Implicit neural representations for laryngeal aerodynamics

Hauser SL, Kist A, Döllinger M, Kniesburges S (2025)

Publication Type: Journal article

Publication year: 2025

Journal

Book Volume: 37

Article Number: 081901

Journal Issue: 8

DOI: 10.1063/5.0267402

Abstract

Simulating and representing the phonation process is a computationally intense problem. In this study, we address this issue using implicit neural representations to determine the possibilities of saving computational load by representing computational fluid dynamics simulations through continuous functions represented in a deep neural network. Our work demonstrates the feasibility of using implicit neural representations of a laryngeal aerodynamic simulation containing about 180 × 10⁶ data points within a single neural network. Additionally, we show that with only 20% of the simulated data, we can restore the original resolution with implicit neural representations, showing only nuanced differences compared to the original simulation. We are also confident that with the proposed approach, we can further lower the representation in space and time in future work.

Authors with CRIS profile

How to cite

APA:

Hauser, S.L., Kist, A., Döllinger, M., & Kniesburges, S. (2025). Implicit neural representations for laryngeal aerodynamics. Physics of Fluids, 37(8). https://doi.org/10.1063/5.0267402

MLA:

Hauser, Sophie Louise, et al. "Implicit neural representations for laryngeal aerodynamics." Physics of Fluids 37.8 (2025).

BibTeX: Download