Can I use session.say() with a realtime model?

[LiveKit-Community]

This question originally came up in our Slack community and the thread has been consolidated here for long-term reference.

Has anyone achieved realtime model + session.say() or session.generate_reply()?

I can make the model speak first but it doesn’t follow the exact context.

[LiveKit-Community]

session.say() can’t be called directly for a realtime model.

For realtime agents, use generate_reply() instead. See this example:

examples/voice_agents/realtime_video_agent.py