POST /v1/capture-verbal

When a consumer gives verbal consent during a call, submit the audio segment and transcript to this endpoint. eConsent’s classifier analyzes the transcript for affirmative consent, coercion, and sentiment, then generates a signed consent record if the consent meets the TCPA standard.

This endpoint bills 1 verbal capture unit per request.

Caution

The classifier will reject consent that is coerced, reluctant, or ambiguous. Phrases like “I guess so,” “fine, whatever,” or “if I have to” will return valid: false with a coercion flag. This is intentional: weak consent is a liability in court.

Request

POST https://voice.econsent.org/v1/capture-verbal
Authorization: Bearer ec_voice_YOUR_KEY
Content-Type: application/json

Body

{
  "call_id": "call_xyz789",
  "phone": "+15551234567",
  "calling_entity": "Acme Insurance LLC",
  "captured_at": "2026-04-10T15:42:33Z",
  "audio_segment_url": "https://storage.yourplatform.com/segments/consent.wav",
  "transcript": "Yes I authorize Acme Insurance to call and text me about insurance quotes using AI assistants",
  "agent_disclosure_text": "Do you authorize Acme Insurance to contact you using AI assistants for quote follow-ups?",
  "consent_scope": ["ai_voice", "ai_sms", "telemarketing"],
  "language": "en-US"
}

Field	Type	Required	Description
call_id	string	No	Your internal call identifier
phone	string	Yes	Consumer phone in E.164
calling_entity	string	Yes	Legal entity name
captured_at	string	No	ISO 8601 timestamp (defaults to now)
audio_segment_url	string	Yes	URL to the audio segment (eConsent fetches and stores the SHA-256 hash)
transcript	string	Yes	Text transcript of what the consumer said
agent_disclosure_text	string	No	What the agent said before the consent prompt
consent_scope	array	Yes	Scopes: ai_voice, ai_sms, telemarketing, transactional
language	string	No	BCP 47 code, defaults to en-US

Response: Valid

{
  "consent_id": "cst_lx8ka1b2c3d4e5f6",
  "signature": "9c2f3a8b4d5e6f7a8b9c0d1e2f3a4b5c...",
  "evidence_url": "https://verify.econsent.org/c/cst_lx8ka1b2c3d4e5f6",
  "audio_attestation": {
    "stored": true,
    "sha256": "e3b0c44298fc1c149afbf4c8996fb924..."
  },
  "valid": true,
  "classifier_result": {
    "verdict": "accept_consent",
    "affirmative_consent_score": 1.0,
    "coercion_score": 0.0,
    "sentiment": "positive",
    "backend": "rules"
  },
  "warnings": []
}

Response: Rejected

{
  "consent_id": null,
  "valid": false,
  "rejected_reason": "consent_phrase_too_weak_or_coerced",
  "classifier_result": {
    "verdict": "no_signal",
    "affirmative_consent_score": 0.2,
    "coercion_score": 0.95,
    "sentiment": "negative",
    "backend": "rules"
  },
  "warnings": [
    "Classifier verdict: no_signal",
    "High coercion score detected",
    "Affirmative score: 0.20"
  ]
}

Classifier Verdicts

Verdict	Meaning	Action
accept_consent	Strong, genuine affirmative consent	Consent record created and signed
review_consent	Affirmative phrase detected but flagged (coercion, negative sentiment)	Consent NOT created. Re-prompt the consumer with clearer language.
accept_revocation	Consumer expressed an opt-out	Consent NOT created. Call POST /v1/revoke instead.
no_signal	No consent or opt-out phrase detected	Consent NOT created. The transcript did not contain a recognizable consent phrase.
ambiguous	Conflicting signals	Consent NOT created. Needs human review.

Code Examples

Node.js

const result = await voice.captureVerbal({
  callId: 'call_xyz',
  phone: '+15551234567',
  callingEntity: 'Acme Insurance LLC',
  audioSegmentUrl: 's3://bucket/segment.wav',
  transcript: 'Yes I authorize Acme to call me about insurance',
  agentDisclosureText: 'Do you authorize Acme to contact you?',
  consentScope: ['ai_voice'],
});

if (result.valid) {
  console.log(`Consent captured: ${result.evidenceUrl}`);
} else {
  console.log(`Rejected: ${result.rejectedReason}`);
}

Python

result = voice.capture_verbal(
    phone="+15551234567",
    calling_entity="Acme Insurance LLC",
    audio_segment_url="s3://bucket/segment.wav",
    transcript="Yes I authorize Acme to call me about insurance",
    agent_disclosure_text="Do you authorize Acme to contact you?",
    consent_scope=["ai_voice"],
)

if result.valid:
    print(f"Consent captured: {result.evidence_url}")
else:
    print(f"Rejected: {result.rejected_reason}")