- Fix train not working (
f90cc40)
- Better notebook (
a80a296)
- Better notebook (
b3e9fe3)
- Fix config templates not included (
319ba6e)
- Update civitai model url (
0f015e3)
- Update dependency transformers to v4.43.3 (
bd9262f)
- Update dependency torch to v2.4.0 (
20549f6)
- Update dependency transformers to v4.42.4 (
f949a07)
- Update dependency lightning to v2.3.3 (
31edf05)
- Update dependency fastapi to v0.111.1 (
59ed5f3)
- Update dependency transformers to v4.42.3 (
b9c031c)
- Update dependency tensorboard to v2.17.0 (
e5f3c13)
- Update dependency lightning to v2.3.2 (
a7e299f)
- Update dependency matplotlib to v3.9.1 (
df6adf4)
- Update dependency lightning to v2.3.1 (
89da16b)
- Update dependency scipy to v1.14.0 (
45a1167)
- Update dependency torchcrepe to v0.0.23 (
2d76d82)
- Update dependency torch to v2.3.1 (
cc51418)
- Update dependency sounddevice to v0.4.7 (
4df53c2)
- Update dependency requests to v2.32.3 (
e60876a)
- Update dependency onnx to v1.16.1 (
0d7ed17)
- Replace pysimplegui with pysimplegui-4-foss (
34e2e77)
- Update dependency transformers to v4.41.1 (
42c69fd)
- Update dependency lightning to v2.2.5 (
6a457dc)
- Update dependency requests to v2.32.2 (
28e1be1)
- Update dependency transformers to v4.41.0 (
9d20b50)
- Update dependency matplotlib to v3.9.0 (
ed95519)
- Update dependency tqdm-joblib to ^0.0.4 (
06ea73c)
- Update dependency fastapi to v0.111.0 (
ee70d52)
- Fix format selection for the input audio in non-windows (
8168cb4) - Fix format selection for the input audio in non-windows (
8168cb4)
- Support python 3.12, end support for python 3.8, explicitly specify click as a dependency, update deps (
a7ceffa)
- Replace 3.10 with 3.11 (
a7ceffa)
- Add leading zeros for 4-digit width of the output file name's numeric part #1154 (
41b147f)
- Add annagorshunova as a contributor for bug, and code (
50f6d79) - Update readme.md [skip ci] (
50f6d79) - Update .all-contributorsrc [skip ci] (
50f6d79)
- Set speaker-diarization version to 3.1 for pyannote.audio 3.1.1 compatibility (
9bd3089)
- Update dependency fastapi to v0.110.1 (
eab647c)
- Add priyanshu-hawk as a contributor for code (
d6888db) - Update readme.md [skip ci] (
d6888db) - Update .all-contributorsrc [skip ci] (
d6888db) - Add marcellocirelli as a contributor for bug (
8795709) - Update readme.md [skip ci] (
8795709) - Update .all-contributorsrc [skip ci] (
8795709)
- Add description of repository maintenance status (
3f537b0) - Add samuelswartzberg as a contributor for security (
cddb722) - Update readme.md [skip ci] (
cddb722) - Update .all-contributorsrc [skip ci] (
cddb722) - Update pytorch urls (
c0c5376) - Add 4ll0w3v1l as a contributor for code (
df699c7) - Update readme.md [skip ci] (
df699c7) - Update .all-contributorsrc [skip ci] (
df699c7)
- Disallow pysimplegui>=5, update deps, update pytorch urls in readme.md (
c0c5376) - Disallow pysimplegui>=5 (
c0c5376)
- Fix broken scipy imports in _pqmf.py (
b7639ca)
- Update dependency transformers to v4.39.1 (
a274333)
- Update dependency pebble to v5.0.7 (
e14b62f)
- Update dependency lightning to v2.2.1 (
a84d26b)
- Update dependency onnxsim to v0.4.36 (
12761e8)
- Update dependency transformers to v4.38.2 (
cfc4edb)
- Update dependency rich to v13.7.1 (
21f33d6)
- Update dependency fastapi to v0.110.0 (
29fc759)
- Update dependency torch to v2.2.1 (
bbc73c1)
- Update dependency transformers to v4.38.1 (
c90cfee)
