YouTube Transcript & Subtitle Scraper

Scrape YouTube transcripts and subtitles for any public video. Give one or more video IDs or URLs and get every available caption track — language, whether it is auto-generated (ASR) or human-authored, whether it is translatable, and the downloadable timedtext URL — together with the video's title, channel, length and view count.

Unofficial. This Actor is not affiliated with, authorized, or endorsed by YouTube or Google LLC. It is an independent tool that retrieves publicly available data via a third-party API. Use it in compliance with YouTube's Terms of Service and all applicable laws; you are responsible for how you use the retrieved data.

What it does

• Caption discovery — for each video, lists all caption tracks YouTube exposes (e.g. English, Spanish, auto-generated English), with the language code, a human-readable name, the kind (asr = auto-generated), isTranslatable, and the transcriptUrl (a YouTube timedtext URL).
• Video metadata — every item also carries the parent video's videoTitle, channel, channelId, lengthSeconds, viewCount and shortDescription.
• Filtering — keep only certain languages, or only auto-generated tracks.
• Transcript text (best effort) — optionally tries to download and flatten the caption file into plain text. See the note below.

Input

Field	Type	Default	Description
videoIds	string[]	["dQw4w9WgXcQ"]	Video IDs or full watch / youtu.be / shorts URLs.
languageCodes	string[]	[]	Keep only tracks whose language code matches (e.g. en, es). Empty = all.
autoGeneratedOnly	boolean	false	Keep only ASR (auto-generated) tracks.
fetchTranscriptText	boolean	false	Attempt to download the transcript text (best effort, see note).
maxItems	integer	50	Max total caption tracks across all videos.

Example input

{
  "videoIds": ["dQw4w9WgXcQ", "https://www.youtube.com/watch?v=jNQXAC9IVRw"],
  "languageCodes": ["en"],
  "autoGeneratedOnly": false,
  "fetchTranscriptText": true,
  "maxItems": 100
}

Output

One dataset item per caption track:

{
  "videoId": "dQw4w9WgXcQ",
  "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "videoTitle": "Rick Astley - Never Gonna Give You Up (Official Video) (4K Remaster)",
  "channel": "Rick Astley",
  "channelId": "UCuAXFkgsw1L7xaCfnd5JJOw",
  "lengthSeconds": 213,
  "viewCount": 1779355962,
  "languageCode": "en",
  "language": "English",
  "kind": "asr",
  "isAutoGenerated": true,
  "isTranslatable": true,
  "vssId": ".en",
  "transcriptUrl": "https://www.youtube.com/api/timedtext?v=dQw4w9WgXcQ&...",
  "source": "video:dQw4w9WgXcQ"
}

Notes

• Transcript text is best-effort. YouTube signs each timedtext URL against the IP that requested it, so a server-side download frequently returns an error. When fetchTranscriptText is enabled the Actor still tries, but transcriptText may come back empty. The transcriptUrl is always provided so you can fetch the caption file yourself (append &fmt=json3, &fmt=srv3, or &fmt=vtt) from the appropriate IP.
• Data is sourced live; YouTube / the upstream edge occasionally rate-limits, so the Actor retries transient blocks with exponential backoff.
• Video IDs are de-duplicated within a run.