Commit Graph
142 Commits
Author SHA1 Message Date
Hervé BREDIN 1c37c39dd7 Créé avec Colaboratory 2022-07-20 22:49:12 +02:00
Hervé BREDIN 2decda1d04 Créé avec Colaboratory 2022-07-18 10:49:08 +02:00
J-PetiotandHervé BREDIN d094a3e43d feat: add pyannote.diff and pyannote.review Prodigy recipes
* fix: fixes #816 

Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2022-04-08 16:11:10 +02:00
FrenchKrabandHervé BREDIN 01904eb4ca * feat: add support for custom validation metrics (#913)
* BREAKING: use task and protocol full names in logs and checkpoint paths
* BREAKING: use diarization error rate for validation of segmentation

Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2022-03-15 16:49:04 +01:00
FrenchKrab c0bf7dd50c improve: update the AMI-diarization-setup submodule (#903) 2022-02-21 09:31:43 +01:00
Hervé BREDINandJ-Petiot 538b00476f feat: add pyannote.audio Prodigy recipe (#896)
Co-authored-by: J-Petiot <jpetiot@sylvebarbe.localdomain>
2022-02-17 16:45:16 +01:00
Hervé BREDIN a7d9a73679 Update training_with_cli.md 2022-02-17 09:02:02 +01:00
Hervé BREDIN cfa1632185 Update training_with_cli.md 2022-02-15 22:07:33 +01:00
Hervé BREDIN 065c6ff22e feat: add pyannote-audio-train CLI (#881)
feat: add pyannote.audio.utils.preprocessors utility module
BREAKING: update PyanNet.sincnet.stride default value to 10
BREAKING: rename Segmentation Hydra config to SpeakerSegmentation
2022-02-11 16:13:52 +01:00
FrenchKrab 4eda05c33f fix: fix outdated example in the training tutorial 2022-02-11 12:46:52 +01:00
Hervé BREDIN 0471053b46 Créé avec Colaboratory 2022-02-08 21:28:43 +01:00
Hervé BREDIN b5794f46d1 feat: add initial documentation (#841) 2022-01-08 22:20:41 +01:00
Hervé BREDIN 6466fa6371 Update README.md 2021-12-16 15:16:06 +01:00
Hervé Bredin 70dbf764a0 doc: update VAD tutorial 2021-12-16 15:06:42 +01:00
Hervé BREDIN d02142abb4 doc: remove useless cell 2021-12-15 13:38:07 +01:00
Hervé BREDIN 2e83450678 feat: add fine-tuning tutorial (#631) 2021-12-15 13:36:56 +01:00
Hervé Bredin 062bd7bd9a git: add AMI-diarization-setup submodule 2021-04-27 13:44:58 +02:00
Hervé BREDIN 977c3de8f7 feat: add "Defining a custom model" tutorial (#607) 2021-02-05 15:06:49 +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 035368428a feat: reproduce original segmentation results 2020-12-07 14:42:50 +01:00
Hervé BREDIN a9778a4d62 feat: add pipelines (and tutorials) 2020-11-16 15:07:44 +01:00
Harry f1906bcbbc Merging in v2 repo 2020-11-08 22:44:29 +01:00
Dan ErezandHervé BREDIN d2279b5ae9 doc: explain how to use own set of scores in pipeline training
Fixes #439 #430

Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2020-09-23 11:32:06 +02:00
Hervé BREDIN c683897866 fix: fix typo similar to #420 2020-07-23 09:20:01 +02:00
Hervé BREDIN 3f274eb4ba fix: fixes #420 2020-07-23 09:17:24 +02:00
Michael ScofieldandHervé BREDIN cd57d9cc6a doc: add more details in "pretrained" tutorial
Co-authored-by: Hervé BREDIN <hbredin@users.noreply.github.com>
2020-07-19 11:27:56 +02:00
Hervé BREDIN aba831b97b fix: fix typo in speaker embedding tutorial 2020-07-02 09:57:45 +02:00
kan-cloud ca03dc1a38 fix: fix typo in SAD tutorial 2020-06-19 19:39:10 +02:00
kan-cloud 52f1fe7d13 fix: fix typo in SCD tutorial 2020-06-19 19:38:41 +02:00
kan-cloud 34b3632b80 fix: fix typo in SCD tutorial 2020-06-19 19:37:54 +02:00
Айдос Сарсембаев 5bcc577828 doc: fix typo in tutorial 2020-06-18 09:29:50 +02:00
Hervé BREDIN c3791bc02c setup: switch to pyannote.database 4.0 2020-06-15 18:12:55 +02:00
Hervé BREDIN e5fe4f8a0d feat: add prodigy recipes (#394)
setup: switch to pyannote.core 4.1
2020-06-09 14:40:45 +02:00
Hervé BREDIN d4c58ab566 fix: fix #392 2020-06-04 09:20:17 +02:00
Hervé Bredin f886996cc2 fix: fixes #381 2020-05-20 09:17:58 +02:00
Hervé BREDIN e131e309b2 fix: fixes #362 2020-04-27 14:59:54 +02:00
Wes Bouaziz 562b74e4e2 fix: fix typos in tutorials 2020-04-17 14:58:52 +02:00
PaulLerner 8c3ec73706 fix: fix typo in SCD tutorial 2020-04-14 12:59:04 +02:00
Hervé BREDIN 8593cc5270 feat: switch to pyannote.database 3.0 2020-03-31 23:48:59 +02:00
Hervé Bredin 3d951ef023 feat: add tutorial for applying pretrained pipeline 2020-03-06 14:16:37 +01:00
Hervé Bredin bd9623c058 doc: update pipeline training tutorial (fixes #306) 2020-03-06 14:15:08 +01:00
Hervé BREDIN 88e24ee24f BREAKING: use f-score for validation everywhere
- BREAKING: remove `--purity` and `--precision` options at they were actually making the pyannote-audio CLI too complicated with little to no benefit. "validation" is just meant to help decide when a model starts to overfit, and should ideally only be used for this purpose.  `pyannote-pipeline` should be used to actually build and evaluate a system.
- doc: update tutorials accordingly
2020-02-27 13:02:55 +01:00
Hervé Bredin 3ea0c09487 doc: update SAD pipeline tutorial 2020-02-26 09:32:56 +01:00
Hervé Bredin 4ba6d4f57f fix: fix typos in SAD pipeline tutorial 2020-02-25 14:33:04 +01:00
Hervé Bredin 3a8288b18d fix: update SAD pipeline tutorial to match latest API
... and also show how to use "apply" pretrained models
2020-02-25 14:27:36 +01:00
Hervé Bredin f6ca360f94 feat: add --cpu option and use GPU by default when available 2020-02-25 13:11:44 +01:00
Hervé Bredin 9183200745 improve: add support for fine-tuning without a configuration file
* feat: add fine-tuning tutorial
2020-02-24 17:10:31 +01:00
Hervé BREDIN bf728bb622 fix: fixes #301 2020-02-17 13:17:36 +01:00
Hervé BREDIN 8a133f197f fix: fixes #302 2020-02-17 13:14:20 +01:00
Hervé BREDIN 0d82f87833 doc: ask user to use models in develop until properly released
Fixes #278
2020-02-17 12:53:55 +01:00