Skip to main content
. Author manuscript; available in PMC: 2019 Dec 31.
Published in final edited form as: Proc IEEE Int Conf Acoust Speech Signal Process. 2018 Sep 13;2018:6677–6681. doi: 10.1109/ICASSP.2018.8461556

Fig. 1:

Fig. 1:

Schematic representation of the employed U-net architecture. Each blue box represents a feature map, whereas while boxes denote copied feature maps. The arrows stand for the different operations.