Krisp

Krisp installs as a virtual microphone and speaker on your device, intercepting audio before it reaches whatever conferencing tool you are already using — Zoom, Teams, Google Meet, or anything else. In-meeting, it runs noise cancellation, echo removal, and cross-talk suppression in real time. After the call, the AI note taker surfaces a transcript, recording, and summary without you having to prompt it. The entire workflow is passive: you join the call, Krisp handles the rest.

The accent conversion capability is where Krisp makes a claim most competitors in the noise-cancellation space have not entered. The vendor describes it as real-time AI reshaping of a speaker’s accent for the listener — intended to reduce miscommunication and reduce the cognitive load on agents who field calls in a non-native accent. For individual users, this runs on the speaker side. For call centers, a listener-side variant is available through the Call Center AI platform.

For individuals and small remote teams, Krisp fits cleanly into any stack because it requires no changes on the meeting platform side. For call centers and BPOs, the vendor segments this into a separate product — Call Center AI — which adds speech analytics, call scoring, compliance monitoring, and a real-time agent assist layer. These are not features accessible from the consumer install; they are a distinct platform with a distinct sales motion.

An API is available for developers building voice AI pipelines. The vendor describes voice isolation and turn-taking improvements as specific targets for AI agent use cases, with noise cancellation and accent conversion available for human-to-human call integrations. Voice translation via API is listed as available on a self-serve basis.