Commit Graph

48 Commits

Author SHA1 Message Date
WeberJulian 61b67ef16f Fix read_json_with_comments 2023-12-11 23:58:52 +01:00
Eren Gölge 623ea41634
Fix model tests (#2943) 2023-09-14 15:21:48 +02:00
Eren G??lge ee7cee0e35 Fixup 2023-09-13 18:21:44 +02:00
Eren Gölge d309f50e53
Implement FreeVC (#2451)
* Update .gitignore

* Draft FreeVC implementation

* Tests and relevant updates

* Update API tests

* Add missings

* Update requirements

* :(

* Lazy handle for vc

* Update docs for voice conversion

* Make style
2023-03-25 18:33:23 +01:00
Julian Weber a07397733b
Multilingual tokenizer (#2229)
* Implement multilingual tokenizer

* Add multi_phonemizer receipe

* Fix lint

* Add TestMultiPhonemizer

* Fix lint

* make style
2023-01-02 10:03:19 +01:00
Eren Gölge 9e5a469c64
d-vector handling (#1945)
* Update BaseDatasetConfig

- Add dataset_name
- Chane name to formatter_name

* Update compute_embedding

- Allow entering dataset by args
- Use released model by default
- Use the new key format

* Update loading

* Update recipes

* Update other dep code

* Update tests

* Fixup

* Load multiple embedding files

* Fix argument names in dep code

* Update docs

* Fix argument name

* Fix linter
2022-09-13 14:10:33 +02:00
Edresson Casanova 371772c355
Replace pyworld by pyin (#1946)
* Replace pyworld by pyin

* Fix unit tests
2022-09-09 10:43:14 +02:00
Stanislav Kachnov 2c9f00a808
Fix tune wavegrad (#1844)
* fix imports in tune_wavegrad

* load_config returns Coqpit object instead None

* set action (store true) for flag "--use_cuda"; start to tune if module is running as the main program

* fix var order in the result of batch collating

* make style

* make style with black and isort
2022-08-22 09:55:32 +02:00
Edresson Casanova f81892483d
REBASED: Transform Speaker Encoder in a Generic Encoder and Implement Emotion Encoder training support (#1349)
* Rename Speaker encoder module to encoder

* Add a generic emotion dataset formatter

* Transform the Speaker Encoder dataset to a generic dataset and create emotion encoder config

* Add class map in emotion config

* Add Base encoder config

* Add evaluation encoder script

* Fix the bug in plot_embeddings

* Enable Weight decay for encoder training

* Add argumnet to disable storage

* Add Perfect Sampler and remove storage

* Add evaluation during encoder training

* Fix lint checks

* Remove useless config parameter

* Active evaluation in speaker encoder test and use multispeaker dataset for this test

* Unit tests fixs

* Remove useless tests for speedup the aux_tests

* Use get_optimizer in Encoder

* Add BaseEncoder Class

* Fix the unitests

* Add Perfect Batch Sampler unit test

* Add compute encoder accuracy in a function
2022-03-11 14:43:40 +01:00
Edresson Casanova 917f417ac4
Add alphas to control language and speaker balancer (#1216)
* Add alphas to control language and speaker balancer

