Solving Room Impulse Response Inverse Problems Using Flow Matching with Analytic Wiener Denoiser

Kyung Yun Lee, Nils Meyer-Kahlen, Sebastian J. Schlecht, Vesa Välimäki · 2026

Read Solving Room Impulse Response Inverse Problems Using Flow Matching with Analytic Wiener Denoiser by Kyung Yun Lee, Nils Meyer-Kahlen, Sebastian J. Schlecht, Vesa Välimäki (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-2602-00652.txt.

Description

Kyung Yun Lee and colleagues present RIRFlow, a training-free Bayesian method for room impulse response problems like denoising and deconvolution, where learned priors need large data sets and generalise poorly. They derive an analytic prior from impulse-response statistics, modelling them as a Gaussian process whose variance decays exponentially. This yields a closed-form Wiener denoiser for minimum mean squared error. That denoiser serves as the prior inside a flow-based inverse solver using guided posterior sampling, extended to nonlinear cases by a local Gaussian approximation.

Catalog record

CREATORKyung Yun Lee, Nils Meyer-Kahlen, Sebastian J. Schlecht, Vesa Välimäki
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62905
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.00652

Browse the full library · About the Tower of Babel