careful-tiger-18580
02/13/2026, 11:46 AMtall-airline-16756
02/13/2026, 11:46 AMtall-airline-16756
02/13/2026, 11:46 AM"krisp_enabled": true option is only available on LiveKit Cloud or if you can use it on a self-hosted LiveKit server running on your own CPU VM.
Short answer: You can use krisp_enabled: true with self-hosted LiveKit. It is not limited to the paid LiveKit Cloud.
What to keep in mind when self-hosting:
• Krisp binaries or the noise-cancellation components must be available in your deployment environment or properly linked into your server image.
• Real-time noise cancellation is CPU work — ensure your VM has enough CPU resources to handle the processing load for the number of concurrent calls/participants you expect.
If you’d like, tell me:
• How you’re deploying LiveKit (Docker, Kubernetes, a single VM image), and
• Roughly how many simultaneous audio streams you expect,
and I can suggest practical deployment tips and CPU sizing guidance for running Krisp-enabled audio reliably.
Sources: sip | livekit/livekit_composite | DeepWiki | Announcing LiveKit Cloudcareful-tiger-18580
02/13/2026, 11:52 AM# Standard enhanced noise cancellation
<http://noise_cancellation.NC|noise_cancellation.NC>()
# Background voice cancellation (NC + removes non-primary voices
# that would confuse transcription or turn detection)
noise_cancellation.BVC()
# Background voice cancellation optimized for telephony applications
noise_cancellation.BVCTelephony()
are these all three models only available with livekit cloud api?
and cannot be use with self hosted livekit on my cpu vm?tall-airline-16756
02/13/2026, 11:52 AMNC() (standard enhanced noise cancellation)
• BVC() (background voice cancellation)
• BVCTelephony() (BVC optimized for telephony)
—are Cloud-only or if they can run on a self-hosted LiveKit server on your CPU VM.
Short answer: You can use all three models with a self-hosted LiveKit deployment. They are not limited to LiveKit Cloud.
What to keep in mind when self-hosting:
• These models are part of the livekit-plugins-noise-cancellation package and run locally in the agent process, so the CPU on your VM does the processing.
• You must install the noise-cancellation plugin in your agent environment and download the required model files (for example, via the agent tooling) before starting the agent.
• Real-time noise cancellation uses CPU resources — size your VM according to the number of concurrent audio streams you expect.
If you want, tell me:
• How you plan to deploy LiveKit (Docker, Kubernetes, single VM), and
• An estimate of concurrent audio streams/participants,
and I’ll suggest specific packaging and CPU-sizing tips and list the exact installation/download steps you’ll need.
Sources: Voice AI quickstart | LiveKit Documentation | Choosing the right noise cancellation for your agent | LiveKitcareful-tiger-18580
02/13/2026, 11:55 AMtall-airline-16756
02/13/2026, 11:55 AMlivekit-plugins-noise-cancellation~=0.2 is sufficient to use LiveKit's noise cancellation plugin.
Short answer: Yes — livekit-plugins-noise-cancellation~=0.2 is the correct package to get the noise cancellation plugin, and installing it is the main requirement to use the NC(), BVC(), and BVCTelephony() models.
Additional points to ensure it works on a self-hosted VM:
• Install the livekit-plugins-noise-cancellation package in the agent environment where the LiveKit agent/processing runs.
• Download and place the required model files (the agent tooling or plugin documentation provides the download step) before starting the agent so the models are available to the plugin.
• Ensure your VM has sufficient CPU capacity for real-time processing — noise cancellation runs locally in the agent process and consumes CPU proportional to the number of concurrent audio streams.
If you’d like, tell me:
• whether you’re using Docker, Kubernetes, or a plain VM image, and
• an estimate of concurrent audio streams,
and I’ll provide concrete install/download commands and CPU-sizing recommendations.