Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

strange generations issue... #78

Open
zethfoxster opened this issue Sep 29, 2023 · 9 comments
Open

strange generations issue... #78

zethfoxster opened this issue Sep 29, 2023 · 9 comments

Comments

@zethfoxster
Copy link

im completely brand new to text to speech gen, and im not sure if this is just a limitation of where we currently are in text to speech...or if this is a bug.

sometimes in minute or longer gens, the voice just goes off talking about random stuff thats not in the text that i wrote. if this is normal i apologize for opening an issue on it. otherwise, how can i keep the generation on topic so to speak in longer generations?

@zethfoxster
Copy link
Author

i will say tho, it sounded so natural that i had to do a double take lol...i wasnt ready for that. it almost sounded on topic. but i would much rather prefer it stuck to the script instead of adlibing....

@yesbroc
Copy link

yesbroc commented Oct 30, 2023

thats completely to do with bark lol, you can try lowering the temps to avoid randomness, otherwise you can regen it

@AIhasArrived
Copy link

@yesbroc hello, did you have good results with bark gui? If yes, how what configs etc?

@yesbroc
Copy link

yesbroc commented Oct 30, 2023

i had decent results, i used bark models 0, 6, and 8 (v2) with default config settings, text temp at 0.5, it did do the thing where it unnecessarily generated but not often

@AIhasArrived
Copy link

could you break down the process?

  1. train? Or use trained models (or both)
  2. open some tab to do cloning or use another repo with TTS something?
    etc
    This kind of stuff ^^^thx

@yesbroc
Copy link

yesbroc commented Oct 30, 2023

i used pretrained bark models and the v2 voices, these are automatically installed

@AIhasArrived
Copy link

What parameters and configs? hw much time, etc etc, more details please

@yesbroc
Copy link

yesbroc commented Oct 30, 2023

default settings you'd get from a fresh install, cpu offloading, bark v2 models (not small), i also made more than 1 and chose the best one.

@AIhasArrived
Copy link

Do you mind sharing a screenshot of your settings? Sorry for bothering

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants