Krisp, best known for AI noise cancellation, has opened its enterprise voice translation engine to developers. The new Krisp Voice Translation API — launched alongside Voice Translation v3 — does real-time, bidirectional speech-to-speech translation across 60+ languages, the same engine that hit 96% accuracy in a live healthcare deployment.
## What the Krisp Voice Translation API does
The API is built around a single WebSocket: speech goes in, translated speech and text come out, with SDKs in JavaScript and Python at launch and C++ on the way. Because it’s aimed at regulated, high-stakes use, domain control is there from day one — Custom Vocabulary and Dictionary features tune it to your terminology, while Accuracy QA automatically scores 100% of translated calls across four quality dimensions. A Quick Phrases feature delivers pre-written, compliant content as translated speech in any supported language.
## Why it matters
Live voice translation has been demoed for years but rarely trusted in production. Krisp’s pitch is the track record: in that healthcare deployment, 90% of multilingual calls completed end-to-end with no human interpreter and zero safety incidents across 8+ languages. Putting that behind a self-serve API is what turns it from a demo into real infrastructure for support, telehealth, and global teams.

Leave a comment