- Update dependency transformers to v4.38.0 (
4dec304)
- Update dependency matplotlib to v3.8.3 (
e8eab7f)
- Update dependency tqdm to v4.66.2 (
4516483)
- Update dependency lightning to v2.2.0 (
f7b2a42)
- Update dependency fastapi to v0.109.2 (
c570f8e)
- Update dependency fastapi to v0.109.1 (
6ee83d5)
- Update dependency lightning to v2.1.4 (
33334fd)
- Update dependency torch to v2.2.0 (
8750059)
- Update dependency transformers to v4.37.2 (
69c59b8)
- Add onako2 as a contributor for doc (
b204663) - Update readme.md [skip ci] (
b204663) - Update .all-contributorsrc [skip ci] (
b204663)
- Update dependency transformers to v4.37.1 (
d8be0d0)
- Update dependency transformers to v4.37.0 (
7b405c6)
- Update dependency fastapi to v0.109.0 (
565be56)
- Update dependency transformers to v4.36.2 (
7e18425)
- Update dependency fastapi to v0.108.0 (
091805c)
- Update dependency torch to v2.1.2 (
77586fd)
- Update dependency pebble to v5.0.6 (
546db40)
- Update dependency lightning to v2.1.3 (
47b15e6)
- Update dependency pebble to v5.0.4 (
a8dc5d7)
- Update dependency matplotlib to v3.8.2 (
68eb536)
- Update dependency torch to v2.1.1 (
1911035)
- Update dependency lightning to v2.1.2 (
58c8d5a)
- Update dependency rich to v13.7.0 (
1be5442)
- Update dependency transformers to v4.35.2 (
77ee0c0)
- Update dependency transformers to v4.35.1 (
fa503ce)
- Add shinyjustyz as a contributor for bug, and code (
acd6a8e) - Update readme.md [skip ci] (
acd6a8e) - Update .all-contributorsrc [skip ci] (
acd6a8e)
- Make pyanote.audio use gpu (
c9d49ca)
- Update dependency lightning to v2.1.1 (
ce8efce)
- Update dependency transformers to v4.35.0 (
bb05569)
- Update dependency fastapi to v0.104.1 (
dbd4490)
- Update dependency onnx to v1.15.0 (
5736bf7)
- Update python to >=3.8,<3.13 (
031712a)
- Update dependency onnxsim to v0.4.35 (
dd89347)
- Update dependency onnxsim to v0.4.34 (
3d2d4af)
- Update dependency transformers to v4.34.1 (
78c2d4c)
- Update dependency fastapi to v0.104.0 (
6440667)
- Update dependency rich to v13.6.0 (
9ae0737)
- Update dependency fastapi to v0.103.2 (
02cea64)
- Replace "34j" with "voicepaw" (
c1e6c0c)
- Update python to >=3.11,<3.12 (
a5455b9)
- Update dependency rich to v13.5.3 (
e692e8c)
- Update dependency transformers to v4.33.2 (
7a8e54f)
- Update dependency lightning to v2.0.9 (
dcde3d1)
- Update dependency matplotlib to v3.7.3 (
302d5a7)
- Update dependency transformers to v4.33.1 (
f3e3b68)
- Update dependency transformers to v4.33.0 (
146d3ae)
- Update dependency fastapi to v0.103.1 (
f7473aa)
- Update dependency lightning to v2.0.8 (
825fa44)
- Update dependency transformers to v4.32.1 (
da7a72f)
- Add instructions for pipx installation, update torch urls (
0b02c49) - Add shenberg as a contributor for usertesting, ideas, and code (
319ddf3)
- Remove weight norm on inference so metal backend will work without cpu fallback (
39ea0bc)
- Add zerui18 as a contributor for code, and ideas (
4e74fc4) - Add ph0rk0z as a contributor for bug, and code (
8dc25c7)
- Add batched loading to clustering & max length per clip to split (
4179ec9)
- Add star3lord as a contributor for bug, and code (
b3e2cfe)
- Pass str instead of path in sf.load() and sf.write() (
561cbfe)
- Fix typo in core.py (
6a87d32)
- Fix window scaling (
9cd720c)
- Update pretrained model url, raise error if there are no files to preprocess, shuffle files consistently (
c4c719c) - Update urls for pretrained models (
