briefly experimenting with whisper, I put together a small api using fastAPI, just ~20 lines of py and you have a voice-to-text api made with MIT software... Awesome!