A audio-to-gif GAN

Thie repository implement a audio-visual GAN model.

Our code is based of StackGAN reference code https://github.com/hanzhanggit/StackGAN-Pytorch/tree/master/code

What we implement/modified:

1. GAN:
   • Modify StackGAN to work with new data sources
   • Implement WGAN options in StackGAN
   • Implement two-stream network in StackGAN
2. Audio:
   • All data preprocessing, feature extract
   • Implement and train AudioSet embedding
3. Other:
   • Evaluation code for incpetion score + Fine tue VGG net for inception score
4. Training:
   • Trained a lot of GANs ..... on both AudioSet and COCO (for testing) .