They’d just have an array of microphones everywhere and isolate each voice - rooms only need n+1 microphones where n is the maximum number of people. That’s already simple to do today, and it’s not even that expensive.