File transcription

Submit audio or video, receive a timecoded transcript and optional summary.

Submitting a job

POST /v1/transcriptions
{
  "media_url": "https://your-bucket.example.com/meeting.mp4",
  "language": "zh",
  "target_languages": ["en"],
  "diarization": true,
  "summary": true
}

You receive a job_id. Processing is asynchronous and typically takes about 0.2× the duration of the audio.

Retrieving the result

GET /v1/transcriptions/{job_id}

On completion you receive an array of passages separated by speaker and carrying timecodes, plus translations and a summary if requested.

Callbacks

You can also supply a callback_url when submitting, and we will POST the result there on completion. Callbacks carry an X-Winteam-Signature header — verify it before using the payload.

Export formats

Results export as JSON, SRT, VTT, Word or PDF.