MAJOR release for GenVox. Complete revamp of code and dataflow.
- Cleaned up README and created a logo (582a3ff)
- Code structure changed and is now more readable and easy to understand (4dea463, 1dfd058, 0114a95)
- Removed typeguard, but integrated typing hints to a higher extent (5238132)
- Highly generalized model base that allows to easily add new models (96a4a35, ec06296)
- Upgraded trainer, with extra features like
debug_run
(349d6bc), single batch eval structure (75c333a) - Added
Synthesizer
to easily generate speech from text using pretrained models (b08681e) - Annotated, cleaned and optimized
Tacotron2
(f393fe2) - Upgraded resample.py to do parallel processing (514fc7c)
- Updated data analysis and added custom dataset creation from
Youtube
in tools (98b8c0b) - Added support for ULCA dataset: formatter and pipeline (1de8570)