Share Email Print
cover

Proceedings Paper

Repairing highly corrupted speech and images with U-net autoencoders
Author(s): Michael S. Lee; John S. Hyatt; Samuel N. Edwards
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Recovering data from high amounts of loss and corruption would be useful for a wide variety of civilian and military applications. Highly corrupted data (e.g., speech and images) has been less studied relative to the problem of light corruption, but would be advantageous for applications such as low-light imagery and weak signal reception in acoustic sensing and radio communication. Unlike milder signal corruptions, resolving strong noise interference may require a more robust approach than simply removing predictable noise, namely actively looking for the expected signal, a type of problem well suited for machine learning. In this work, we evaluate a variant of the U-net autoencoder neural network topology for accomplishing the difficult task of denoising highly corrupted images and English speech when noise floors are 2-10x stronger than the clean signal. We test our methods on corruptions including additive white Gaussian noise and channel dropout.

Paper Details

Date Published: 21 April 2020
PDF: 12 pages
Proc. SPIE 11413, Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications II, 114130E (21 April 2020);
Show Author Affiliations
Michael S. Lee, CCDC U.S. Army Research Lab. (United States)
John S. Hyatt, CCDC U.S. Army Research Lab. (United States)
Samuel N. Edwards, CCDC U.S. Army Research Lab. (United States)
Parsons, Inc. (United States)


Published in SPIE Proceedings Vol. 11413:
Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications II
Tien Pham; Latasha Solomon; Katie Rainey, Editor(s)

© SPIE. Terms of Use
Back to Top
PREMIUM CONTENT
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?
close_icon_gray