Transcribe a video

Transcription is asynchronous — it calls an external transcriber. The request returns 202 with an execution; you track that to completion, then read the transcript.

1. Kick off the transcript

curl -X POST https://api.clipping.cc/v1/items/itm_video_123/transcript \
  -H "Authorization: Bearer ck_live_…" \
  -H "Content-Type: application/json" \
  -d '{ "lang": "pt", "embed": true }'

{ "execution": { "execution_id": "exc_…", "status": "pending" } }

Field	Meaning
`lang`	Target language (e.g. `pt`). Omit to auto-detect.
`embed`	`true` → also generate embeddings so the transcript is searchable (consumes AI tokens).
`text`	Provide a transcript yourself — skips the external transcriber (no provider cost).

2. Wait for it

Poll the execution until done (see Asynchronous operations for a ready-made polling helper):

curl https://api.clipping.cc/v1/executions/exc_… \
  -H "Authorization: Bearer ck_live_…"

3. Read the transcript

curl "https://api.clipping.cc/v1/items/itm_video_123/transcript?lang=pt" \
  -H "Authorization: Bearer ck_live_…"

{
  "transcript": {
    "item_id": "itm_video_123",
    "lang": "pt",
    "text": "…",
    "segments": [{ "start": 0.0, "end": 4.2, "text": "…" }]
  }
}

Only embedded content is searchable. Pass embed: true when transcribing (or run a workflow that produces documents) so it shows up in semantic search.

Cost

Running the transcript draws a 0.01 coin flat fee plus the transcriber’s real cost (pass-through). Reading it back is 0.005 coin. A manual transcript (text provided) skips the provider cost.

Get started

Core concepts

Guides

Transcribe a video

1. Kick off the transcript

2. Wait for it

3. Read the transcript

Cost

​1. Kick off the transcript

​2. Wait for it

​3. Read the transcript

​Cost

1. Kick off the transcript

2. Wait for it

3. Read the transcript

Cost