Hello, everyone. Does anyone know if it's possible to access the model's output (GPT's text) while user is talking with Vapi? We need GPT's answer in real-time for lip sync and emotions as it contains SSML.