YouTube Transcript Scraper

Pull the full transcript and timestamped captions from any YouTube video — pick your language, get clean text plus segment-level timing, and export to JSON or CSV.

What you get

• Full plain-text transcript of each video as a single field, ready for search, summarisation, or LLM input
• Timestamped segments array with startMs, durationMs, and text per caption line — perfect for jumping to a moment in the video or building chapter markers
• Language metadata showing the language code returned and whether captions were human-written or auto-generated
• Available languages list so you can see every caption track the video offers, even when you only kept one
• Per-video metadata: videoId, title, durationSeconds, uploader
• Multi-language preference order — list ["en", "es", "fr"] and the actor picks the first one that exists for each video
• Residential proxy by default so the request reliably returns captions; switch to Auto cascade to save proxy bandwidth

Use cases

• Content repurposing — turn long videos into blog posts, newsletters, or tweet threads with timestamped quotes
• SEO research — analyse competitor video scripts and topics for keyword and topic gaps
• Accessibility — generate searchable text archives of training videos, lectures, and webinars
• Language learning — pull dual-language transcripts for the same video to build study material
• Podcast and interview notes — feed transcripts into summarisers to produce show notes and chapter markers

How to use

1. Paste one or more YouTube URLs into YouTube Video URLs (youtube.com/watch?v=..., youtu.be/..., or youtube.com/shorts/...)
2. Set Languages in preference order — e.g. en for English-only, or en, es, fr to fall back through Spanish then French
3. Leave Prefer Human-Written Captions on to favour manual subtitles over auto-generated ones when both exist
4. Leave Proxy Mode on Residential (default) — switch to Auto if you want to try cheaper paths first
5. Run the actor — one transcript record per video appears in the Dataset tab

Output format

Each dataset record:

{
  "inputUrl": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "videoId": "jNQXAC9IVRw",
  "title": "Me at the zoo",
  "durationSeconds": 19,
  "uploader": "jawed",
  "language": "en",
  "isAutoGenerated": false,
  "transcript": "All right, so here we are in front of the elephants...",
  "segments": [
    { "startMs": 0, "durationMs": 2400, "text": "All right, so here we are" },
    { "startMs": 2400, "durationMs": 3100, "text": "in front of the elephants" }
  ],
  "availableLanguages": ["en", "en-US", "es", "fr"],
  "proxyUsed": "residential",
  "status": "success",
  "error": null
}

When no caption track matches the requested languages, status is "error" and error is "No captions available in requested languages" — availableLanguages still lists everything the video does have so you can retry with the right code.

Notes & limits

• Up to 10 videos per run. Schedule multiple runs for larger jobs.
• Captions must exist on the video. Some music videos and shorts have no captions at all — those return status: "error".
• Auto-generated captions are word-level transcribed by YouTube and may include noise or incorrect punctuation. The isAutoGenerated flag tells you which kind you got.
• Region-locked or private videos will fail; try Residential proxy mode if Auto can't fetch them.
• Language codes are BCP-47, e.g. en, en-US, es, fr, pt-BR. If you specify en and only en-US exists, the actor will use en-US automatically.
• Only scrape content you have the right to use. Respect copyright and YouTube's Terms of Service.