Physics-Guided Variational Model for Unsupervised Sound Source Tracking
Luan Vinícius Fiorio, Ivana Nikoloska, Bruno Defraene, Alex Young, Johan David, Ronald M. Aarts · 2026
Read Physics-Guided Variational Model for Unsupervised Sound Source Tracking by Luan Vinícius Fiorio, Ivana Nikoloska, Bruno Defraene, Alex Young, Johan David, Ronald M. Aarts (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-2602-08484.txt.
Description
Tracking a sound source usually needs array-processing math or costly labeled positions. Luan Vinícius Fiorio and five co-authors build a model that learns tracking with no labels at all, guided by physics. Their arXiv paper pairs a variational encoder with a decoder that injects geometry through time-delay likelihoods between microphone pairs.
Directions are estimated straight from array signals, with no ground-truth positions in training. Tests on real data show the method beating classical baselines and matching supervised models on accuracy and cost. It also handles mismatched array shapes and corrupted microphone-position data robustly. Readers get an unsupervised tracker that carries its physics with it, plus a sketched path to multiple sources.
Catalog record
| CREATOR | Luan Vinícius Fiorio, Ivana Nikoloska, Bruno Defraene, Alex Young, Johan David, Ronald M. Aarts |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08484