How come you have a killer audio input feature but can't transcribe the audio? Would be a killer feature to add a whisper integration or a way to transcribe audio. Don't even get me started with the possibilities.