anyone have best practices to implement a turn-based speaking solution via VAPI? like if someone is done talking they are able to hit a button on the UI and that signals to VAPI that the person is done speaking