Commit Graph
2480 Commits
Author SHA1 Message Date
Hervé BREDIN 88ce811a3d fix(pipeline): fallback to K-means for VBx with constraints on number of speakers 2025-09-07 15:05:54 +02:00
Hervé BREDIN c301c96add setup: loosen lightning dependency (#1904) 2025-08-28 16:32:54 +02:00
antoinelaurent 5363b0483e fix: cast segment boundaries to float (and pad tensors in collate_X) 2025-08-28 15:42:18 +02:00
Hervé BREDIN 000277fee4 doc: update SpeakerDiarization docstring 2025-08-28 15:16:57 +02:00
Hervé BREDIN f75e659dd1 fix(hub): fix handling of revision with models in Pipeline subfolders (#1907, PYA-940) 2025-08-28 10:38:54 +02:00
Clément Pagés 36e5c8fd8c feat(cli): add --per-file option in benchmark command (#1901) 2025-08-25 13:45:54 +02:00
785618c246 feat(io): switch from torchaudio to torchcodec for audio I/O (PYA-901)
Co-authored-by: Juan Coria <juanmc2005@hotmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
2025-08-22 16:08:29 +02:00
Hervé BREDIN 4315c12ddb BREAKING(pipeline): remove unmaintained pipelines (#1902, PYA-858) 2025-08-22 15:09:39 +02:00
Hervé BREDIN f735ba843d BREAKING(task): remove OverlappedSpeechDetection task (#1903) 2025-08-22 15:05:26 +02:00
Hervé BREDIN d496192aac feat: add get_pipeline utility function (#1899) 2025-08-17 11:41:32 +02:00
Hervé BREDIN 24f202dd51 feat: add exclusive_diarization flag (#1898) 2025-08-17 11:38:34 +02:00
365704d336 feat: add support for opentelemetry (#1896)
Co-authored-by: Valentín Mari <valentinmari@hotmail.com>
Co-authored-by: Juan Coria <juanmc2005@hotmail.com>
Co-authored-by: Valentín Mari <vmari@users.noreply.github.com>
2025-08-14 16:20:18 +02:00
Hervé BREDIN f235b19504 feat: add Powerset.powerset_classes property 2025-08-14 08:42:24 +02:00
Hervé BREDIN 3dfb2e5473 doc: update benchmark with latest numbers 2025-08-14 08:39:10 +02:00
Hervé BREDIN 159b769e1b feat: add skip-dependency-check option 2025-08-13 15:01:20 +02:00
Hervé BREDIN bb8cee9093 feat: add skip-dependency-check option 2025-08-13 14:55:08 +02:00
Hervé BREDIN 23ae2d03a7 doc: update README 2025-08-12 18:34:15 +02:00
Hervé BREDINandPetr Pálka 29c883fad6 PYA-849 feat: add support for VBx clustering (#1894)
Co-authored-by: Petr Pálka <82486052+Selesnyan@users.noreply.github.com>
2025-08-12 17:39:10 +02:00
Hervé BREDIN 78924e2fa2 feat(cli): add CLI to strip checkpoints to their bare minimum (#1858) 2025-06-13 17:11:11 +02:00
Juan Coria 9178d88b87 feat(core): allow loading Calibration object from an in-memory dict directly (#1883) 2025-06-13 13:40:05 +02:00
Hervé BREDIN 114c981400 feat(cli): benchmark speaker count estimation (#1882) 2025-06-13 11:39:02 +02:00
Hervé BREDIN ac4dc884b4 feat(cli): add --metric option to "pyannote-audio optimize" (#1880) 2025-06-13 09:49:48 +02:00
Hervé BREDIN 89980bcf62 doc: update pyannoteAI benchmark 2025-06-12 18:31:08 +02:00
Clément Pagés c4c31dde00 fix(cli): fix optimization corner case and file format 2025-06-12 17:11:26 +02:00
Clément Pagés 0942fadcb8 cli(benchmark): add --progress option (#1878) 2025-06-11 16:11:45 +02:00
Clément PagésandHervé BREDIN c97b3d0720 cli(benchmark): add --optimize option for fair comparison (#1876)
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2025-06-02 09:06:30 +02:00
Juan Coria 799fffd768 improve(utils): add option to load model checkpoint from BytesIO (#1873) 2025-05-21 14:23:46 +02:00
Hervé BREDIN 5e56f06866 feat: add option to skip dependency check (#1870) 2025-05-20 21:19:57 +02:00
Hervé BREDIN 0fa7e086b6 feat(cli): log processing hardware and speed 2025-05-02 10:13:50 +02:00
Hervé BREDIN a9d8183a72 improve(utils): improve dependency check when loading pretrained models and/or pipeline 2025-04-15 23:13:48 +02:00
Hervé BREDIN 89fd1ad81a setup: update pytorch dependencies (#1859) 2025-04-15 18:13:42 +02:00
Hervé BREDIN 469f776a92 chore: stringify track names 2025-03-31 20:36:25 +02:00
Hervé BREDIN 97f1a1d8ce improve(train): speed up dataloaders 2025-03-28 11:06:21 +01:00
Hervé BREDIN d381589409 setup: switch to lightning (from pytorch_lightning) 2025-03-27 11:16:49 +01:00
Hervé BREDIN 963b6cf83e feat(pyannoteAI): add wrapper around pyannoteAI SDK 2025-03-14 22:40:52 +01:00
Hervé BREDIN ce495767be feat(cli): add "download" and "optimize" commands (#1843) 2025-02-24 13:00:19 +01:00
Hervé BREDIN b8d20cafad fix(metrics): handle corner case in SegmentationErrorRate (#1847) 2025-02-21 17:10:17 +01:00
Hervé BREDIN 93ad8b90ca BREAKING: switch to uv, namespace package, Python 3.10+, remove old CLI 2025-02-12 13:07:16 +01:00
Hervé BREDIN d9389defa7 fix: fix Calibration.from_pretrained support for directory 2025-02-04 23:13:55 +01:00
Hervé BREDIN 27bf53c3aa fix: fix Calibration.from_pretrained support for pathlib.Path 2025-02-04 23:07:39 +01:00
Hervé BREDIN bb6bc1ea0a fix: set revision to None when loading from local directory 2025-02-04 22:37:18 +01:00
Hervé BREDIN e7b7c6178c fix: fix #1832 2025-02-04 22:07:07 +01:00
Hervé BREDIN 912ef14677 fix: fix HF token usage in SpeechBrain interface 2025-02-04 22:05:49 +01:00
Ilya Gulya 74705e9a5d doc: add progress hook in README.md example (#1826) 2025-02-04 15:28:21 +01:00
Clément Pagés d6136a8d69 fix(separation): fix division by zero in separation pipeline corner cases (#1815) 2025-02-04 14:17:58 +01:00
Hervé BREDIN 43bb258a5c feat: add CLI to apply (and benchmark) pretrained pipelines 2025-01-27 11:07:13 +01:00
Hervé BREDIN fd91866315 doc: improve HF-related docstring 2025-01-27 10:53:04 +01:00
Hervé BREDIN 9cf27e6d62 feat: add segmentation error rate (#1836) 2025-01-23 14:41:34 +01:00
benniekiss 2dba593bc5 BREAKING: fix cache_dir support, rename use_auth_token to token, and other changes 2025-01-20 22:01:06 +01:00
Hervé BREDIN ea1ad3429b setup: update torch_audiomentations, pyannote.database, pyannote.metrics and typer dependencies 2025-01-15 21:45:45 +01:00