One thing that I've learned is that making sure that whisper is running on Kuda will accelerate the transcription process. And given the amount of time it takes, the transcribers good, but we could potentially go to a slightly higher model like Turbo instead of Tiny.