For AI agents: a documentation index is available at /llms.txt. Markdown versions of all pages can be requested by appending `.md` to the URL, or by setting the `Accept` header to `text/markdown`.
Skip to main content
Speech to TextFeatures

Entity Detection

Find and label personal, health, and payment card information in Batch transcripts.

Available in: Batch. Not available in Realtime or Agent STT.

Entity Detection finds potentially sensitive data in a transcript, such as names, dates of birth, health conditions, and credit card numbers, and labels each value with its entity class. For example, use it to:

  • Find sensitive data before you store or share a transcript, or redact it automatically with PII Redaction.
  • Extract details such as names, addresses, and account numbers into CRM records or forms.
  • Flag conversations for compliance review, such as calls where a card number was read out.
  • Analyze clinical conversations by the conditions, drugs, and procedures mentioned.

Enable Entity Detection​

To enable Entity Detection, set enable_entities to true and entities_config.detect_advanced_entities to true:

{
"type": "transcription",
"transcription_config": {
"language": "en",
"enable_entities": true,
"entities_config": {
"detect_advanced_entities": true
}
}
}

entities_config is only accepted when enable_entities is true.

Entity Detection output​

For the utterance "Mark was born on the 25th of July 2000.", the JSON transcript looks like this (timings omitted):

{
"results": [
{
"type": "entity",
"entity_class": "name_given",
"alternatives": [{ "content": "Mark", "speaker": "S1" }]
},
{ "type": "word", "alternatives": [{ "content": "was", "speaker": "S1" }] },
{ "type": "word", "alternatives": [{ "content": "born", "speaker": "S1" }] },
{ "type": "word", "alternatives": [{ "content": "on", "speaker": "S1" }] },
{ "type": "word", "alternatives": [{ "content": "the", "speaker": "S1" }] },
{
"type": "entity",
"entity_class": "dob",
"alternatives": [{ "content": "25th of July 2000", "speaker": "S1" }]
},
{ "type": "punctuation", "alternatives": [{ "content": ".", "speaker": "S1" }] }
]
}

Each detected value has type set to entity and an entity_class, such as name_given or dob. Entity Detection returns the most specific class that applies. For the full list, see Entity Detection classes.

To redact the values instead of labeling them, use PII Redaction. You can enable both in the same job.

Entity Detection and smart formatting entities​

enable_entities returns one of two kinds of entities:

Smart formatting entitiesEntity Detection
Configurationenable_entities: trueenable_entities: true and entities_config.detect_advanced_entities: true
Entity classescardinal, date, money, percentage, and other smart formatting classesEntity Detection classes, such as name_given, dob, and credit_card
spoken_form and written_formIncludedNot included
ModelsStandard and EnhancedAll models

A job returns one kind or the other, not both. On Melia 1 and Oak 1, enable_entities is only accepted with detect_advanced_entities set to true.

Supported languages for Entity Detection​

Entity Detection supports all transcription languages, except:

  • Bashkir
  • Basque
  • Esperanto
  • Galician
  • Interlingua
  • Irish
  • Maltese
  • Marathi
  • Mongolian
  • Urdu
  • Uyghur
  • Welsh

For Mandarin, use the Simplified output locale (cmn-Hans, the default). Entity Detection of Traditional characters (cmn-Hant) isn't supported yet.

Next steps​