Replacing a retired cloud speaker recognition service
Major cloud providers retired their speaker recognition services in 2025 and 2026. Formanta replaces them, as a cloud API or on your own servers, and adds voice clone and deepfake detection on the same call.
What changes, what stays
Customers re-enroll, without extra effort
Voiceprints are vendor-specific and do not transfer. With consent, you can enroll customers from call recordings you already have, so they do not need to repeat phrases.
Your audio path
Formanta reads the call audio you already capture.
Deepfake detection and a choice of deployment
Every call can be checked for voice clones and replays. Use the cloud API, or run Voice Server on your own servers so no vendor's end-of-support notice can switch it off.
Three steps
Map the current flow
Enrollment, thresholds, fallback routes and what happens to unverified callers today.
Run on recorded calls
Formanta scores your historical audio, so you set thresholds on your own data and enroll consenting customers from the same recordings.
Switch with a fallback
Go live with your current step-up method kept in place until results hold. Unclear results keep going to it.
Plan your migration
Tell us which service you ran and how many calls go through voice verification.
← Back to Formanta