70 Commits
Author SHA1 Message Date
Hervé BREDIN c2b12c5aa5 fix: avoid replaying training samples on mid-epoch resume (#2105) 2026-11-19 21:39:03 +01:00
Hervé BREDIN 47fa6ac0dd BREAKING: perform channel selection per-file according to 'channel' key 2026-11-17 08:51:15 +01:00
Kayvan Zahiri 0d6f291328 fix: accumulate discrete DER frame counts in float64 (#2049) 2026-11-09 06:42:58 +01:00
Mohammad Hijjawi 755a98c23c fix(pipeline): fix VBxClustering crashing with requested speakers than embeddings (#2079) 2026-11-06 17:30:54 +01:00
Hervé BREDIN 08d6f3160a feat(cli): add support for multi-objective optimzation (#2056) 2026-11-06 17:14:43 +01:00
AnthonyandHervé BREDIN cc679dc7d6 fix(model): honor ToTaToNet linear hidden size (#2077)
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2026-11-02 15:17:42 +01:00
ea18bc70d8 Align audio CI and releases; require Python 3.11 (#2081)
* feat(cli): add --subfolder option to benchmark command (#2063)
* fix(utils): fix `SpeakerDiarizationMixin.optimal_mapping` when `reference` is a mapping
* fix(task): fix `balance` option crashing training with `KeyError: 'metadata'` (#2067)
* fix(io): fix `Audio.crop` raising when an in-memory waveform is cropped up to its very end (#2068)
* fix(model): make Model.default_activation return identity for unsupported tasks
* fix: compute sample duration from preloaded waveform
* fix(pipeline): omit unset runtime options from model references (#2080)

Co-authored-by: abeeranajam31 <128603901+abeeranajam31@users.noreply.github.com>
Co-authored-by: antoinelaurent <antoinelaurent@me.com>
Co-authored-by: Kayvan Zahiri <123409429+Kayvan-Zahiri@users.noreply.github.com>
Co-authored-by: Mohammad Hijjawi <51020382+MohammadHijjawi97@users.noreply.github.com>
2026-10-22 15:07:01 +02:00
Hervé BREDIN 5dd8d59148 feat: add support for subfolder in Pipeline.from_pretrained (#2029) 2026-06-25 09:54:25 +02:00
Hervé BREDIN 3f2f34deeb chore: remove speechbrain and fix io unit tests (#1938) 2025-10-22 18:21:44 +02:00
Hervé BREDIN f735ba843d BREAKING(task): remove OverlappedSpeechDetection task (#1903) 2025-08-22 15:05:26 +02:00
Hervé BREDIN d381589409 setup: switch to lightning (from pytorch_lightning) 2025-03-27 11:16:49 +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
benniekiss 2dba593bc5 BREAKING: fix cache_dir support, rename use_auth_token to token, and other changes 2025-01-20 22:01:06 +01:00
Adel Moumen bd7b977c21 fix: fix support for speechbrain==1.0 (#1659)
* fixes #1727 
* fixes #1677
2024-06-19 13:58:31 +02:00
Clément Pagés 99ed3d6de3 improve(ux): improve error message on missing specifications 2024-02-25 21:00:25 +01:00
Clément Pagés d83c0f8ea9 fix(cli): update CLI with latest APIs 2024-02-25 20:49:08 +01:00
Hervé BREDIN fea5c2951e feat(metric): add reduce option to diarization_error_rate metric (#1635) 2024-01-31 21:55:36 +01:00
FrenchKrabandHervé BREDIN d115e9ac94 feat(powerset): add Powerset.permutation_mapping to help with permutation in powerset space
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2024-01-30 11:21:08 +01:00
Clément Pagés 9d22113e69 ci: update ci model used for tests 2024-01-25 17:34:23 +01:00
Hervé BREDIN e43138278a feat: add pyannote.audio.sample.SAMPLE_FILE (#1629) 2024-01-25 10:42:32 +01:00
Clément PagésandHervé BREDIN b41b176e3f fix: fix support for non-ASCII characters
Fixes #1608 #1612 

Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2024-01-15 13:11:21 +01:00
Clément PagésandHerve Bredin 9e4ec5f6e0 feat(task): add option to cache training metadata to disk
Co-authored-by: Herve Bredin <hbredin@users.noreply.github.com>
2024-01-12 10:46:46 +01:00
Clément PagésandHervé BREDIN 884913e1ac feat(model): add support for multi-speaker statistics pooling (#1386)
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2023-11-14 10:33:56 +01:00
0e1a726ad2 fix(pipeline): fix AgglomerativeClustering to honor num_clusters when provided
Co-authored-by: Olivier Birot <obirot@babbage.paris.inria.fr>
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2023-11-07 12:02:41 +01:00
Hervé BREDIN e9a92740a0 improve(test): use pyannote.database.registry (#1413) 2023-09-20 17:39:30 +02:00
Hervé BREDIN d19a728a0c fix(test): force test training to run on CPU 2023-05-24 17:17:08 +02:00
Hervé BREDIN 2623cf1980 BREAKING(io): update Audio behavior for multi-channel audio
- BREAKING(io): channels are now 0-indexed (used to be 1-indexed)
- BREAKING(io): multi-channel audio is no longer downmixed to mono by default.
  You should update how `pyannote.audio.core.io.Audio` is instantiated:
  * replace `Audio()` by `Audio(mono="downmix")`;
  * replace `Audio(mono=True)` by `Audio(mono="downmix")`;
  * replace `Audio(mono=False)` by `Audio()`.
2023-04-21 15:14:56 +02:00
Hervé BREDIN 9faa8fc8a8 setup: update most dependencies (#1323)
- setup: switch to torch 2.0+ and lightning 2.0+
- setup: switch to torchaudio 2.0+ and soundfile 0.12+
- setup: switch to pyannote.core 5.0+ and pyannote.database 5.0+
- setup: switch to speechbrain 0.5.14+

