YouTube Transcript Scraper Plus

Extract transcripts from YouTube videos together with real video metadata, in a clean, flat, one-row-per-video format. Built on the youtube-transcript-api library for transcripts and YouTube's own watch-page data (InnerTube JSON + oEmbed) for metadata.

Accepts single videos, bulk lists, Shorts / embed / live links, bare 11-character video IDs, and channel or playlist URLs (expanded to their videos).

What it extracts

Each dataset row is one video and contains:

Transcript

• plainText — the full transcript as clean text
• segments[] — per-segment start, duration, end, timestamp (hh:mm:ss), text, and a watch?t= deep-link url
• srt — ready-to-use SRT subtitle string
• vtt — ready-to-use WebVTT subtitle string
• segmentCount, wordCount, characterCount, readingTimeMinutes

Language / caption fields (authoritative, from the library)

• languageCode, language
• isGenerated — true for auto-generated captions (from the library's is_generated, not a display-name guess)
• isTranslatable, translationLanguages[], translationLanguageCount
• availableLanguages[] — every caption track on the video (language, languageCode, isGenerated, isTranslatable)
• languageCount
• translatedTo — set when you request translation via translateTo

Video metadata

• title, channel, channelId, channelUrl
• viewCount, likeCount (best-effort; null when YouTube does not expose it)
• publishDate, uploadDate, durationSeconds
• description, category, keywords[], thumbnail, isLiveContent

Bookkeeping

• status — one of ok, no_captions, unplayable, error (never faked — a video with no captions is reported honestly)
• scrapedAt (ISO 8601 UTC), error (message when status is not ok)

Optional AI enrichment (off by default)

When aiEnhancement is on and an API key is provided, each row also gets:

• aiSummary, aiKeyPoints[], aiTopics[], aiChapters[]

Providers are auto-detected from the model name: Claude (Anthropic), GPT/o-series (OpenAI), Gemini (Google), Grok (xAI), DeepSeek, Sonar (Perplexity), Mistral.

Input

Field	Type	Default	Description
startUrls	array	—	Video URLs / IDs / Shorts / embed / live links / channel / playlist URLs
preferredLanguages	array	["en"]	ISO codes in priority order; first match wins, falls back to best track
transcriptType	enum	any	any / manualOnly / autoOnly
translateTo	string	""	ISO code to translate into (only if the track is translatable)
includeSegments	boolean	true	Include the per-segment array
includeSrt	boolean	true	Include SRT output
includeVtt	boolean	true	Include WebVTT output
maxVideosPerChannel	integer	10	Cap when expanding a channel/playlist
concurrency	integer	5	Videos processed in parallel
aiEnhancement	boolean	false	Turn on AI summary/key points/topics/chapters
aiModel	enum	claude-haiku-4-5	Model/provider for AI enrichment
aiApiKey	secret	—	API key for the AI provider (or set the provider env var)
proxyConfiguration	proxy	Residential	Residential proxy recommended — YouTube blocks datacenter IPs

Example input

{
  "startUrls": ["https://www.youtube.com/watch?v=4KbrxIpQgkM"],
  "preferredLanguages": ["en"],
  "transcriptType": "any",
  "includeSrt": true,
  "includeVtt": true,
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Example output (abridged, real fields)

{
  "videoId": "4KbrxIpQgkM",
  "url": "https://www.youtube.com/watch?v=4KbrxIpQgkM",
  "status": "ok",
  "title": "...",
  "channel": "...",
  "channelId": "UC...",
  "channelUrl": "https://www.youtube.com/channel/UC...",
  "viewCount": 123456,
  "likeCount": 4321,
  "publishDate": "2022-01-01",
  "durationSeconds": 615,
  "languageCode": "en",
  "language": "English",
  "isGenerated": false,
  "isTranslatable": true,
  "availableLanguages": [
    { "language": "English", "languageCode": "en", "isGenerated": false, "isTranslatable": true }
  ],
  "segmentCount": 210,
  "wordCount": 1500,
  "readingTimeMinutes": 7.5,
  "plainText": "Full transcript text...",
  "segments": [
    { "start": 0.0, "duration": 3.2, "end": 3.2, "timestamp": "00:00:00", "text": "Hello", "url": "https://www.youtube.com/watch?v=4KbrxIpQgkM&t=0s" }
  ],
  "srt": "1\n00:00:00,000 --> 00:00:03,200\nHello\n...",
  "vtt": "WEBVTT\n\n00:00:00.000 --> 00:00:03.200\nHello\n...",
  "scrapedAt": "2026-07-02T00:00:00Z"
}

When a video has no captions, the row is still emitted with status: "no_captions" and the transcript fields left null — no fabricated text. Only genuine result rows (status: "ok") are charged.

Notes

• Residential proxy is strongly recommended; YouTube blocks datacenter IPs for transcript requests.
• likeCount and some metadata can be null when YouTube does not surface them — these are reported honestly rather than faked.
• Legality: this retrieves captions and metadata already publicly served by YouTube. Review YouTube's Terms of Service for your use case; republishing content may require the creator's permission.