Can I currently put a VAD module in the pipeline and only send audio when there is an active conversation? Feel like just that would solve the problem?