Comment by mightytravels

Use this Whisper derivative repo instead - one hour of audio gets transcribed within a minute or less on most GPUs - https://github.com/Vaibhavs10/insanely-fast-whisper