Transcripts

Each result can include a signed link to its transcript. Links are for the requester and expire.

Examples use VIDEO_ID and clp_live_... as placeholders.

The link

A result with ok: true has a transcript_url, on a 200 and on a succeeded job. It's a path, /v1/transcript/:id with a token query parameter, so add the API's origin in front of it. The id is the result's artifact.cache_digest.

The token expires 15 minutes after the response that gave it to you.
The token only works for the account that requested it, and the request also needs an API key from that same account. A link on its own fetches nothing.

Fetch a transcript

Call GET /v1/transcript/:id with your key. In the example, replace the id and token placeholders, or use the transcript_url as it came:

curl -X GET "https://clypso-eta.vercel.app/v1/transcript/{id}?token={token}" \
  -H "Authorization: Bearer clp_live_..."

The response is the transcript as plain text in transcript:

JSON

{
  "id": "ID",
  "transcript": "TEXT"
}

Fetching a transcript counts toward your rate limit, and doesn't use your monthly limit.

When a link has expired

Get a new link in either of these ways:

Poll the job again, if the result came from a job. Each poll of a succeeded job signs a new link.
Send the same POST /v1/extract again with no Idempotency-Key, or a new one. While the result is cached, this returns right away and doesn't count against your monthly limit.

Don't reuse the old Idempotency-Key for this. A replay returns the stored response, including the old link.

Errors

Status	Code	Meaning
`401`	`invalid_api_key`	The API key is missing or invalid.
`401`	`invalid_signed_url`	The token is expired, altered or for another transcript, or your key is from another account. These all look the same on purpose.
`404`	`transcript_not_found`	The token is valid, but no transcript text is stored for this ID.
`429`	`rate_limited`	See Rate limits.