c4c719c)
- Add gui command for module root entrypoint (
3940a4c) - Add gui command to main (
3940a4c) - Add gui command to main (
3940a4c) - Add gui cli command (
3940a4c)
- Replace pyinputplus with normal input (
2b507da)
- Add batch inference, enhance gui, add custom theme (
3ce110b)
- Add meldoner as a contributor for ideas, and code (
880fea8)
- Complete removal of ckpts in colab (
e8964c6)
- Always refresh output path if input path changed (
f79de0c)
- Remove filehandler to avoid permissionerror (
38e0c4e)
- Add maximxlss as a contributor for code (
435ca3c)
- Add max_chunk_seconds option (
101b948)
- Add scorpi as a contributor for code (
542d3a8)
- Fix epoch variable name to log in checkpoint save/load functions (
0530ea3)
- Add pre-classify command to manually classify files (
7a0319c)
- Decouple lf0 predictor from speaker embeddings (
7ab47f4)
- Add highupech as a contributor for bug (
8eedc24) - Fix typo in readme.md (
1773940) - Add acekagami as a contributor for translation (
958b9fd) - Update readme.md [skip ci] (
958b9fd) - Update .all-contributorsrc [skip ci] (
958b9fd) - Update readme_zh_cn.md (
1ccd594)
- Specify encoding to utf-8 in read_text() and write_text() (
e947336)
- Add alexanderkoumis as a contributor for code (
5e032a3) - Update readme.md [skip ci] (
5e032a3) - Update .all-contributorsrc [skip ci] (
5e032a3)
- Configurable output file (#452) (
d2e3596)
- Fix so-vits-svc style contentvec usage (
6d35139)
- Only save checkpoints on main device (
1aaaac6)
- Add sbersier as a contributor for bug (
58b936d) - Add escoolioinglesias as a contributor for video (
69f097f) - Add garrettconway as a contributor for review (
c1e4ada) - Add blueamulet as a contributor for maintenance (
514ed84) - Add guranon as a contributor for bug, ideas, and code (
b9eb3fd) - Add zerohackz as a contributor for bug, and code (
66d5adc) - Add tybantarnusa as a contributor for bug (
e6e57b3) - Add blacksingh as a contributor for bug (
7bc76ba) - Add escoolioinglesias as a contributor for bug, and usertesting (
f00fe6e) - Add outhipped as a contributor for bug (
7497175) - Add yxlllc as a contributor for ideas, and code (
42e35d2) - Add lordmau5 as a contributor for ideas, maintenance, and 2 more (
352451c) - Add tonyco82 as a contributor for bug (
036ce90) - Add 75aosu as a contributor for bug (
5afc28b) - Add hxl9654 as a contributor for bug (
0953f1f) - Add ducttapegames as a contributor for bug (
b0f4d39) - Add likkkez as a contributor for bug (
4a12109) - Add alondan as a contributor for bug (
662ec4b) - Add mmodeusher as a contributor for bug (
6a78df9) - Add meldoner as a contributor for bug (
5586bec)
- Don't save model when tuning for auto batch size (
2311a35)
- Properly stop training after
epochshas been reached (f9bb3d8)
- Support ddp in windows (gloo backend) (
bcb0507)
- Replace
fairseqwithtransformers(a2fe0f3)
- Set persistent_workers = true in dataloader for performance, do not save checkpoints, fix logging issue and multiple warning issues, do not do validation when global_step == 0 (
6cab9af)
- Always use "spawn" context in processpool (
5d7fb77)
- Fix subprocess errors in linux and fix wrong error logging (
fd67db6)
- Fix y_mel length (
2d71992)
- Add option to name ckpts by epochs (
bba24c4)
- Patch stft and add mps to get_optimal_device() (
da928aa)
- Automatically decide batch_size (
8ffa128)
- Show errors raised in inference (
99833c5)
- Suppress pytorch logs for deprecated typedstorage (
