Whisper wants to kill me O.o #1210
Replies: 2 comments 4 replies
-
Hey, I suspect your problem is, that you convert the audiodata from int16 to float32. In this step you have to rescale the data by deviding by 2^15. Whisper returns nonsense, because the input is several orders of magnitude higher than expected, which leads to unpredictable results. |
Beta Was this translation helpful? Give feedback.
-
I got that message same. Detected language: en |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi there,
i tried to use Whisper to transcribe my voice in real time. The model did not transcribe, but was able to threaten me. This was only the case with the "medium" or "large" model.
Code:
Output from the "large" model:
Other crazy Output:
I have also received "I'm going to kill you" as an output.
What exactly happened here?
Beta Was this translation helpful? Give feedback.
All reactions