Screen Cut Pro

WebRTC (LiveKit)

Experimental. The LiveKit room recorder works for capturing per-track files, but the editor side is still maturing — multi-cam timeline UI, per-track grid composite, and per-stream lanes are in flight. Expect rough edges, particularly around editing sessions with more than two participants.

Screen Cut Pro can join a LiveKit room as a non-publishing observer and capture each remote participant’s video and audio to its own file — ready to edit, mix, and export. Originally built for recording AI‑agent interactions cleanly, the same flow works for podcasts, interviews, and small remote calls.

Why this is useful

Cloud egress (LiveKit’s server-side recording, Zoom Cloud Record, etc.) composites the call into one mixed file. You lose the ability to fix one person’s audio, mute someone in post, re-time shots, or rebalance the conversation. Screen Cut Pro records each participant’s tracks separately so the editor has real material to work with.

Connecting

Click the LiveKit button in the HUD. In the dialog, paste your LiveKit project URL and a JWT access token, then click Connect.

The token must grant roomJoin + the room name you want to record. The recorder doesn’t need publish permissions — subscribe-only is enough.

Generate a token with the LiveKit CLI: 

livekit-cli create-token \
  --api-key $LIVEKIT_API_KEY --api-secret $LIVEKIT_API_SECRET \
  --room my-room \
  --identity recorder \
  --join

Recording

Once connected, the dialog shows the live participant list with each participant’s currently published tracks (camera, microphone, screen-share, etc.). Click Record to start; click Stop when you’re done. Late joiners' tracks are picked up automatically.

What gets saved

Each recording produces a session bundle inside Screen Cut Pro’s temporary recordings folder, containing:

  • One .mp4 per remote video track (H.264).
  • One .m4a per remote audio track (AAC).
  • A livekit-room.json manifest with each track’s participant SID, identity, source, start/end offsets, and dimensions or sample rate.

Editor handoff

Click Open in Editor after Stop. The editor mounts the first participant’s video and audio as the main video + microphone tracks. Every other audio track lands on the music lane labeled <participant name> (microphone), so all voices play back mixed. Other participants’ video tracks stay in the recording bundle for the multi-cam timeline UI we’re still building.

Privacy

The recorder joins the room with whatever identity your token assigns. Other participants can see it in the room’s participant list. Be transparent about recording — use a clearly named identity like recorder and disclose recording to the call.

Previous iOS Screen