Acoustic seminar | 20 February 2026

 
Stepan Malykh
Chebyshev Research Center
Noise Reduction: Contemporary Methods and Lessons Learned from Training a Neural Denoiser
Abstract

An overview of modern noise reduction techniques. We present and compare mask-based, waveform-based, and generative neural models, discuss evaluation pitfalls, and share empirical findings from developing and testing a streaming neural denoiser.

https://arxiv.org/abs/2205.05474
https://arxiv.org/abs/2006.12847
https://arxiv.org/abs/2204.05841