* Add docs for speaker and language samplers

* Change the Samplers weights to float for save memory

* Change the test_samplers to unittest format

* Add get_sampler method in BaseTTS

* Fix rebase issues

* Add language and speaker samplers support for DDP training

* Rename distributed sampler wrapper

* Remove the DistributedSamplerWrapper and use the one from Trainer

* Bugfix after rebase

* Move the samplers config to tts config
2022-03-10 14:56:09 +01:00
Eren Gölge aa81454721 Update BaseTrainingConfig 2022-02-25 11:26:59 +01:00
Eren Gölge 1f0c8179da Make style 2022-02-25 11:26:59 +01:00
Eren Gölge 04202da1ac Make style 2022-02-25 10:48:03 +01:00
Eren Gölge d724984be1 Fix language assignment 2022-01-02 11:11:24 +00:00
Eren Gölge 61874bc0a0 Fix your_tts inference from the listed models 2021-12-31 13:45:05 +00:00
Eren Gölge 633dcc9c56 Implement RMS volume normalization 2021-12-22 15:51:14 +00:00
Eren Gölge c9c1fa0548 Fix multi-speaker init in Synthesizer 2021-12-21 09:44:07 +00:00
Eren Gölge 4c50f6f4df Add functions to get and check and argument in config and config.model_args 2021-12-20 11:54:10 +00:00
Edresson 90eac13bb2 Rename ununsed_speakers to ignored_speakers 2021-12-20 11:54:10 +00:00
Edresson f996afedb0 Implement multilingual dataloader support 2021-12-20 11:54:09 +00:00
Eren Gölge e62d3c5cf7 Use absolute imports for tts configs and models 2021-10-21 16:29:06 +00:00
Eren Gölge 8584f2b82d Update docstring format 2021-09-06 15:16:58 +00:00
Eren Gölge 94e8e0d416 Fix configs 2021-09-06 15:16:58 +00:00
Eren Gölge 18b2e41e5a Use `coqui_tts` as the default run name 2021-08-30 12:56:47 +00:00
Ayush Chaurasia 28870f8df4 update docstring 2021-08-09 18:35:35 +00:00
Ayush Chaurasia 79b74a989d Update: add_text 2021-08-09 18:34:38 +00:00
Ayush Chaurasia f63cf46c55 Unified logger API 2021-08-09 18:34:00 +00:00
Ayush Chaurasia f606741dc4 Add artifacts logging , wandb args 2021-08-09 18:31:16 +00:00
Eren Gölge 060e746e21 Add `do_amp_to_db` option 2021-08-09 18:02:36 +00:00
Eren Gölge 960a35a121 Add `scheduler_after_epoch` to `BaseTrainingConfig` 2021-08-09 18:02:36 +00:00
Agrin Hilmkil ced4cfdbbf Allow saving / loading checkpoints from cloud paths (#683)
* Allow saving / loading checkpoints from cloud paths

Allows saving and loading checkpoints directly from cloud paths like
Amazon S3 (s3://) and Google Cloud Storage (gs://) by using fsspec.

Note: The user will have to install the relevant dependency for each
protocol. Otherwise fsspec will fail and specify which dependency is
missing.

* Append suffix _fsspec to save/load function names

* Add a lower bound to the fsspec dependency

Skips the 0 major version.

* Add missing changes from refactor

* Use fsspec for remaining artifacts

* Add test case with path requiring fsspec

* Avoid writing logs to file unless output_path is local

* Document the possibility of using paths supported by fsspec

* Fix style and lint

* Add missing lint fixes

* Add type annotations to new functions

* Use Coqpit method for converting config to dict

* Fix type annotation in semi-new function

* Add return type for load_fsspec

* Fix bug where fs not always created

* Restore the experiment removal functionality
2021-08-09 18:02:36 +00:00
Eren Gölge 6b265ae8e3 Docstring update 2021-06-28 17:03:47 +02:00
Eren Gölge ab563ce7cd Start training by config.json using `register_config` 2021-06-28 17:03:47 +02:00
Eren Gölge 15fa31b595 fixup configs 2021-06-28 17:03:47 +02:00
Edresson c90037c2e9 solve merge problems 2021-05-26 16:01:30 -03:00
Eren Gölge 218af1d9a2 change `list` to `List` in config 2021-05-18 17:30:27 +02:00
Eren Gölge 4df31f7fbd unused_speakers argument for ignoring speaker ids in multi-speaker
training
2021-05-18 14:50:03 +02:00
Eren Gölge 12722501bb styling 2021-05-15 23:48:31 +02:00
Eren Gölge 8b1014d188 add docstrings with default value fixes 2021-05-15 23:45:10 +02:00
Eren Gölge 0213e1cbf4 update configs for tts models to match the field typed with the expected
values
2021-05-12 00:57:38 +02:00
Eren Gölge 843d1b3d98 linter fixes 2021-05-11 11:30:00 +02:00
Eren Gölge 19fb1d743d style update 2021-05-11 11:30:00 +02:00
Eren Gölge 10de40bba1 make num_workers mandatory config field 2021-05-11 11:29:18 +02:00
Eren Gölge df1ddd3539 allow read_json_with_comments for backward compat 2021-05-11 11:29:18 +02:00
Eren Gölge 3fde2001b1 train_encoder refactoring for coqpit 2021-05-11 11:29:18 +02:00
Eren Gölge 9ee70af9bb code styling 2021-05-11 11:29:18 +02:00
Eren Gölge 10db2baa06 global shared Coqpit configs 2021-05-11 11:29:18 +02:00
Eren Gölge 757e90b1cc load_config function to initialize the right Coqpit for the given model 2021-05-11 11:29:18 +02:00