Representing Sounds as Neural Amplitude Fields: A Benchmark of Coordinate-MLPs and A Fourier Kolmogorov-Arnold Framework

Linfei Li, Lin Zhang, Zhong Wang, Fengyi Zhang, Zelin Li, Ying Shen · 2026

Read Representing Sounds as Neural Amplitude Fields: A Benchmark of Coordinate-MLPs and A Fourier Kolmogorov-Arnold Framework by Linfei Li, Lin Zhang, Zhong Wang, Fengyi Zhang, Zelin Li, Ying Shen (2026) 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-2601-06406.txt.

Description

A sound wave can be drawn by a small net that takes a time point and gives the height of the wave. Linfei Li and 5 co-authors test that trick for sound in a 2026 paper. Such nets shine for light fields and 3D shapes. For sound, few had checked which parts make them work.

Their test mixed three ways to code the time point with twice eight switch shapes for the net's nodes. The nets proved picky: they need fine tuning and a start that fits the pitch of the sound. The team's new frame drops the time code. It builds on wave math and an old split rule for hard forms, and it learns high pitches with care so low ones do not swamp the fit. On speech and music, it held up well with far less tuning.

Catalog record

CREATORLinfei Li, Lin Zhang, Zhong Wang, Fengyi Zhang, Zelin Li, Ying Shen
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-06406

Browse the full library · About the Tower of Babel