Krisp.ai Developer Hub

Welcome to the Krisp AI Voice SDK developer hub. You'll find comprehensive guides and documentation to help you start working with the AI Voice SDK as quickly as possible, as well as support if you get stuck. Let's jump right in!

Voice Isolation for Voice AI (VIVA) models are designed for inbound audio streams for Voice AI. Deployed on servers in front of the VAD to improve turn-taking, false interrupts and sometimes WER. Removes background voices and noise from TV, music, kids, cars and many other noise sources. Language and accent independent.

Noise Cancellation (NC-o, NC-i) models designed for microphone (outbound/uplink) and speaker (inbound/downlink) audio streams. Noise robust for use in contact centers, offices and home environments. Includes de-reverberation to remove room echo. Physical microphone, language and accent independent.

Background Voice Cancellation (BVC-o) models designed for microphone (outbound/uplink) audio stream to remove other voices near the primary speaker wearing a headset or ear buds and all other noises (includes AI Noise Cancellation). Fixes the cross-talk problem, as well as blocking competing voices in office and at home. No enrollment or training of the user’s voice is required. Language and accent independent.

Our AI Voice SDK is integrated and used across many different application and device types, supporting use cases where clear voice is critical to user experience. More details are available in the SDK Features section.

Model designed for converting the accent of a contact center agent in real-time from their local accent to that of the customer, improving customer experience while providing a significant ROI. Models are available for India-to-US, Philippines-to-US and LatAM-to-US. Local accent robust, gender independent, noise-robust and microphone/headset independent. No enrollment or training of the user’s voice is required.