It's an interesting idea but the tech is probably not really there yet. It's one thing to have one person say something simple in a "command voice." It's another to transcribe, in real-time, overlapping conversations in a space. I wonder if people wore mics though.