e67ac62)
- Fix check for notebook / colab (
7f69814)
- Add option to specify tensorboardlogger version parameter support (
a685123)
- Fix torch.load and save to use file objects and weights_only and remove unidecode (
4aad701)
- Fix gradient logging (
73ef3dc)
- Support sola algorithm (
0fcbf99)
- Do not use rich in notebook (
03c8240)
- Run inference in thread and disable button (
c55caa8)
- Make num_workers configurable (
e8df714)
- Specify num_workers in dataloader (
6042164)
- Use richprogressbar (
17e937a)
- Add optional
accumulate_grad_batchesconfig param (1172b23) - Add accumulate_grad_batches hparam (
1172b23)
- Normalize loss when using gradient accumulation (
1172b23)
- Fix ddp not working (
bec43fc)
- Fix init_logger not showing debug messages in certain conditions as intended (
d3ab7d6)
- Improves inference (
d3228df) - Improves and nb_clean (
d3228df) - Improves inference (
d3228df) - Improves inference (
d3228df) - Unix formatting (
d3228df) - Step lr schedulers at end of epoch (
3af223e)
- Fix fp16_run not being mix precision and fix bf16 errors (
b0dd0ed)
- Fix wrong commands in "before training" (
e056ad9)
- Improve quality of training (
7ed71d6) - Initialize
_temp_epochvariable (7ed71d6) - Fix order of optimizer as per lightning.ai documentation (
7ed71d6) - Remove
with torch.no_grad():call for generator loss (7ed71d6) - Ensure
log_audio_dictuses correcttotal_batch_idx(7ed71d6) - Only save checkpoints for first
batch_idx(7ed71d6)
- Fix checkpoint not properly loaded (
0979147)
- Fix optim_d functions called in wrong order (
13d6346)
- Add bf16 and fp16 support (
4229fd8)
- Update dependency starlette to v0.26.1 (
5eb574b)
- Remove wrong test and trigger release (
9ea77e4) - Remove pydantic constraints (
f446e3b) - Fix fastapi version to 0.88 (
a26f387) - Fix get_optimal_device (
79e4b5a)
- Update dependency fastapi to <0.96 (
29c8cc0)
- Migrate to lightning (
824ecbd)
- Fix train_cluster (
b0c93e4)
- Fix default config type to revert breaking changes (
e05c0b5)
- Fix issues when loading legacy checkpoint and fix pre-hubert n_jobs (
15f1e7f)
- Move methods from dataloader to pre-hubert (
d5a4456)
- Remove possible leak in unused code (
e921c3d)
- Better implementation of repeat_expand_2d (
ef30a9d)
- Add quickvc, fix usage of contentvec, remove onnx support (
1a6c021)
- Update allcontributors link for @mashirosa (
650524b) - Add paperspace referral (
7280012) - Add paperspace link and add more description, add a link for zh-cn docs (
bc4b122)
- Update dependency tensorboard to v2.12.1 (
0ccda1c)
- Update dependency gradio to v3.24.1 (
4fa141b)
- Update dependency gradio to v3.24.0 (
4e441cb)
- Add sbersier as a contributor for ideas, and usertesting (
a655bf4) - Add coldcawfee as a contributor for bug (
87a09e6)
- Update dependency rich to v13.3.3 (
8bdefa9)
- Add nerdyrodent as a contributor for video (
78ab661) - Add heyfixit as a contributor for doc (
32a2a63) - Add desuka-art as a contributor for bug (
fe3c6bf) - Add ruckusmattster as a contributor for bug (
2b971db) - Add pierluigizagaria as a contributor for usertesting (
6fabe8d) - Add satisfy256 as a contributor for bug (
ee72aee) - Add dl909 as a contributor for bug (
a5e6651)
- Add an option to launch tensorboard in
traincommand (ef22cce)
- Fix preprocessing and convert bool options to flags, use
unidecodeto decode non-ascii filenames inpre-resample(98d7ee2)
- Add yt tutorial vid link (
