article, we introduce the fundamentals of physics-informed machine learning (PIML) for sound field estimation and overview current PIML-based sound field estimation methods.

Sound,Audio and Speech Processing

What problem does this paper attempt to address?

The paper is primarily dedicated to addressing the problem of sound field estimation, which involves reconstructing a spatial sound field from a set of discrete microphone measurements. Specifically, sound field estimation techniques are crucial in various applications, including room acoustics analysis, sound field visualization/auditoryization, spatial audio reproduction using speaker arrays or headphones, and active noise cancellation in specific areas. In recent years, VR/AR audio has become a significant application of this technology, as it requires capturing the sound field over a large area using multiple microphones. The paper points out that traditional physics-based methods, while providing accurate estimates, have faced challenges with the rise of data-driven methods. Combining physical prior knowledge with machine learning algorithms to improve estimation accuracy has become a research hotspot. To this end, researchers have introduced Physics-Informed Machine Learning (PIML), particularly Physics-Informed Neural Networks (PINNs), which incorporate physical equations as part of the loss function to constrain the model. This approach prevents unnatural distortions in the estimated results that do not conform to the actual sound field, especially when observational data is sparse. In summary, the paper aims to introduce the basic principles of Physics-Informed Machine Learning and provide an overview of current PIML-based sound field estimation methods. It focuses on how to embed physical properties into different machine learning techniques, such as linear regression, kernel regression, and neural networks. The paper also summarizes the current PIML-based sound field estimation methods, their limitations, and future prospects.

Physics-Informed Machine Learning For Sound Field Estimation

Physics Based Sparsity Level Determination for Acoustic Scattered Far-Field Prediction.

Acoustic Field Visualization and Source Localization Via Physics-Informed Learning of Sparse Data with Adaptive Sampling

A Variational Bayesian Approximation Approach Via A Sparsity Enforcing Prior In Acoustic Imaging

Proceedings of Meetings on Acoustics

Sound Field Estimation around a Rigid Sphere with Physics-informed Neural Network

An Efficient Variational Bayesian Inference Approach Via Studient's-t Priors for Acoustic Imaging in Colored Noises

Kernel Learning for Sound Field Estimation with L1 and L2 Regularizations

Physics-informed neural networks for one-dimensional sound field predictions with parameterized sources and impedance boundaries

Physics-Informed Neural Network for Volumetric Sound field Reconstruction of Speech Signals

Implicit neural representation with physics-informed neural networks for the reconstruction of the early part of room impulse responses

Room impulse response reconstruction with physics-informed deep learning

Mapping ambient sound levels using physics-informed machine learning

Physics-Informed Machine Learning: A Survey on Problems, Methods and Applications

Bayesian inference and neural estimation of acoustic wave propagation

When Physics Meets Machine Learning: A Survey of Physics-Informed Machine Learning

Kernel Interpolation of Incident Sound Field in Region Including Scattering Objects

Spatial domain decomposition-based physics-informed neural networks for practical acoustic propagation estimation under ocean dynamics

Noise-aware Physics-informed Machine Learning for Robust PDE Discovery

Physics-informed machine learning