Deep Learning for Text to Speech

It would be really fun to have two language models talk to each other on a fake phone conversation