Torch not compiled with CUDA enabled
报错 AssertionError: Torch not compiled with CUDA enabled
You have the wrong version of torch installed because of some AI tool that you've installed in the past. You need to uninstall torch, anything that touches it and then do pip purge so that the caches are cleared as well.
If you want to hit the easy button then I suggest you try out my tool transcribe-anything --device insane which uses insanely-fast-whisper but uses an internal isolated environment so that this problem doesn't happen.
This link may help you.
Note: You need to change the 'cu000' part to match your CUDA version. For example, change it to 'cu121' if you're using CUDA 12.1 or 'cu118' if you're using CUDA 11.8.