Video transcription - Cralo

AI agents call video.transcript to get a transcript of supported public YouTube and Vimeo media. The CLI’s data guarantees only text: timestamped segments are written to a local sidecar file referenced by meta.segments_file, and detected language, source, and duration are not part of the CLI’s compacted output.

Call with a public URL

cralo tools call video.transcript \
  --input '{"url":"https://www.youtube.com/watch?v=example","language":"en"}' \
  --max-price 2.00 \
  --json

For URL input, the CLI lazily downloads and verifies yt-dlp. For YouTube, it also sets up a separately licensed token helper that runs locally. All media fetching happens on your machine; the media is then sent for transcription and priced by resolved duration.

Prefer official captions

cralo tools call video.transcript \
  --input '{"url":"https://www.youtube.com/watch?v=example","use_official_transcript":true}' \
  --json

This YouTube-only path does not fall back to ASR when captions are unavailable.

Pricing and limits

ASR is billed per minute, with a one-minute minimum.
Maximum duration is 120 minutes.
Supported URL hosts are YouTube and Vimeo.
Use --async for long media and query the invocation later.

Agents must not use Cralo to bypass private, DRM-protected, or members-only media restrictions. Respect source-platform terms.

⌘I