Some notes on mlx-whisper - it's now really easy to run transcriptions through an Apple Silicon (and GPU) optimized Whisper model using Python on macOS