YouTube Transcript Scraper for AI & RAG

Turn any YouTube video into clean, structured text ready for LLM context windows, RAG pipelines, or full-text search. Extracts captions directly from YouTube's internal player data — no browser, no API key, no rate-limit headaches. Process hundreds of videos in parallel with configurable concurrency.

Features

• Full transcript extraction — pulls the complete caption track from any video with available subtitles
• Timed segments — returns individual segments with offset and duration for precise referencing
• Bulk processing — extract transcripts from hundreds of videos in a single run
• Parallel execution — configurable concurrency to balance speed and reliability
• AI-ready output — concatenated plain text field perfect for LLM ingestion
• No API key required — works directly with YouTube's public caption data

Input

Field	Type	Required	Default	Description
urls	array	Yes	—	YouTube video URLs to extract transcripts from
concurrency	integer	No	10	Number of videos to process in parallel. Increase for large batches, decrease if you see errors.

Input Example

{
  "urls": [
    "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "https://www.youtube.com/watch?v=jNQXAC9IVRw"
  ],
  "concurrency": 10
}

Output

Each video produces one dataset row containing the full transcript as a single text block, plus an array of timed segments for granular access.

Key output fields:

Field	Type	Description
inputUrl	string	The video URL you provided
videoId	string	YouTube video ID
fullText	string	Complete transcript as a single text block
segments	array	Individual caption segments with timing data
segments[].text	string	Text content of the segment
segments[].offset	number	Start time in milliseconds
segments[].duration	number	Segment duration in milliseconds

Output Example

{
  "inputUrl": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "videoId": "jNQXAC9IVRw",
  "fullText": "Welcome to the presentation. Today we will cover three key topics in machine learning. First, let us look at supervised learning and how it differs from unsupervised approaches...",
  "segments": [
    { "text": "Welcome to the presentation.", "offset": 0, "duration": 2400 },
    { "text": "Today we will cover three key topics", "offset": 2400, "duration": 3100 },
    { "text": "in machine learning.", "offset": 5500, "duration": 1800 }
  ]
}

Pricing

Pay per event on Apify platform.

Event	Description
Transcript	One video transcript extracted

Use Cases

• RAG pipelines — feed video transcripts into vector databases for retrieval-augmented generation
• Content repurposing — convert video content into blog posts, summaries, or social media threads
• Research analysis — make lecture and conference talk content searchable and quotable
• SEO audits — analyze what competitors say in their videos without watching them
• Training data — build domain-specific datasets from educational YouTube content

Related Actors

Actor	What it adds
YouTube Transcript MCP Server	AI-powered transcription via MCP — works even on videos without captions
YouTube Video Metadata Extractor	Get views, likes, tags, and publish dates alongside your transcripts
Markdown Extractor	Convert web pages to clean markdown — complement video transcripts with written sources