Commit Graph

17 Commits

Author SHA1 Message Date
Marcus Llewellyn
0e08760896 Fixed silly lack of EOF blank line, indentation 2022-06-06 15:13:29 -05:00
Marcus Llewellyn
5a74461c1e read.py combines all candidates
If candidates where greater than 1 on in read.py, only the fist candidate clips would be combined. This adds a bit of code to make a combined file for every candidate.
2022-06-04 17:47:29 -05:00
James Betker
412315ab7d Update read.py to support multiple candidates 2022-05-22 05:26:01 -06:00
James Betker
d96d55a8b4 Fix faulty merge 2022-05-19 10:37:57 -06:00
James Betker
a1c131bde9 Merge remote-tracking branch 'origin/main'
# Conflicts:
#	tortoise/read.py
2022-05-19 10:34:54 -06:00
Johan Nordberg
00730d2786 Allow setting models path from environment variable 2022-05-19 21:02:09 +09:00
James Betker
8fdf516e62 Remove CVVP
After training a similar model for a different purpose, I realized that
this model is faulty: the contrastive loss it uses only pays attention
to high-frequency details which do not contribute meaningfully to
output quality. I validated this by comparing a no-CVVP output with
a baseline using tts-scores and found no differences.
2022-05-17 12:21:25 -06:00
James Betker
a1ae84c49d Add a way to get deterministic behavior from tortoise and add debug states for reporting 2022-05-17 12:11:18 -06:00
Johan Nordberg
a8fa71b82d Improve sentence splitting 2022-05-13 11:02:17 +00:00
James Betker
33d4226a7d read.py: allow user-specified splits 2022-05-12 11:24:55 -06:00
James Betker
12acac6f77 Fix default output path 2022-05-02 21:37:39 -06:00
James Betker
00e84bbd86 fix paths 2022-05-02 20:56:28 -06:00
James Betker
5663e98904 misc fixes 2022-05-02 18:00:57 -06:00
James Betker
ee24d3ee4b Support totally random voices (and make fixes to previous changes) 2022-05-02 15:40:03 -06:00
James Betker
66805da4bd add support for specifying the model_dir 2022-05-01 17:29:25 -06:00
James Betker
01b783fc02 Add support for extracting and feeding conditioning latents directly into the model
- Adds a new script and API endpoints for doing this
- Reworks autoregressive and diffusion models so that the conditioning is computed separately (which will actually provide a mild performance boost)
- Updates README

This is untested. Need to do the following manual tests (and someday write unit tests for this behemoth before
it becomes a problem..)
1) Does get_conditioning_latents.py work?
2) Can I feed those latents back into the model by creating a new voice?
3) Can I still mix and match voices (both with conditioning latents and normal voices) with read.py?
2022-05-01 17:25:18 -06:00
James Betker
23a3d5d00b Move everything into the tortoise/ subdirectory
For eventual packaging.
2022-05-01 16:24:24 -06:00