Rebuilding the Statistical Model with Cleaner Voice

Categories Artificial Intelligence, Blog, Computer Science, Machine Learning0 Comments

I’ve been doing some research recently in Machine Learning. Resurrecting my previous “VertoPellis” project and cleaning up the source data we fed the system. Upon building the voice, we fed it speeches which had echos and high ambient noises.

This resulted in a very difficult learning process as well as grainy/dirty output. I’m quickly fixing that before I move to implementing some form of Recurrent Neural Network to speed up speech output.

Stay tuned!