- BREAKING(task): rename `Segmentation` task to `SpeakerDiarization`
- BREAKING(task): remove support for variable chunk duration

- BREAKING(pipeline): remove `SpeakerSegmentation` pipeline (in favor of `SpeakerDiarization` pipeline)
- BREAKING(pipeline): remove support `FINCHClustering` and `HiddenMarkovModelClustering`

- BREAKING: drop support for Python 3.7
2023-04-17 15:32:37 +02:00
Hervé BREDIN 7121a73f08 feat: add support for label scope and missing classes (#1296)
- setup: switch to pyannote.database 5.0
- feat(task): add support for label scope in speaker diarization task (from pyannote.database 5.0)
- feat(task): add support for missing classes in multi-label segmentation task (from pyannote.database 5.0)
- improve(task): load metadata as tensors/arrays rather than pyannote.core instances
- BREAKING(task): rename `Segmentation` task to `SpeakerDiarization`
- BREAKING(task): remove support for variable chunk duration
2023-04-13 21:15:57 +02:00
Ewald Enzinger cb7bfd5a4d setup: switch to pytorch-lightning >= 1.8 (#1210) 2023-01-20 12:46:56 +01:00
FrenchKrabandHervé BREDIN 12dc20bde1 feat: add support for "powerset" speaker segmentation training
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2023-01-20 11:22:18 +01:00
hadwareandHervé BREDIN 1dac64a162 feat: add support for generic multi-label segmentation
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2022-07-01 10:35:14 +02:00
Hervé Bredin 39633d9d3a fix: remove augmentation tests 2021-09-08 14:34:31 +02:00
Hervé BREDIN 8ab7ebe044 fix: fix fine-tuning and transfer learning API (#737)
* fix: fixes #736 
* feat: add more unit tests
2021-09-08 14:23:01 +02:00
Hervé BREDIN c69e92df77 chore: remove multi-task support (#675) 2021-06-04 16:35:32 +02:00
Harry Coultas Blum 3c4d0a68a3 feat: add support for file-like objects (#640) 2021-04-12 11:31:13 +02:00
Hervé BREDIN e4d6896500 fix: bring back notebook-based tests (#654) 2021-04-07 13:55:38 +02:00
Hervé BREDIN f8786e2b5d feat: add resegmentation pipeline (#652)
* feat: add Introspection.frames property
* feat: add support for mismatched number of classes in `permutate` utility function
2021-03-23 15:25:20 +01:00
Hervé BREDIN 3091df3549 test: only run notebooks in notebook directory (#647) 2021-03-19 16:32:54 +01:00
Hervé BREDIN b17a6db36f feat: add "skip_aggregation" option to Inference (#612) 2021-02-11 13:03:52 +01:00
Hervé BREDIN 0b9be93ee1 chore: replace load_from_checkpoint by Model.from_pretrained (#588)
* chore: import Model from package root
* chore: import Inference from package root
2021-01-20 16:39:08 +01:00
Hervé BREDIN c2b4247f22 chore: remove Diarization alias and MultiTaskSegmentation import (#586) 2021-01-20 16:13:55 +01:00
Hervé BREDIN a55b31ea8b chore: rename Scale to Resolution (#577) 2021-01-16 15:44:56 +01:00
Hervé BREDIN 8a86468d44 chore: simplify the API even more 2021-01-14 14:33:17 +01:00
Julien Chaumond 61be5122ae feat: add huggingface model hub integration 2021-01-05 11:24:46 +01:00
Hervé BREDIN 78a16a2ca3 feat: add generic "permutate" utility function
* feat: add "vad_loss" option to "Segmentation"
* improve: improve permutation-invariant support in "Inference"
2021-01-04 16:08:59 +01:00
Hervé BREDIN 035368428a feat: reproduce original segmentation results 2020-12-07 14:42:50 +01:00
Harry Coultas Blum 35e39136d6 feat: add codecov report 2020-12-02 18:52:13 +01:00
Hervé BREDIN a9778a4d62 feat: add pipelines (and tutorials) 2020-11-16 15:07:44 +01:00
Harry Coultas BlumandHervé BREDIN 3ffd31a688 feat: switch to torchaudio for audio IO
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2020-11-10 21:00:09 +01:00