Medix-AI/coqui-tts
Medix-AI
/
coqui-tts
mirror of https://github.com/coqui-ai/TTS.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
4,935 Commits 80 Branches 98 Tags 314 MiB
Commit Graph

1055 Commits

Author SHA1 Message Date
Enno Hermann e5c208d254 feat(cleaners): add multilingual phoneme cleaner 
This doesn't convert numbers into English words.
 2024-06-14 15:06:03 +02:00
Enno Hermann 03de4b889e docs: fix readthedocs links 
[ci skip]
 2024-06-13 22:48:34 +02:00
Enno Hermann 07cbcf825c fix(espeak_wrapper): read phonemize() input from file 
Avoids utf8 encoding issues on Windows when passing the text directly.
Fixes https://github.com/coqui-ai/TTS/discussions/3761
 2024-05-29 10:10:05 +02:00
Enno Hermann 49fcbd908b fix(espeak_wrapper): avoid stuck process on windows 
Fixes #24
 2024-05-29 07:39:03 +02:00
Enno Hermann 203f60f1e1 refactor(espeak_wrapper): remove sync argument 
_espeak_exe is always called with sync=True, so remove code for sync==False
 2024-05-28 21:30:55 +02:00
Enno Hermann df088e99df
Merge pull request #19 from idiap/toml 
Move from setup.py to pyproject.toml, simplify requirements
 2024-05-27 08:59:09 +01:00
Enno Hermann 018f1e6453 docs(bark): update docstrings and type hints 2024-05-15 22:56:55 +02:00
Enno Hermann 59a6c9fdf2 fix(bark): add missing argument for load_voice() 
Fixes https://github.com/coqui-ai/TTS/issues/2795
 2024-05-15 22:56:28 +02:00
Enno Hermann 6d563af623 chore: remove obsolete code for torch<2 
Minimum torch version is 2.1 now.
 2024-05-08 18:08:40 +02:00
Enno Hermann 865a48156d fix: make korean g2p deps optional 2024-05-08 18:08:40 +02:00
Enno Hermann 55ed162f2a fix: make chinese g2p deps optional 2024-05-08 18:08:40 +02:00
Enno Hermann ea893c3795 fix: make bangla g2p deps optional 2024-05-08 18:08:40 +02:00
Enno Hermann ec50006855 style: run pre-commit 
Automatic changes from: pre-commit run --all-files
 2024-05-08 12:17:47 +02:00
Enno Hermann fb92e13ebb build: remove unused/obsolete code 2024-05-08 12:13:41 +02:00
Enno Hermann 962f9bbbcf refactor(espeak_wrapper): fix ruff lint suggestions 2024-05-01 13:31:39 +02:00
Enno Hermann 7b2289a454 fix(espeak_wrapper): capture stderr separately 
Fixes https://github.com/coqui-ai/TTS/issues/2728

Previously, error messages from espeak were treated as normal output and also
converted to phonemes. This captures and logs them separately.
 2024-05-01 12:31:49 +02:00
Enno Hermann 52a52b5e21 fix(LanguageManager): allow initialisation from config with language ids file 
Previously, running `LanguageManager.init_from_config(config)` would never use
the `language_ids_file` if that field is present because it was overwritten in
the next line with a new manager that manually parses languages from the
datasets in the config. Now that is only used as a fallback.
 2024-04-19 11:57:27 +02:00
Enno Hermann b3c9685aee fix(tokenizer): add debug logging 2024-04-11 16:58:12 +02:00
Enno Hermann 2ad790d169
Merge pull request #4 from idiap/hindi 
feat(xtts): support Hindi for sentence-splitting and fine-tuning
 2024-04-11 16:49:44 +02:00
Enno Hermann d41686502e feat(xtts): support hindi for sentence-splitting and fine-tuning 
The XTTS model itself already supports Hindi, it was just in these components.
 2024-04-08 15:57:56 +02:00
Enno Hermann b711e19cb6 refactor: remove verbose arguments 
Can be handled by adjusting logging levels instead.
 2024-04-03 15:19:45 +02:00
Enno Hermann b6ab85a050 fix: use logging instead of print statements 
Fixes #1691
 2024-04-03 15:19:45 +02:00
Enno Hermann d772724125 fix: update repository links, package names, metadata 2024-04-03 12:02:44 +02:00
Enno Hermann 7630abb43f refactor(bin.find_unique_chars): use existing function 2024-03-30 22:22:40 +01:00
Enno Hermann adbcba06da refactor(dataset): get audio length with torchaudio 
Removes a (GPL) dependency
 2024-03-14 20:48:29 +01:00
Enno Hermann e5c6da1c98
Merge pull request #20 from eginhard/return-complex 
fix: torch.stft will soon require return_complex=True
 2024-03-13 13:50:21 +01:00
Enno Hermann e95f8950eb fix: torch.stft will soon require return_complex=True 
Refactor that removes the deprecation warning:
torch.view_as_real(torch.stft(*, return_complex=True)) is equal to
torch.stft(*, return_complex=False)

