#"entities" is not returned in scribe_v2 model response.

1 messages · Page 1 of 1 (latest)

dawn hedgeBOT
#
AI-generated Answer for ak_deepankar

Short answer: entities is only returned when you explicitly enable entity detection in your request.

What to do

  • Use the batch Speech-to-Text endpoint (POST /v1/speech-to-text) with model_id=scribe_v2 and set the entity_detection parameter. It accepts 'all', a single category/type, or a list (e.g., ["pii"] or ["pii","phi"]). When enabled, the response includes an entities array. (elevenlabs.io)
  • See the entity detection guide for an example response showing the entities field and SDK usage (e.g., transcription.entities). Note that entity detection is billed extra. (elevenlabs.io)

References

  • API reference: Create transcript (entity_detection parameter; entities returned when enabled). (elevenlabs.io)
  • Guide: Entity detection (example with entities in the response; additional cost; how to enable). (elevenlabs.io)

Sources:
Create transcript | ElevenLabs Documentation
Entity detection | ElevenLabs Documentation
Create transcript | ElevenLabs Documentation
Entity detection | ElevenLabs Documentation

steep crypt
#

still it is not returning. here is my request - const scribeConfig = {
model_id: "scribe_v2",
tag_audio_events: true,
diarize: true,
language_code: "eng",
entity_detection: entityTypes && entityTypes.length > 0 ? entityTypes : ["all"],
};

        const transcription = await client.speechToText.convert({
            file: audioFile,
            ...scribeConfig,
        } as any);