Commit Graph

163 Commits

Author SHA1 Message Date
Brandon Thomas c43c5ae99c Back to "thin" venv, also print cuda stats 2023-04-26 23:50:00 -07:00
Brandon Thomas fe22a102be "thick" python venv 2023-04-26 08:49:13 -04:00
Brandon Thomas f367e91033 Merge pull request #11 from storytold/bt-fix-svc-docker
We can't run `svc` from another k8s container
2023-04-26 05:11:42 -07:00
Brandon Thomas 80a3d86205 cleanup 2023-04-26 08:10:46 -04:00
Brandon Thomas 280c694783 also install venv 2023-04-26 08:09:49 -04:00
Brandon Thomas d4a963413a report env vars 2023-04-26 07:41:21 -04:00
Brandon Thomas 3735576654 can call using python -m "module as script" invocation 2023-04-26 06:44:45 -04:00
Brandon Thomas 127672e571 Attempting to fix svc to work with k8s 2023-04-26 05:53:26 -04:00
Brandon Thomas 22c8ca104e test 2023-04-26 01:07:36 -07:00
Brandon Thomas 2ed94a5658 venv ignore 2023-04-26 00:50:38 -07:00
Brandon Thomas a1aabd5f3e build so-vits-svc 2023-04-25 21:35:51 -07:00
Brandon Thomas 5207696bcf adding ffmpeg lets svc transcode ogg 2023-04-25 04:11:38 -04:00
Brandon Thomas ec3c16f8b5 fixed legacy TT2 docker - install python3.6 deps directly instead of venv 2023-04-18 18:47:35 -04:00
Brandon Thomas b784b2a203 cleanup legacy TT2 docker image 2023-04-18 15:13:49 -07:00
Brandon Thomas 8cd777b463 stupid fix for f-string interpolation fail 2023-04-18 12:46:23 -07:00
Brandon Thomas cad3c6dee5 We can't run onnx for download integrity check, so run inference 2023-04-13 19:38:48 -04:00
Brandon Thomas e2b8bd586d Michael's copy of so-vits-svc adds cross-fading 2023-04-09 22:19:43 -07:00
Brandon Thomas e7efe757e5 Merge pull request #7 from storytold/feature/proper-audio-chunking
Cross Faded Chunking for Lower Memory usage on Inference
2023-04-10 01:19:16 -04:00
Brandon Thomas f8957a63e9 Add back original so-vits-svc 2023-04-09 22:18:14 -07:00
Brandon Thomas 2148c77395 Move Paul's copy of so-vits-svc so I can experiment with three copies 2023-04-09 22:16:33 -07:00
Brandon Thomas 8923c4952f Merge pull request #8 from storytold/saj-updated-sovits-svc
update to so-vits-svc-fork repo, reduce memory usage, add dockerfile, use dio instead of praat
2023-04-10 01:15:28 -04:00
Brandon Thomas 16ed2e13d5 Hack for forcing nltk to use our cache 2023-04-09 06:49:02 -04:00
Brandon Thomas 7462bf1838 include ffmpeg, etc dependencies 2023-04-07 00:26:37 -04:00
Brandon Thomas 31bfa49bec Generate metadata output file (but not spectrogram) 2023-04-06 15:58:37 -07:00
Brandon Thomas 0c2f709a4f change docker base image to older cuda 2023-04-04 15:43:25 -04:00
Brandon Thomas 35772ff318 docker shouldn't include build outputs! 2023-04-04 05:49:21 -04:00
Brandon Thomas e2597bdde3 Dockerfile for VITS 2023-04-04 02:20:40 -07:00
Brandon Thomas 54e48388cc Merge pull request #10 from storytold/wav2lip-update
Wav2lip update
2023-04-04 03:39:39 -04:00
Paul Sajna 3e7affb2a8 update deps for wav2lip to support python 3.10.10 2023-04-03 13:59:37 -07:00
Brandon Thomas 6c64d725f5 put all tts models under tts/ (we can change later) 2023-04-03 00:02:15 -07:00
Brandon Thomas 36fde12212 Merge pull request #9 from storytold/zdisket-neotts-rebase
Rebase of ZDisket's TT2 and VITS (neo-tts)
2023-04-03 02:57:19 -04:00
ZDisket b3aaee9549 Add DE-VITS ARPA 2023-04-03 02:53:30 -04:00
ZDisket 3f657d8ede add train, check notebook 2023-04-03 02:53:29 -04:00
ZDisket 4aa96d4697 Add inference notebooks 2023-04-03 02:53:29 -04:00
ZDisket e928e6b44a Add Tacotron2-CAC
- Tacotron2 based on https://github.com/keonlee9420/Comprehensive-Tacotron2
- Removed DeepSpeaker and Tensorflow dependency
- Added ARPA function
- Replaced guided attention loss with Convolutional Attention Consistency
2023-04-03 02:53:29 -04:00
ZDisket 5d78edd6b5 add this random stuff 2023-04-03 02:53:29 -04:00
Paul Sajna a9d643488b fix bugs introduced during slicing 2023-03-25 22:41:53 -07:00
Paul Sajna 572400ea85 update to so-vits-svc-fork repo 81551ce9c6fb7924d184c3c5a4cf9035168b28d2, add Dockerfile, reduce memory usage, remove unnecessary deps, use dio instead of parselmouth or crepe 2023-03-21 22:37:01 -07:00
ArEnSc af65e49da6 Got subchunking for audio lower than 15 seconds to be chunked a bit further then cross faded. 2023-03-21 21:10:55 -04:00
Brandon Thomas 8bac589838 fix vocoder paths 2023-03-21 05:53:07 -04:00
Brandon Thomas fc27fb26c7 Fix other script issues 2023-03-21 05:45:17 -04:00
Brandon Thomas 548ce49535 Fixes to inference script to match Rust 2023-03-21 05:35:30 -04:00
Brandon Thomas 85d9b7afec New entry point for subprocess/CLI-based inference 2023-03-21 03:58:36 -04:00
Brandon Thomas 89a7055c35 Update requirements, doc 2023-03-13 07:12:44 -04:00
Brandon Thomas b7ec3a2269 add so-vits-svc @ f22819cb8718ead5e50dea4fdbd05195c72fb1a8 2023-03-13 04:54:54 -04:00
Brandon Thomas 0082747d3c Try to suppress github scrollbar 2023-03-13 04:30:40 -04:00
Brandon Thomas cabd6ca6f7 ascii art 2023-03-13 04:29:45 -04:00
Brandon Thomas c4ec6ffc08 Update README. Root-level readme is going to go out of date
Since the root readme can't give us much information on each model,
we'll just add some ASCII art.
2023-03-13 04:26:27 -04:00
Brandon Thomas 9fe76017d4 update dockerfile to work locally 2023-03-13 03:59:24 -04:00
Brandon Thomas 9e011e9863 Move dockerfile and update build script 2023-03-13 01:48:03 -04:00