Enhanced noise cancellation

Overview

LiveKit cloud offers access to advanced models licensed from Krisp to remove background noise and ensure the best possible audio quality. The models run locally, with no audio data sent to Krisp servers as part of this process and negligible impact on audio latency or quality.

The filter is available in either an outbound configuration (where noise is removed by the sender) or in an inbound configuration (where noise is removed by the receiver).

Background voice cancellation (BVC) is available for inbound server-side SDKs, such as Python. This model removes extra background speakers in addition to background noise, providing the best possible experience for voice AI applications.

Recommendation

For most 1:1 voice AI applications, LiveKit recommends using the BVC model in the LiveKit Agents Python SDK as the most effective solution.

Here is a comparison to illustrate the improvement over standard WebRTC noise suppression:

Original

WebRTC noiseSuppression

LiveKit Cloud enhanced noise cancellation

Supported platforms

Consult the following table for a complete overview of feature support.

Platform	Outbound	Inbound	BVC	Package
Web	✅	❌	❌	@livekit/krisp-noise-filter
Swift	✅	❌	❌	LiveKitKrispNoiseFilter
Android	✅	❌	❌	io.livekit:krisp-noise-filter
Flutter	✅	❌	❌	livekit_noise_filter
React Native	✅	❌	❌	@livekit/react-native-krisp-noise-filter
Unity	❌	❌	❌	N/A
Python	❌	✅	✅	livekit-plugins-noise-cancellation
Node.js	❌	❌	❌	N/A
Telephony	✅	✅	❌	LiveKit SIP documentation

Usage instructions

Use the following instructions to integrate the filter into your app.

Installation

Install the noise cancellation package from PyPI:

pip install livekit-plugins-noise-cancellation

Usage in LiveKit Agents

Include the filter in the constructor for VoicePipelineAgent or MultimodalAgent:

from livekit import agents
from livekit.plugins import noise_cancellation

# For voice pipeline agent
pipeline_agent = agents.pipeline.VoicePipelineAgent(
    # ...,
    noise_cancellation=noise_cancellation.BVC()
)

# For multimodal agent
multimodal_agent = agents.multimodal.MultimodalAgent(
   # ...,
   noise_cancellation=noise_cancellation.BVC()
)

Usage with AudioStream

Apply the filter to any individual inbound AudioStream:

stream = rtc.AudioStream.from_track(
    track=track,
    noise_cancellation=noise_cancellation.NC(),
)

Available models

There are two noise cancellation models available:

# Standard enhanced noise cancellation
noise_cancellation.NC()

# Background voice cancellation (NC + removes non-primary voices 
# that would confuse transcription or turn detection)
noise_cancellation.BVC()

Compatibility

The Python noise filter is currently supported only on Mac and Linux platforms.