/audiogan

GAN for (raw) audio generation

Primary LanguagePython

Speech generation using GAN (work in progress)

Current code is in PyTorch, and only the following files are relevant.

  • audiogan.py
  • dataset.py
  • timer.py
  • preprocess-fisher.py for preprocessing the Fisher English Training corpus.

Other files are obsolete and they are only kept for references (we used WGAN-GP before but we reverted to classical GAN now).