Controlling the Parameterized Multi-channel Wiener Filter using a tiny neural network

2025

Read Controlling the Parameterized Multi-channel Wiener Filter using a tiny neural network (2025) 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-2507-13863.txt.

Description

A team of speech engineers, including Eric Grinstein, Ashutosh Pandey, and Cole Li, wrote a 2025 paper on cleaning speech with a classic filter steered by a tiny neural network. Multi channel noise removal faces a trade off: neural models suppress noise best but often bend the voice itself. The parameterized multi channel Wiener filter, a classical beamformer, exposes a dial for that trade off but needs someone to turn it.

Their system, NeuralPMWF, lets a small, fast network run that dial in real time. The design keeps computation low enough for live devices. In tests, the hybrid beat several competing systems of similar cost on both perceptual and objective quality measures. It pairs the control of signal processing with the judgment of learning.

Catalog record

YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-13863

Browse the full library · About the Tower of Babel