Speech Generator

Published:

This is an implementation of word-level language modeling. The texts are trained for 64 epoches from ~30 formal speeches (200kb). There are 6645 unique words in the dictionary.
Built with tensorflow
Source Code