Thanks — I’m responding to your question about whether you can detect client-side noise cancellation so the worker can avoid applying it as well. Short answer: There is no built‑in way to query or detect whether a client is running noise cancellation. The documentation shows how to enable noise cancellation on the client (for example

krisp_enabled

in SIP) and how to enable agent/worker noise cancellation (BVC/BVCTelephony), but it does not provide a signal or API to tell a worker that the client already applied noise suppression. Recommended approaches you can use instead • Centralize the decision in your app configuration: Decide up front whether noise cancellation is applied on the client or on the worker, and enforce that via config (e.g., a flag in your session/room metadata). This prevents double-processing and audio degradation. • Coordinate via signaling or metadata: Have the client send a short metadata field or an initial control message when joining (e.g.,

noise_cancellation=client

), and have the worker read that to decide whether to enable server-side noise cancellation. • Use deployment policies: If you control the client builds (mobile/desktop/web), distribute builds or a server-side policy that enables only one side’s cancellation by default. • Default to worker-side when client unknown: If you cannot guarantee client cooperation, default to enabling noise cancellation on the worker and disable it on clients you control, or vice versa, depending on your priorities for latency and quality. Why this matters • Running both client and worker noise cancellation can degrade audio quality. The guidance is to ensure only one layer performs suppression. If you’d like, I can: • Suggest a minimal metadata schema or signaling message you can send from client to worker. • Provide an example flow for enforcing a single-source policy (client-first or worker-first). Which would help you most? Sources: sip | livekit/livekit_composite | DeepWiki | Choosing the right noise cancellation for your agent | LiveKit