Enhanced noise cancellation
Krisp uses advanced AI models to identify and remove background noise in realtime. Noise cancellation can be applied to both inbound or outbound audio streams and can be integrated either on the client side or server side.
LiveKit uses the Krisp SDK and embedded AI models. Noise cancellation is done locally in your agent, LiveKit client, or LiveKit SIP.
Inbound and outbound noise cancellation
Inbound noise cancellation removes unwanted noise from the incoming audio, ensuring clearer sound for the listener. For example, when a customer calls your service, Krisp eliminates their background noise so your agent can hear them clearly.
Outbound noise cancellation removes background noise from your microphone, so the person on the other end hears only your voice. For instance, when making a call from a noisy contact center, Krisp suppresses background sounds, allowing your voice to come through clearly.
Client-side and server-side noise cancellation
Client-side noise cancellation runs on the user’s device, filtering noise before transmitting the audio.
Server-side noise cancellation processes audio centrally, making it particularly useful for applications such as conversational AI. In this setup, Krisp removes background noise and secondary voices before the audio reaches Voice Activity Detection (VAD) or Speech-to-Text (STT) models, improving transcription accuracy and turn detection.
LiveKit's Krisp integration
You can enable Krisp outbound (microphone) noise cancellation using LiveKit client SDKs for JavaScript, Android, and iOS. To learn more, see Noise & echo cancellation.
You can also enable Krisp noise cancellation for incoming and outgoing calls. LiveKit SIP supports both inbound and outbound, server-side noise cancellation. To learn more, see Noise cancellation for calls.