
Speech Synthesis for the Chuvash Language with Deep Neural Networks
I trained a few popular speech synthesizers to produce intelligible Chuvash audio clips from written samples of the language. I used Python as the primary programming language and R for processing non-audio data. Special permission was granted to train neural networks for each speech synthesis solution on IU's Big Red II supercomputer.
Read the paper here, and the Request for Proposal (RFP) here. Code is avaliable in my class repository on GitHub. Final project for LING-L 445 "Computation and Linguistic Analysis," an undergraduate/graduate level course.