You can record the audio of a LiveKit room using the Egress feature, which allows you to capture what the audio sounds like from the agent's perspective. The recommended approach is to start a room composite recorder when the agent joins the room. This will capture all audio (and video, if desired) from the room, effectively recording what an agent participant would hear. Recording ends when all participants leave and the room is closed. For audio-only recording, set the

audio_only

parameter to

True

in your request. The recordings can be stored in your preferred cloud storage provider (Amazon S3, Google Cloud Storage, or Azure Blob Storage). Here is a Python example for starting an audio-only room recording: Sources: examples/receive-audio/README.md | Recording agent sessions | LiveKit Docs | Session recording and transcripts | LiveKit Docs | examples/save_to_disk/src/main.rs