YouTube Transcript Extractor - AI Training Data

Extract YouTube video transcripts optimized for AI and machine learning workflows.

---

Why This Actor?

Feature	Free Tools	This Actor
AI chunking	❌	✅ Split by token limit
Token counting	❌	✅ Estimated tokens
Clean transcripts	❌	✅ Remove ♪ [music]
SRT/VTT formats	❌	✅ All included
Video metadata	❌	✅ Title, author, thumbnail
Affordable	Limited	$0.01 per video

---

Use Cases

AI Training Data

Build high-quality training datasets from YouTube videos. Chunked transcripts fit any LLM context window.

Content Analysis

Analyze video content. Get word counts, token estimates, and structured metadata.

Subtitle Generation

Export transcripts in SRT or VTT format for video editing, captions, or accessibility.

Academic Research

Extract lectures, interviews, and documentaries. Clean transcripts ready for analysis.

---

Features

🎯 AI-Optimized Output

• Smart chunking - Split transcripts to fit your LLM's context window
• Token estimation - Know exactly how many tokens you're working with
• Clean mode - Remove music symbols (♪), [applause], [laughter] for cleaner training data

📄 Multiple Formats

• Plain text - Raw transcript
• SRT subtitles - For video editors
• VTT subtitles - Web-compatible
• Timestamps - Optional [MM:SS] markers

📊 Metadata Enrichment

• Video title and author
• Thumbnail URL
• Duration (formatted and raw)
• Word count and character count
• Detected language

🔒 Reliability

• Automatic proxy fallback (Direct → Datacenter → Residential)
• YouTube Shorts support
• Multi-language transcripts

---

Pricing

$0.01 per transcript extraction

Videos	Cost
10	$0.10
100	$1.00
1,000	$10.00

No monthly commitment. Pay only for what you use.

---

Quick Start

Input

{
  "videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "language": "en",
  "chunkSize": 2000,
  "cleanTranscript": true,
  "outputFormat": "text"
}

Output

{
  "videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "videoId": "dQw4w9WgXcQ",
  "transcript": "♪ We're no strangers to love ♪",
  "transcriptClean": "We're no strangers to love",
  "chunks": [
    {
      "id": 0,
      "text": "We're no strangers to love...",
      "start": 1.36,
      "end": 110.0,
      "wordCount": 230
    }
  ],
  "metadata": {
    "title": "Rick Astley - Never Gonna Give You Up",
    "author": "Rick Astley",
    "thumbnailUrl": "https://img.youtube.com/vi/dQw4w9WgXcQ/maxresdefault.jpg",
    "duration": 211.32,
    "durationFormatted": "03:31",
    "wordCount": 367,
    "estimatedTokens": 488,
    "language": "en"
  },
  "transcriptSRT": "1\n00:00:01,360 --> 00:00:03,040\n♪ We're no strangers to love ♪",
  "transcriptVTT": "WEBVTT\n\n00:00:01.360 --> 00:00:03.040\n♪ We're no strangers to love ♪"
}

---

Input Parameters

Parameter	Type	Default	Description
videoUrl	string	required	YouTube video URL
language	string	"en"	Preferred transcript language
chunkSize	integer	2000	Max chars per chunk (0 = off)
cleanTranscript	boolean	false	Remove music symbols and filler
includeMetadata	boolean	true	Include video metadata
outputFormat	string	"text"	Format: text, srt, or vtt
includeTimestamps	boolean	true	Add [MM:SS] timestamps

---

Supported URLs

• https://www.youtube.com/watch?v=VIDEO_ID
• https://youtu.be/VIDEO_ID
• https://www.youtube.com/shorts/VIDEO_ID

---

FAQ

Q: What if a video has no transcript? A: The actor will return an error for that video.

Q: Can I extract transcripts in other languages? A: Yes. Set language to the ISO code (e.g., "es" for Spanish).

Q: What's the maximum chunk size? A: Default is 2000 characters (~500 tokens). Set to 0 to disable chunking.

Q: How accurate is token estimation? A: We use ~1.33 tokens per word as a rough estimate.

---

Support

Open an issue on GitHub or contact for enterprise pricing on large volumes.

---

$0.01 per transcript | Try it now on Apify