YouTube Transcript Scraper — Bulk, With Timestamps

Extract transcripts (subtitles/captions) from YouTube videos in bulk — manual or auto-generated, in any language, with timestamped segments and video metadata. Clean JSON built for AI/LLM pipelines, summarisation, research, SEO and content repurposing.

You only pay for transcripts actually delivered. Videos with no captions, or unavailable videos, are reported in a failure list — never billed as results.

What it does

• Accepts any mix of YouTube URLs (watch, youtu.be, Shorts, embed) or bare video IDs
• Language preference with sensible fallbacks (manual captions → auto-generated → any language, all configurable)
• Timestamped segments (start, duration, text) plus the joined plain-text transcript
• Video metadata: title, channel, duration, view count
• Failure transparency: a FAILED_VIDEOS record explains every undelivered video (no captions / unavailable / blocked)

Output example

{
  "videoId": "dQw4w9WgXcQ",
  "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "title": "Rick Astley - Never Gonna Give You Up (Official Video)",
  "channelName": "Rick Astley",
  "durationSeconds": 213,
  "language": "en",
  "isAutoGenerated": false,
  "wordCount": 488,
  "transcript": "We're no strangers to love You know the rules and so do I...",
  "segments": [
    { "startSeconds": 18.64, "durationSeconds": 3.96, "text": "We're no strangers to love" }
  ]
}

Typical uses

• AI pipelines — feed transcripts into LLM summarisation, RAG indexes, or agents
• Content repurposing — turn videos into articles, newsletters, show notes
• Research — analyse messaging across a channel or topic at scale
• SEO — mine competitor video content for topics and phrasing
• Accessibility & archiving

Notes

• Works with public videos that have captions (manual or auto-generated)
• For very large batches, the automatic proxy setting handles rotation; switch to residential proxies if you see botwall entries in the failure list
• Only publicly available caption data is collected