Archived in favour of using the IMS-Toucan toolkit. (https://github.com/DigitalPhonetics/IMS-Toucan)
Adapted from Tomiinek Multilingual Text to Speech
This repository contains an implementation of Tacotron 2 that supports gradient accumulation along with some other modifications from the sourced code to accomplish the goal of generating correct Cherokee language speech using the Uchihara "Modified Community Orthography".
At a later date using Syllabary may be attempted in combination with MCO.
-Michael Conrad