1694f44)
- Specify samplerate to reduce memory usage (
6217eda)
- Initialize logging in logger file and move version log (
441d51f) - Fix dtype in sf.read() to save memory and fix preprocess_resample (
0af1e13) - Fix audio resampled to 22khz (
4203f37)
- Update notebook and readme.md (
38d9744)
- Fix some parameters not passed (
6cfe3d3)
- Add 2 more preprocessing commands (
45eba0f)
- Add blueamulet as a contributor for code (
6a7e8ba)
- Allow float32 audio to be processed properly (
13943b6)
- Change default f0 method from crepe to dio (
baf58d2)
- Add lordmau5 as a contributor for bug, and code (
4df46ee) - Update readme.md [skip ci] (
4df46ee) - Update .all-contributorsrc [skip ci] (
4df46ee)
- Fix old checkpoint deletion by sorting the models properly (#65) (
287dc94)
- Fix devices list and fix tqdm error in gui (
59724cd)
- Add mashirosa as a contributor for doc, and bug (
495b7cb) - Fix cluster inference command and improve cluster training command (
7642594)
- Fix defaut for auto_play (
07920a4) - Fix speaker not automaticlly set to the first one if not found in cluster inference (
a643e4f)
- Add cluster training and inference (
9ffb621)
- Better error handling (
985704b)
- Fix onnx export and fix gui (
3e9a47d)
- Fix cluster not working (
29b209c)
- Fix speakers and devices not updated and fix default presets (
a851150)
- Update dependency torchcrepe to v0.0.18 (
4fda479)
- Shorten docs (
e0c1572)
- Fix clean_checkpoints (
e5169bf)
- Add blueamulet as a contributor for question (
8d073e3) - Add garrettconway as a contributor for doc (
6c6cbc6) - Update readme.md regarding installation, update. wsl audio support (
4f1323b)
- Disable checkbox if cuda is not available and show errors for vc (
3fdd983)
- Fix rtf calculation (
fb25500)
- Fix latest_checkpoint_path (
00b9f4a)
- Update dependency onnxsim to v0.4.19 (
f8a4cf6)
- Update dependency onnxoptimizer to v0.3.10 (
d0137f9)
- Add links for pretrained models and fix gui pic height (
34ac39f) - Add more explanation to notebook (
9b3c483)
- Use librosa.load() instead of soundfile.read() (
b343106) - Fix window too big to show in a fhd environment (
259e6e6)
- Add presets (
e8adcc6)
- Update dependency gradio to v3.23.0 (
a2bdb48)
- Update gui screenshot (
58d06aa)
- Enhance realtimevc (
81551ce)
- Update dependency scipy to v1.10.1 (
e0253bf)
- Add throwawayaccount01 as a contributor for bug (
15e31fa) - Add blueamulet as a contributor for ideas (
a3bcb2b)
- Do dummy inference before running vc (
4066c43)
- Fix default dataset path (
ac47fed)
- Fix compute_f0_crepe returning wrong length (
afb42b0)
- Update dependency librosa to v0.10.0 (
8e92f71)
- Add more f0 calculation methods (
6b3b20d)
- Update dependency gradio to v3.22.1 (
f09fc23)
- Allow nested dataset (
0433151)
- Update notes about vram caps (
0a245f4)
- Add garrettconway as a contributor for bug (
31d9671) - Launch tensorboard (
52229ba) - Add 34j as a contributor for example, infra, and 6 more (
1b90378) - Add garrettconway as a contributor for code (
716813f)
- Use hubert preprocess force_rebuild argument (
87cf807)
- Better performance (
668c8e1)
- Configurable input and output devices (
a822a60)
- Fix option names (
7ff34fe)
- Update readme.md (
b988101)
- Call init_logger() (
e6378f1)
- Enhance realtime algorythm (
d789a12)
- Add gui (
34aec2b)
- Update notebook (
7b74606)
- Fix notebook (
3ed00cc)
- Realtime inference (
4dea1ae)
- Main feat (
faa990c)