Thanks — this is a great question. I’m responding to your question about whether you can capture agent and user audio in separate channels without using RoomCompositeEgress because of cost. Short answer: Yes — you can record agent and user audio separately with lower-cost egress options. Options and trade-offs • Track Egress (most cost-effective) ◦ What it does: Exports individual raw tracks (audio as Ogg/Opus) without transcoding or browser rendering. ◦ Why it helps: No Chrome rendering or heavy transcoding, so it uses far fewer compute/transcode minutes and is the lowest-cost option. ◦ How it works for dual-channel recording: Start two Track Egress jobs — one for the agent audio track, one for the user audio track. You’ll get two separate files (one per participant) which you can keep separate or combine later if needed. ◦ Limitation: Track Egress produces one track per job; it does not produce a single dual-channel file natively. • Track Composite Egress (recommended when you want per-participant AV files) ◦ What it does: Produces a file that muxes a participant’s audio (and optional video) into a single output per participant. ◦ Why it helps: Lower cost than RoomCompositeEgress because there's no full-room layout rendering; still performs some muxing/transcoding. ◦ How it works for dual-channel recording: Run two Track Composite Egress sessions — one for the agent, one for the user — to get independent files that already contain that participant’s audio channel. ◦ Trade-off: More processing than Track Egress, so slightly higher cost, but cheaper than full room composition. • Room Composite Egress ◦ You’re already using this and it supports dual-channel in one file, but it’s the most expensive because it renders the whole room (browser/Chrome rendering + composition + encoding). Recommendation • If you only need raw audio for transcription, archival, or post-processing: use Track Egress (two jobs). This is the cheapest and simplest. • If you want a per-participant muxed file (audio + video) without full-room composition: use Track Composite Egress (two jobs). Cheaper than RoomCompositeEgress but has some transcoding cost. • If you specifically need a single file with both channels already mixed into separate channels inside one container, RoomCompositeEgress is the straightforward option — otherwise you can merge the two track outputs offline (or via a lightweight serverless job) to produce a single dual-channel file. If you’d like, I can: • Show an example request for starting Track Egress or Track Composite Egress for two tracks in your SDK/language. • Help compare estimated costs between your current RoomCompositeEgress usage and Track Egress (I’ll need approximate egress durations or file sizes). • Suggest a simple FFmpeg command to merge two Ogg/Opus tracks into a single dual-channel file if you want a combined output without RoomCompositeEgress. Which would you like next — example code for starting Track Egress, a cost estimate (please share typical durations), or a merge command for combining two audio files? Sources: Universal Egress | Universal Egress | Universal Egress | Understanding and Estimating Pricing for Video Conference and Livestream Us...