LMFCA-Net: A Lightweight Model for Multi-Channel Speech Enhancement with Efficient Narrow-Band and Cross-Band Attention
2025
Read LMFCA-Net: A Lightweight Model for Multi-Channel Speech Enhancement with Efficient Narrow-Band and Cross-Band Attention (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-2502-11462.txt.
Description
Yaokai Zhang and co-authors present a light speech cleanup network in a 2025 paper called LMFCA-Net. Multi channel systems can lift speech from noise well, but they often need heavy compute that small devices lack. The new design splits its focus work along time and along pitch bands to catch both near and far links in sound.
Two lean attention blocks replace the loop based units that slow older models down. Tests show the network cleans speech about as well as top rivals while cutting compute cost and delay by a wide margin. That trade makes it a strong fit for phones, hearing gear, and other small tools used in daily life.
Catalog record
| YEAR | 2025 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-11462