Challenges for this feature would be identifying who is currently speaking, and detecting changes to the current speaker.