Thanks — great question. Yes, you can use noise cancellation in LiveKit. Brief details and recommendations: • LiveKit provides enhanced noise cancellation models such as

NC()

and

BVC()

(Background Voice Cancellation). • For voice-AI agents, use

BVC()

— it filters background noise and non-primary speakers so the user’s voice that reaches your STT engine is cleaner. • For SIP/telephony participants, use

BVCTelephony()

which is optimized for narrow-band telephone audio. • Apply noise cancellation on the agent side using the provided plugins (for example,

noise_cancellation.BVC()

for WebRTC participants or

BVCTelephony()

for SIP). • Use only one noise-cancellation model at a time to avoid double-processing and degraded audio quality. • If you apply agent-side noise cancellation, coordinate with frontend and SIP trunk settings (for example, disable

krisp_enabled

on the frontend when using agent-side filtering) to prevent conflicts. • Processing is local on the agent’s CPU (adds minimal latency, typically under 10 ms). There’s no direct API charge for this processing, but it consumes CPU resources. If you want, tell me: • Whether your participants are WebRTC or SIP/telephony, • If you’re using a voice-AI agent with STT, and • Whether you already have frontend-level noise suppression enabled (e.g.,

krisp_enabled

). With that, I can recommend the exact plugin and show a short example of how to enable it. Sources: Choosing the right noise cancellation for your agent | LiveKit