Voice ID API
Enroll once. Then verify (1:1) or identify (1:N) by voice.

How we compare

Voice biometrics is a thin market: AWS bundles it into Connect (per-minute, contact-center oriented), and Microsoft put Speaker Recognition behind Limited Access approval. Brainiall exposes the primitive directly — enroll / verify / identify — at a flat per-verification price, with no contact-center platform to adopt.

Prices are list-price approximations for orientation, not quotes. Always check each vendor's current pricing page.

Pricing

Enroll-once / verify-many: enrollment and verification are billed the same. Free tier is generous enough to wire up an end-to-end flow.

PAYG: $0.0075 / verification (Brainiall Voiceprint engine). Enrollment is billed at the same rate. Identification across a group counts as one verification regardless of group size.

Three calls: enroll, verify, identify

# 1. Enroll a voiceprint (repeat with more clips to strengthen it)
POST https://api.brainiall.com/v1/voice-id/enroll
  {"audio": "<base64 wav>", "speaker_id": "agent-42", "group_id": "acme"}
  -> {"speaker_id": "agent-42", "n_samples": 1, "enrolled": true}

# 2. Verify (1:1) — is this clip the enrolled speaker?
POST https://api.brainiall.com/v1/voice-id/verify
  {"audio": "<base64 wav>", "speaker_id": "agent-42", "group_id": "acme"}
  -> {"similarity": 0.81, "match": true, "threshold": 0.65}

# 3. Identify (1:N) — rank everyone enrolled in the group
POST https://api.brainiall.com/v1/voice-id/identify
  {"audio": "<base64 wav>", "group_id": "acme", "top_k": 5}
  -> {"candidates": [{"speaker_id": "agent-42", "similarity": 0.81}, ...]}

# Manage enrollments
GET    https://api.brainiall.com/v1/voice-id/speakers?group_id=acme
DELETE https://api.brainiall.com/v1/voice-id/speakers/agent-42?group_id=acme

~2 seconds of clear speech is enough for a usable voiceprint; more clips per speaker make it more robust. The threshold is returned in every response and accepted as a request parameter so you can tune the false-accept / false-reject trade-off for your use case.

What it's for

• Caller authentication: confirm a returning caller is who they claim to be from the first few seconds of the call, before any knowledge-based questions.
• Account-takeover & fraud signals: flag when the voice on a call doesn't match the account's enrolled voiceprint, or matches a known-fraud list.
• Speaker labelling at scale: pair with the Diarization API — diarize a recording into turns, then identify which enrolled person each turn belongs to.
• Access control for voice interfaces: gate a voice assistant or kiosk action on a 1:1 verification rather than a shared PIN.
• Privacy by design: the API stores only a fixed-length, irreversible embedding — the original audio is never persisted. You delete a person's voiceprint with a single call.

Consent & biometric data

A voiceprint is biometric data. By calling the enroll endpoint you attest that you have obtained the end-user's consent to create and store their voiceprint, and that your use complies with applicable biometric-privacy law (e.g. Illinois BIPA, GDPR special-category data). This mirrors the consent-attestation model used by comparable services such as AWS Connect Voice ID. See the Terms of Service and Data Processing Addendum for the full conditions.

Press kit & resources

What reviewers, integrators and procurement teams typically ask for.

More specialty APIs

Same single API key, same usage-based pricing, different problem solved.

Speaker DiarizationWho spoke when — standalone, no transcription

Identity VerificationFace detection, embedding & liveness

Pronunciation AssessmentPhone / word / sentence scoring

Voice ProPremium text-to-speech voices