EveryVoiceTTS / EveryVoiceTTS/EveryVoice
Create a regression Test
Open
enhancement
- Dominant language
- Python
- Stars
- 45
- Forks
- 4
- Avg merge
- 1d 8h
- Merged PRs (30d)
- 14
Description
#### Regression test overview
- [x] Use LJ, maybe 10000 sentence sample
- [x] run new-project with a canned --resume-from file
- [x] preprocess
- [x] train fs2
- [x] train a vocoder on same
- [x] synthesize a bunch of stuff
- [x] run the demo -- maybe use playwright to enter a couple sentences to synthesize
#### Synthesis variants to include
- [ ] synthesize with all output types
- [ ] synthesize from text (characters, phonemes, phonological features)
- [x] synthesize from spectrograms
- [ ] try synthesizing spectrograms and then running vocoder fine-tuning on the generated spectrograms
Contributor guide
Assessment
This issue has not been assessed yet.