search

silvervens / Coqui-TTS-Models-Sonic-Characters

Instead of using other greedy "AI" websites, why not run it on your PC! You can use the "issues" to post your model here after training it via coqui TTS!!! Make sure to list which is trained (Tacotron, VITS, and more) including which one have a vocoder or NOT!
2 stars 0 forks source link

Training Model Attempt #1

Open skshadan opened 1 year ago

skshadan commented 1 year ago

Hello, guide me how can i train it locally, on my system???

skshadan commented 1 year ago

also, how can i use your trained model?????

silvervens commented 1 year ago

Oof, this has been a long time. To use my pre-trained model.

also, how can i use your trained model?????

To run you can download TTS via python and setup a basic script. https://youtu.be/MYRgWwis1Jk

The YT channels that host local trainings can also be found by searching up Google. Or literally using Jupyter and requirements. fulfilled.

skshadan commented 1 year ago

everything is working fine, what i want is a good model ( any pre trained model ) or how can i train my own model easily locally?

skshadan commented 1 year ago

Do you have any pre-trained models available?

silvervens commented 1 year ago

Do you have any pre-trained models available?

Only Silver unfortunately, spent all of my google colab monthly subscription on it. No more powers left. If you wished to submit other characters feel free to do so :D

skshadan commented 1 year ago

also i am getting this error when i continue my training?

Traceback (most recent call last): File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1591, in fit self._fit() File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1548, in _fit self.test_run() File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1466, in test_run test_outputs = self.model.test_run(self.training_assets) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\models\vits.py", line 1442, in testrun wav, alignment, , _ = synthesis( File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\synthesis.py", line 186, in synthesis model.tokenizer.text_to_ids(text, language=language_name), File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\tokenizer.py", line 108, in text_to_ids text = self.text_cleaner(text) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\cleaners.py", line 125, in phoneme_cleaners text = en_normalize_numbers(text) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\english\number_norm.py", line 92, in normalize_numbers text = re.sub(_comma_number_re, _remove_commas, text) File "C:\Users\shada\AppData\Local\Programs\Python\Python38\lib\re.py", line 208, in sub return _compile(pattern, flags).sub(repl, string, count) TypeError: expected string or bytes-like object

silvervens commented 1 year ago

also i am getting this error when i continue my training?

Traceback (most recent call last): File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1591, in fit self._fit() File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1548, in _fit self.test_run() File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1466, in test_run test_outputs = self.model.test_run(self.training_assets) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\models\vits.py", line 1442, in testrun wav, alignment, , _ = synthesis( File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\synthesis.py", line 186, in synthesis model.tokenizer.text_to_ids(text, language=language_name), File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\tokenizer.py", line 108, in text_to_ids text = self.text_cleaner(text) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\cleaners.py", line 125, in phoneme_cleaners text = en_normalize_numbers(text) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\TTS\tts\utils\text\english\number_norm.py", line 92, in normalize_numbers text = re.sub(_comma_number_re, _remove_commas, text) File "C:\Users\shada\AppData\Local\Programs\Python\Python38\lib\re.py", line 208, in sub return _compile(pattern, flags).sub(repl, string, count) TypeError: expected string or bytes-like object

I have no idea, perhaps check requirements?

skshadan commented 1 year ago

yeah! i tried! btw loved your model! it's soo good mann!! but too much noisy!

silvervens commented 1 year ago

yeah! i tried! btw loved your model! it's soo good mann!! but too much noisy!

Is it that you need the audios and things to set it up??? Like the CIVS and the Audio samples?

skshadan commented 1 year ago

Yess! can u pls guide me on- how to make high quality dataset? how to train properly? how much data is required to get a good voice? how to finetune a model?

silvervens commented 1 year ago

Yess! can u pls guide me on- how to make high quality dataset? how to train properly? how much data is required to get a good voice? how to finetune a model?

I literally poured in tons of Audio samples and used the subtitle tool to make a subtitle out of it and then did more. It's a painful process. All it was from a tutorial, I don't remember much, but seemingly I kinda gave up since my voice is actually Silver-like. So...

Don't think I have that in my google drive, I use the google colab and inserted it via directories ending up with a finished fine-tuned model... Wait... Lemme see...

Go to google colab and skip to the fine tune and continue from there. image

silvervens commented 1 year ago

That way you can continue training my Silver's voice model :D

Don't forget to send it here, I will mark this as a contribution and will replace the model with yours trained, since my google colab subscription is cancelled.

skshadan commented 1 year ago

can u please the google collab link?

silvervens commented 1 year ago

The link is with the page... It's there, just scroll DOOOWN!

Scrolly If I'm late :(

skshadan commented 1 year ago

Yeah! got it man! Thanks! by the way do you have any other good models available for coqui??? also do u know about finetuning?? how can i finetune the model??? also one more question - your model have some noise when it is generating the audio.

silvervens commented 1 year ago

Yeah! got it man! Thanks! by the way do you have any other good models available for coqui??? also do u know about finetuning?? how can i finetune the model??? also one more question - your model have some noise when it is generating the audio.

I don't know much on how to fine tune things. Although, when fine tuning I leave it on google colab doing its thing. Apparently, it keeps on getting better the more iterations it had.

I have no idea what the noise is, but I think... Hmm... Prolly' something that has to do with tensor flow thingy...

I got no other models unless someone decided to train/make a model from Coqui that is for example Shadow, Sonic and more. I will be accepting Sonic characters sent from the "issues" to be added here.

skshadan commented 1 year ago

sure! i will try to train sonic model :))) also much data needed??

silvervens commented 1 year ago

sure! i will try to train sonic model :))) also much data needed??

https://drive.google.com/drive/folders/1B2ZiSnMjNdcQiP8i2DkX7fCo_q6fy8jB?usp=sharing

Here's the training data of Silver the hedgehog... It's a lot...

skshadan commented 1 year ago

hey, when i am training getting this error - Traceback (most recent call last): File "train_vits.py", line 96, in main() File "train_vits.py", line 88, in main trainer.fit() File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\trainer.py", line 1610, in fit remove_experiment_folder(self.output_path) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\trainer\generic_utils.py", line 64, in remove_experiment_folder fs.rm(experiment_path, recursive=True) File "E:\KL2.0\CODEZ\Coqui\tts-coqui\TTS\venv\lib\site-packages\fsspec\implementations\local.py", line 172, in rm shutil.rmtree(p) File "C:\Users\shada\AppData\Local\Programs\Python\Python38\lib\shutil.py", line 730, in rmtree return _rmtree_unsafe(path, onerror) File "C:\Users\shada\AppData\Local\Programs\Python\Python38\lib\shutil.py", line 608, in _rmtree_unsafe onerror(os.unlink, fullname, sys.exc_info()) File "C:\Users\shada\AppData\Local\Programs\Python\Python38\lib\shutil.py", line 606, in _rmtree_unsafe os.unlink(fullname) PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'E:/KL2.0/CODEZ/Coqui/tts-coqui/TTS/recipes/ljspeech/vits_tts/bheem-voice-August-24-2023_11+27AM-409db505\events.out.tfevents.1692856674.Shadaen'

silvervens commented 1 year ago

I personally have no Idea... I use google colab! Did you use jupyter notebook?

Check all the important libraries.

skshadan commented 1 year ago

Hello, thanks for the help. I want to know what the process is for creating a dataset to analyze emotions in speech. Is the dataset created based on emotions, or is there a process for capturing emotions in speech?

skshadan commented 1 year ago

???

skshadan commented 1 year ago

can you please tell, how much time it took for you to train the model??

silvervens commented 1 year ago

Only audio clips unfortunately... You can go to the link and see how much data I've put... No emotions unfortunately, due to how limiting audio resources are...

silvervens commented 1 year ago

sure! i will try to train sonic model :))) also much data needed??

https://drive.google.com/drive/folders/1B2ZiSnMjNdcQiP8i2DkX7fCo_q6fy8jB?usp=sharing

Here's the training data of Silver the hedgehog... It's a lot...

this!

skshadan commented 1 year ago

hey bro- try this and tell me!!

https://huggingface.co/sk0032/coqui-tts-model

silvervens commented 1 year ago

hey bro- try this and tell me!!

https://huggingface.co/sk0032/coqui-tts-model voice.zip

Is it my model that you trained or silver's????

Tested now ended up with these results:

silvervens commented 1 year ago

voice.zip

skshadan commented 1 year ago

yoo!! hey man, can i get ur mail or discord?

silvervens commented 1 year ago
Sure! Discord as “silvervens” LITERALLY!!! Sent from Mail for Windows From: ShadanSent: Friday, September 29, 2023 6:18 PMTo: silvervens/Coqui-TTS-Models-Sonic-CharactersCc: SilverVens; CommentSubject: Re: [silvervens/Coqui-TTS-Models-Sonic-Characters] Training Model Attempt (Issue #1) yoo!! hey man, can i get ur mail or discord?—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>