250 character Limit - How to get over it? #3548
Replies: 5 comments
-
I guess if you meant the limitation of text prompt in hugging face demo which is (200 characters ) you can modify this changes by app.py --> line 200 you can make changes to 500 (if len(prompt) > 500:) |
Beta Was this translation helpful? Give feedback.
-
i tried making that change and it is still stopping at 200, is it somewhere else in the code as well? |
Beta Was this translation helpful? Give feedback.
-
Hi ,
you can modify changes with tokenizer.py which you can find it in TTS/TTS/tts/layers/xtts/tokenizer.py
line 624 you can defined the limitation for text input .
Also in training.py that you pass the language and dataset and ... search for " max_text_length=200," and change it to your desirer but it is not recomended .
after all the above if your still get error of text length your dataset must be chunk of audio between 3-6 second and not more than 12 second if you intend to fine tune .
Best regards
Bahareh Arghavani Nobar
Graduate research assistant
SAIL LAB<https://sail-lab.org/>
University of New Haven
Email: ***@***.***
Tel: 860 944 5353
[cid:732fe70c-00c2-4538-91b8-ed2b7368b091]
[https://ci3.googleusercontent.com/proxy/DkROAS12GWCwh0_8bxbCv6RlE4gMGP6HARzsNsA2KcSH0BycdFLGLeaOe4xIrHHuu7Ub2MNs7TSnzBaxYL-R32Ol5miKoswHnkzrCTseRr9h-S7aErKd5xwNqlVtl08jE15GNC3tTIgN=s0-d-e1-ft#http://orig13.deviantart.net/ab15/f/2017/047/9/f/signature_1_by_mehrjou-daz8smo.png]<http://thinkbeforeprinting.org/>
…________________________________
From: TalonPower ***@***.***>
Sent: Monday, March 4, 2024 2:34 PM
To: coqui-ai/TTS ***@***.***>
Cc: Arghavani Nobar, Bahareh ***@***.***>; Comment ***@***.***>
Subject: Re: [coqui-ai/TTS] 250 character Limit - How to get over it? (Discussion #3548)
You don't often get email from ***@***.*** Learn why this is important<https://aka.ms/LearnAboutSenderIdentification>
[EXTERNAL SENDER]
i tried making that change and it is still stopping at 200, is it somewhere else in the code as well?
—
Reply to this email directly, view it on GitHub<#3548 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A5O7GQUF7XLZ5L7HJ7TNZCDYWTECVAVCNFSM6AAAAABCQ5TWI6VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DMNZRGE4DO>.
You are receiving this because you commented.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
ok so i forked the coqui tts then i changed the tokenizer.py. in the colab i changed the requirements :https://github.com/TalonPower/xtts2-hf/blob/main/requirements.txt to point at mine and still get same result. the 200 limit. also i made edits in my colab to change the look and feel and it still looks the same my app.py: |
Beta Was this translation helpful? Give feedback.
-
anyone fixed the 250 cahracter limit issue pls |
Beta Was this translation helpful? Give feedback.
-
for xtts v2, how do i change the 250 character limit? i know the api does this by itself, but i don't want to use the api as i need to pass some extra parameters. i can split the text into smaller chunks (sentences or paragraphs) but i need to generate 500 characters, how can i do that?
Beta Was this translation helpful? Give feedback.
All reactions