https://pytorch.org/docs/stable/generated/torch.stft.html
 2024-03-13 12:06:27 +01:00
Enno Hermann 89a061f1d1 docs(tts.models.vits): clarify use of discriminator/generator 
[ci skip]
 2024-03-12 18:59:05 +01:00
Enno Hermann 2e8f47a33d
Merge pull request #10 from eginhard/fix-pinyin 
fix chinese pinyin phonemes
 2024-03-09 16:23:28 +01:00
Enno Hermann 309f39a45f fix(xtts_manager): name_to_id() should return dict 
This is how the other embedding managers work
 2024-03-08 14:47:00 +01:00
Enno Hermann efdafd5a7f style: run black 2024-03-07 11:46:51 +01:00
Enno Hermann 017c84d005 style: make style && make lint 2024-03-06 22:45:35 +01:00
Enno Hermann e05243c4c8 refactor: read/write csv files with standard library 2024-03-06 16:18:09 +01:00
Enno Hermann 24298da5fc
Merge pull request #1 from eginhard/lint-overhaul 
Lint overhaul (pylint to ruff)
 2024-03-06 16:10:26 +01:00
wangjie b184e9f0fe fix chinese pinyin phonemes 2024-01-12 09:11:56 +08:00
Edresson Casanova 5dcc16d193
Bug fix in MP3 and FLAC compute length on TTSDataset (#3092) 
* Bug Fix on XTTS load

* Bug fix in MP3 length on TTSDataset

* Update TTS/tts/datasets/dataset.py

Co-authored-by: Aarni Koskela <akx@iki.fi>

* Uses mutagen for all audio formats

* Add dataloader test wit hall supported audio formats

* Use mutagen.File

* Update

* Fix aux unit tests

* Bug fixe on unit tests

---------

Co-authored-by: Aarni Koskela <akx@iki.fi>
 2023-12-27 13:23:43 -03:00
Aarni Koskela d6ea806469 Run `make style` 2023-12-13 14:56:41 +02:00
Aarni Koskela bd172dabbf xtts/stream_generator: remove duplicate import + code 2023-12-13 14:56:41 +02:00
Aarni Koskela 32abb1a7c4 xtts/perceiver_encoder: Delete duplicate exists() 2023-12-13 14:56:41 +02:00
Aarni Koskela aa549e9028 Fix trailing whitespace 2023-12-13 14:56:41 +02:00
Aarni Koskela 33b69c6c09 Add some noqa directives (for now) 2023-12-13 14:56:41 +02:00
Aarni Koskela 00f8f4892a Ruff autofix unnecessary passes 2023-12-13 14:56:41 +02:00
Aarni Koskela bc2cf296a3 Ruff autofix PLW3301 2023-12-13 14:56:41 +02:00
Aarni Koskela 64bb41f4fa Ruff autofix C41 2023-12-13 14:56:41 +02:00
Aarni Koskela 449820ec7d Ruff autofix E71* 2023-12-13 14:56:41 +02:00
Aarni Koskela 90991e89b4 Ruff autofix unused imports and import order 2023-12-13 14:56:41 +02:00
Eren Gölge 8c1a8b522b
Merge pull request #3405 from coqui-ai/studio_speakers 
Add studio speakers to open source XTTS!
 2023-12-12 16:10:09 +01:00
Eren Gölge 934b87bbd1
Merge pull request #3391 from aaron-lii/multi-gpu 
support multiple GPU training for XTTS
 2023-12-12 13:51:26 +01:00
WeberJulian 5cd750ac7e Fix API and CI 2023-12-11 20:21:53 +01:00
WeberJulian e3c9dab7a3 Make CLI work 2023-12-11 18:49:18 +01:00
WeberJulian a5c0d9780f rename manager 2023-12-11 18:48:31 +01:00
WeberJulian 36143fee26 Add basic speaker manager 2023-12-11 15:25:46 +01:00
Frederico S. Oliveira 163f9a3fdf
Merge branch 'coqui-ai:dev' into dev 2023-12-11 10:04:07 -03:00
Aaron-Li b6e929696a support multiple GPU training 2023-12-08 16:55:32 +08:00
Eren Gölge e49c512d99
Merge pull request #3351 from aaron-lii/chinese-puncs 
fix pause problem of Chinese speech
 2023-12-04 15:57:42 +01:00
Edresson Casanova 5f900f156a
Add XTTS Fine tuning gradio demo (#3296) 
* Add XTTS FT demo data processing pipeline

* Add training and inference columns

* Uses tabs instead of columns

* Fix demo freezing issue

* Update demo

* Convert stereo to mono

* Bug fix on XTTS inference

* Update gradio demo

* Update gradio demo

* Update gradio demo

* Update gradio demo

* Add parameters to be able to set then on colab demo

* Add erros messages

* Add intuitive error messages

* Update

* Add max_audio_length parameter

* Add XTTS fine-tuner docs

* Update XTTS finetuner docs

* Delete trainer to freeze memory

* Delete unused variables

* Add gc.collect()

* Update xtts.md

---------

Co-authored-by: Eren Gölge <erogol@hotmail.com>
 2023-12-01 23:52:23 +01:00
Aaron-Li 7b8808186a fix pause problem of Chinese speech 2023-12-01 23:30:03 +08:00
Frederico S. Oliveira bcd500fa7b Fixing bug 
Correction in training the Fastspeech/Fastspeech2/FastPitch/SpeedySpeech model using external speaker embedding.
 2023-11-30 17:27:05 -03:00
Enno Hermann 39321d02be
fix: correctly strip/restore initial punctuation (#3336) 
* refactor(punctuation): remove orphan code for handling lone punctuation

The case of lone punctuation is already handled at the top of restore(). The
removed if statement would never be called and would in fact raise an
AttributeError because the _punc_index named tuple doesn't have the attribute
`mark`.

* refactor(punctuation): remove unused argument

* fix(punctuation): correctly handle initial punctuation

Stripping and restoring initial punctuation didn't work correctly because the
string-splitting caused an additional empty string to be inserted in the text
list (because `".A".split(".")` => `["", "A"]`). Now, an initial empty string is
skipped and relevant test cases are added.

Fixes #3333
 2023-11-30 13:03:16 +01:00
Eren G??lge 3b8894a3dd Make style 2023-11-27 14:15:50 +01:00
Eren G??lge 11ec9f7471 Add hi in config defaults 2023-11-24 15:38:36 +01:00
Eren G??lge 32065139e7 Simple text cleaner for "hi" 2023-11-24 15:14:34 +01:00
Enno Hermann 2af0220996
fix: don't pass quotes to espeak (#3286) 
Previously, the text was wrapped in an additional set of quotes that was passed
to Espeak. This could result in different phonemization in certain edges and
caused the insertion of an initial separator "_" that had to be removed.
Compare:
$ espeak-ng -q -b 1 -v en-us --ipa=1 '"A"'
_ˈɐ
$ espeak-ng -q -b 1 -v en-us --ipa=1 'A'
ˈeɪ

Fixes #2619
 2023-11-24 12:25:37 +01:00
Edresson Casanova 11283fce07
Ensures that only GPT model is in training mode during XTTS GPT training (#3241) 
* Ensures that only GPT model is in training mode during training

* Fix parallel wavegan unit test
 2023-11-17 15:13:46 +01:00
Eren G??lge 44880f09ed Make style 2023-11-17 13:43:34 +01:00
Eren G??lge 26efdf6ee7 Make k_diffusion optional 2023-11-17 13:42:33 +01:00
Julian Weber fbc18b8c34
Fix zh bug (#3238) 2023-11-16 17:51:37 +01:00
Julian Weber 675f983550
Add sentence splitting (#3227) 
* Add sentence spliting

* update requirements

* update default args v2

* Add spanish

* Fix return gpt_latents

* Update requirements

* Fix requirements
 2023-11-16 11:01:11 +01:00
Edresson Casanova 73a5bd08c0
Fix XTTS GPT padding and inference issues (#3216) 
* Fix end artifact for fine tuning models

* Bug fix on zh-cn inference

* Remove ununsed code
 2023-11-15 14:02:05 +01:00
Julian Weber 04901fb2e4
Add speed control for inference (#3214) 
* Add speed control for inference

* Fix XTTS tests

* Add speed control tests
 2023-11-14 16:07:17 +01:00
Eren Gölge ac3df409a6
Merge pull request #3208 from coqui-ai/fix_max_mel_len 
fix max generation length for XTTS
 2023-11-13 14:32:56 +01:00
Eren G??lge 92fa988aec Fixup 2023-11-13 13:44:06 +01:00
WeberJulian b85536b23f fix max generation length 2023-11-13 13:18:45 +01:00
Eren G??lge b2682d39c5 Make style 2023-11-13 13:01:01 +01:00
Eren G??lge a16360af85 Implement chunking gpt_cond 2023-11-13 13:00:08 +01:00
Enno Hermann 3b1e7038bc
fix(formatters): set missing root_path attribute (#3182) 
Fixes #2778
 2023-11-09 16:49:52 +01:00
Aarni Koskela a8e9163fb3
xtts/tokenizer: merge duplicate implementations of preprocess_text (#3170) 
This was found via ruff:

