You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello everyone.
Whisper model works really well for me, but I sometimes get errors like hallucinations (ex a list of dot, or repeated phrases), cutted phrases (ex just on word) or sometimes also have language switch in long audio files.
I was thinking on building on process a anomaly detection tools. The goal is to detect abnormal chuncks with small nlp classification tools and language detections (optional) and if a chunck is abnormal, run it again modifying the prompt or anything, because most of the time if you start with an hallucination, you will destroy the whole process.
What do you guys think of that ? Is that something possible to integrate?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hello everyone.
Whisper model works really well for me, but I sometimes get errors like hallucinations (ex a list of dot, or repeated phrases), cutted phrases (ex just on word) or sometimes also have language switch in long audio files.
I was thinking on building on process a anomaly detection tools. The goal is to detect abnormal chuncks with small nlp classification tools and language detections (optional) and if a chunck is abnormal, run it again modifying the prompt or anything, because most of the time if you start with an hallucination, you will destroy the whole process.
What do you guys think of that ? Is that something possible to integrate?
Beta Was this translation helpful? Give feedback.
All reactions