The question title explained clearly my problem: How can I force avatar to listen until user finish (by sending a signal like pressing a button)? This can be done in OpenAI Voice API.