> F811 Redefinition of unused `preprocess_text` from line 570
 2023-11-09 16:32:12 +01:00
Matthew Boakes 1b9c400bca
PyTorch 2.1 Updates (Weight Norm and TorchAudio I/O) (#3176) 
* Replaced PyTorch weight_norm With parametrizations.weight_norm

* TorchAudio: Migrating The I/O Functions To Use The Dispatcher Mechanism

* Corrected Code Style

---------

Co-authored-by: Eren Gölge <erogol@hotmail.com>
 2023-11-09 16:31:03 +01:00
Gorkem 66a1e248d0
torchaudio should use proper backend to load audio (#3179) 2023-11-09 16:28:39 +01:00
Julian Weber 03ad90135b
Add lang code in XTTS doc (#3158) 
* Add lang code in XTTS doc

* Remove ununsed config and args

* update docs

* woops
 2023-11-08 13:47:33 +01:00
Gorkem 78a596618a
Fix for exception on streaming if last chunk empty (#3160) 2023-11-08 11:32:02 +01:00
Julian Weber ce1a39a9a4
Add char limit warn (#3130) 
* Add char limit warning

* Adding v2 langs

* cached_property for cutlet

* Fix import
 2023-11-08 10:24:23 +01:00
Edresson Casanova 5f9ab6cfaa
Fix style 
Co-authored-by: Aarni Koskela <akx@iki.fi>
 2023-11-06 19:22:34 -03:00
Edresson Casanova 09fb317e6d Remove unused code 2023-11-06 17:36:32 -03:00
Edresson Casanova b146de4ce8 Bug fix on XTTS v2.0 Trainer 2023-11-06 20:26:01 +01:00
Edresson Casanova 1b6f8d0e46 Update unit tests and recipes 2023-11-06 20:25:06 +01:00
Edresson Casanova 72b2bac0f8 Load reference in 24khz to avoid issued with multiple sr references 2023-11-06 20:25:06 +01:00
Edresson Casanova 00294ffdf6 Update XTTS docs 2023-11-06 20:24:06 +01:00
Edresson Casanova 459ad70dc8 Add support for multiples speaker references on XTTS inference 2023-11-06 20:22:35 +01:00
Eren Gölge f0cb19ecca
Drop diffusion from XTTS (#3150) 
* Drop diffusion for XTTS

* Make style

* Drop diffusion deps in code

* Restore thrashed
 2023-11-06 20:15:49 +01:00
Eren G??lge 5d418bb84a Update docs 2023-11-06 18:48:41 +01:00
Eren G??lge 9bbf6eb8dd Drop use_ne_hifigan 2023-11-06 18:43:38 +01:00
Eren G??lge 9d54bd7655 Fixup XTTS 2023-11-06 18:13:58 +01:00
Edresson Casanova e45227d9ff
XTTS v2.0 (#3137) 
* Implement most similar ref training approach

* Use non-enhanced hifigan for test samples

* Add Perceiver

* Update GPT Trainer for perceiver support

* Update XTTS docs

* Bug fix masking with XTTS perceiver

* Bug fix on gpt forward

* Bug Fix on XTTS v2.0 training

* Add XTTS v2.0 unit tests

* Add XTTS v2.0 inference unit tests

* Bug Fix on diffusion inference

* Add XTTS v2.0 training recipe

* Placeholder model entry

* Add cloning params to config

* Make prompt embedding configurable

* Make cloning configurable

* Cheap fix for a cheaper fix

* Prevent resampling

* Update model entry

* Update docs

* Update requirements

* Code linting

* Add xtts v2 to sep tests

* Bug fix on XTTS get_gpt_cond_latents

* Bug fix on rebase

* Make style

* Bug fix in Japenese tokenizer

* Add num2words to deps

* Remove unused kwarg and added num_beams=1 as default

---------

Co-authored-by: Eren G??lge <egolge@coqui.ai>
 2023-11-06 14:58:18 +01:00
Aarni Koskela 38f6f8f0bb
Run `make style` & re-enable it in CI (#3127) 2023-11-06 11:36:37 +01:00
WeberJulian 1c98821359 Remove unused load_audio function 2023-10-27 22:27:18 +02:00
WeberJulian d4e08c8d6c Add features to get_conditioning_latents 2023-10-26 14:57:33 +02:00
WeberJulian c1133724a1 Move lang token add to tokenizer 2023-10-26 14:52:13 +02:00
WeberJulian 6fa46d197d Fix get_conditioning_latents when using only ne 2023-10-26 14:51:35 +02:00
Edresson Casanova 01839af926 Bug fix on XTTS masking training 2023-10-24 18:30:14 -03:00