We can keep it here. It's easier. All I'm curious about is how you imagine the program being able to give detailed feedback. Speech is very complex.