This looks really good, thanks! Really appreciate this and all the other Whisper implementations in this thread as I am sorting up transcriptions for my 120+ podcast episodes.