Version 0.9.6

Build with GPU support no longer available

Notes:
Add clip combiner and duration options to dataset building
Fix blank line handling in synthesis
Fix checkpoint exporting
Standardise alignment JSON

Version 0.9.5

Build with GPU support no longer available

Notes:
Remove waveglow
Update to CUDA 11.1 to support 30 series GPU's

Version 0.9.4

Add checkpoint backup system
Add subtitle support

Version 0.9.3

Add train/test split slider to training
New multi-line synthesis feature
Add max_decoder_steps slider to synthesis

Version 0.9.2

Remove synonym suggestion
Move vocoder upload to settings
Fix checkpoint selection in training

Version 0.9.1

Fix English synthesis
Enable checkpoint selection for export, training & synthesis

Version 0.9

Added support for other languages

Version 0.8.4

Update Kindle extraction guide

Version 0.8.3

Add advanced training options for enabling/disabling checkpoint overwriting & multi GPU

Version 0.8.2

Update Pytorch version and revert CUDA to 10.2