Thanks — you're asking if you can call

AgentSession.say()

while you're receiving a streaming API response. Short answer: yes. You can start

AgentSession.say()

without awaiting it so speech plays while your code continues processing the incoming stream.

say()

returns a

SpeechHandle

you can use to: • interrupt playback if the API response arrives early (

SpeechHandle.interrupt()

), and • check or coordinate playback state from your code. This pattern is useful for “holding” messages like “Let me check that for you” while an async tool/API is running. If you'd like, tell me which SDK/language you're using and I can show a short code example tailored to it. Sources: Audio customization | LiveKit Documentation | Agent speech and audio | LiveKit Documentation, with interactive elements