Because hearing a familiar voice is no longer enough.
DeepFree™ embeds a mathematically derived authentication signal into every live call. AI voice clones cannot replicate it. Neither can recorded audio. The guarantee is cryptographic, not probabilistic.
Each shield represents a distinct technical claim in our US provisional patent filing. Together they form a four-factor authentication system that no synthesis technology — present or future — can defeat.
A voice clone without the enrolled device key cannot generate the correct HMAC carrier. The carrier check returns near zero regardless of which synthesis model is used.
The carrier rotates every 2 seconds. A recorded call contains a signal from an expired time window — cryptographically stale before replay is possible.
Authentication requires simultaneous possession of the user key, device key, and timing offset. A stolen credential without the physical device fails verification.
The authentication signal is embedded with a secret key-derived timing offset. A man-in-the-middle attacker aligns with the live stream, not the secret temporal slot — arriving at the wrong time on every attempt.
These properties are mathematical consequences of the cryptographic architecture. They do not degrade as synthesis technology improves. Click any shield to view the technical claim and mathematical basis.
If someone sent you a DeepFree™ DFV token via text or message, paste it below to confirm its cryptographic authenticity instantly.
All plans include cryptographic verification, Postgres-backed revocation, cross-session HMAC integrity, and NDA on request.
We work with security, fraud, and technology teams at financial institutions. Briefings include a live demonstration, architecture review, and security documentation package.
NDA available on request — Technical white paper and SOC 2 report provided under NDA
Embed a cryptographic carrier into live audio. AI voice clones cannot replicate it — the guarantee is mathematical, not probabilistic.
npm install deepfree
Requires AudioWorklet · Web Crypto API · getUserMedia — Chrome 66+, Firefox 76+, Safari 14.1+
The DeepFree iOS SDK brings cryptographic voice authentication directly into your mobile banking application — no server-side audio processing required for enrollment.
extractable flag is set to false — the key cannot be read by JavaScript, the OS, or any software attack. Authentication requires the physical device.Targeting iOS 16+ · Swift 5.9 · Requires AVAudioEngine & LocalAuthentication frameworks
Authenticate a live call by posting a signed audio chunk to the verification endpoint. Responses arrive in under 15 ms. All requests require a bearer token scoped to your institution.
curl -X POST https://api.deepfree.app/v1/verify \
-H "Authorization: Bearer $DF_API_KEY" \
-H "Content-Type: application/octet-stream" \
--data-binary @chunk.raw
TLS 1.3 required · Bearer token scoped per institution · Rate limit: 600 req/min