The implement of the paper "Time-domain Speech Enhancement with Generative Adversarial Learning".
Demo files for some enhancement results will be provided in Enhancement result demo
Run train.py, noting to modify the path in train.py. train_WGAN() means train the W-TSEGANs, and train_Metric_GAN means to train the M-TSEGANs.
Run clean.py, noting to modify the path in clean.py.
This project is released under the CC BY-NC-ND 4.0 license.