-
-
Notifications
You must be signed in to change notification settings - Fork 71
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
Words repeated in whisper transcription with initial_prompt #165
Comments
Great code!!! My guess is the cpu is too slow. I ran into that issue with whisper-realtime before switching over to cuda. Works fine once I switched to cuda and my mic as audio source. |
Do we have any solution for this problem when using CPU? |
Hi, I tried using the code provided by you, but it is not printing any result other than Listening on the terminal. Any possible solution for the same. |
Hello, the issue with repeating words could be caused by many issues:
|
I am currently passing the audio file directly and not using the microphone as source. Does this account in any way for this strange behaviour of repeating the words? |
@shanky100 most probably not. Make sure to check if Whisper hallucinates when transcribing the entire file at once (instead of streaming it). You can also try removing the text conditioning. If Whisper hallucinates at the beginning and you keep conditioning it on the hallucination, you may be seeing a snowballing effect |
Hello all,
Thank you for doing this great work! I just updated this code to use faster whisper and I facing repeated words issue when I use initial_prompt param in the transcription method. the issue happened when I end my talk in some specific word something like Okay.
The issue:
The code:
Any help ?
The text was updated successfully, but these errors were encountered: