Moving job to FailedJobRegistry (work-horse terminated unexpectedly; waitpid returned 9)

[orjanv]

I'm running this on a laptop with 8GB RAM and get this message when uploading a minute long mp3 file of ~300kb size. The docker stats confirm that the worker goes up in memory use. Is there anything to be done with this, make is spend more time, or how much RAM is required to run this?

[Johannes-Andersen]

Hey, the vram requirements would be the same as noted in the README in the open AI repo
https://github.com/openai/whisper#available-models-and-languages

When I have been developing locally it’s been working fine using the tiny model (I’m on a M1 MacBook Pro)

[orjanv]

Thanks, i just found that table myself. that would explain the issue. Will try with smaller model sizes later today :-) Thanks.

[orjanv]

This worked, the issue can be closed.