-
Notifications
You must be signed in to change notification settings - Fork 955
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
Helping Whisper get certain words right #9358
Comments
Generally speaking, the first thing I'd try is using the --initial_prompt argument with a sentence containing all the terms you expect to encounter. In vanilla Whisper, there's also this: And in Faster-Whisper-XXL (which I recommend), there's this: --reprompt {0,1,2} (experimental) 0) Disabled. 1) Inserts initial_prompt after the prompt resets. 2) Ensures that But this is mostly hypothetical for me since I prefer to just let it do whatever it does and fix it later using search and replace, since it's more consistent that way (and I have different multiple replace groups for different types of content). |
I would recommend not to touch "--initial_prompt" and use "--hotwords" for that. |
I thought there was something like that, but it's not in the help text... |
Hi, guys! I just wanted to know if there is any possibility of feeding Whisper inside Subtitle Edit with a list of terms/names of people that might be on the video for example so it recognizes them better.
Right now I realize that it gets certain names wrong in many different ways, so it is hard to find a pattern.
Have a great day!
The text was updated successfully, but these errors were encountered: