Thanks — I’m responding about whether Egress can be initialized but hold off on recording until the user’s track is published (your countdown scenario). Short answer: Egress cannot be initialized and then wait to start recording only once a track is published. Once an Egress process starts, it will capture media from the room from that start time. That means if you start Egress before the user publishes audio/video, the Egress process is already running and any silence or empty frames (or other published tracks) will be recorded. Three practical ways to implement your countdown without recording the countdown period: 1) Delay starting Egress until the countdown finishes (recommended) • Let the user connect to the room and run the countdown client-side (or via a backend timer). • When the countdown ends, start Egress via your server SDK or Egress API. • Recording will begin only after you start Egress, so the countdown won’t be captured. 2) Start Egress early but only publish the user’s tracks after the countdown • If you must start Egress earlier (for example, to render a composite layout), keep the user silent/unpublished until countdown ends. • Be aware: the Egress process is still running, so any audio/video published by others or default room state could be captured. The countdown itself will not be recorded if the user doesn’t publish, but the Egress start time remains earlier. 3) Start Egress early and trim the recording after the fact • Start recording at countdown start, then programmatically trim or discard the initial portion in post-processing so the final asset begins at the user’s speech. • Use this if you need Egress running early for technical reasons but can do post-processing. Recommended approach: Delay starting Egress until after the countdown. It’s the simplest and guarantees no unwanted recording of the countdown. If you’d like, tell me: • Which Egress type you plan to use (Room Composite, Track Egress, or File/Stream), and • Whether you need Egress running early for layout/render reasons. I can then provide the exact API/SDK call pattern or an example flow for your chosen approach. Sources: Universal Egress | examples/frontdesk | livekit/livekit_composite | DeepWiki