API reference · Agents

Transcribe a dictated description

POSThttps://goblins-call-api.onrender.com/v1/agents/draft/transcribe

Turns a short recording into text, for a description that was spoken rather than typed. The studio uses it where the browser has no speech recognition of its own.

Needs a workspace API key.

Errors

400EMPTY_AUDIO

The body is empty.

413AUDIO_TOO_LARGE

The recording is larger than 8 MB; keep it under a minute.

415UNSUPPORTED_AUDIO

The body is not audio, or Content-Type does not say which kind.

429RATE_LIMITED

More than 60 recordings from this workspace in an hour.

502TRANSCRIBE_FAILED

The transcriber did not answer. Try again, or type it.

Every request can also answer 401, 429 and 500: see Errors.

Notes

  • Send the recording itself as the request body, with its type in Content-Type: audio/webm, audio/mp4, audio/ogg or audio/wav. Up to 8 MB, which is about a minute of speech. The language is detected.
curl -X POST "https://goblins-call-api.onrender.com/v1/agents/draft/transcribe" \
  -H "Authorization: Bearer $CRACKLE_API_KEY"
Response · 200 OK
{
  "text": "A receptionist for my dental clinic in Pune. Books appointments, answers questions about timings and fees, and speaks Hinglish."
}