youtube-transcript

youtube-transcript

Extract transcripts from YouTube videos. Zero dependencies - just Node.js.

When to Use

• "Get the transcript from this YouTube video"
• "Extract transcript from this video"
• "What does this YouTube video say?"

Usage

# Extract plain text transcript
node get-transcript.js "https://www.youtube.com/watch?v=VIDEO_ID"

# With timestamps
node get-transcript.js "https://youtu.be/VIDEO_ID" --timestamps

# JSON output (includes offset/duration for each segment)
node get-transcript.js "https://youtube.com/watch?v=VIDEO_ID" --json

# Save to file
node get-transcript.js "https://youtu.be/VIDEO_ID" --save transcript.txt

# Specific language
node get-transcript.js "https://youtu.be/VIDEO_ID" --lang es

Supported URL Formats

• https://www.youtube.com/watch?v=VIDEO_ID
• https://youtu.be/VIDEO_ID
• https://www.youtube.com/watch?v=VIDEO_ID&t=123
• https://youtube.com/shorts/VIDEO_ID
• Just the video ID: dQw4w9WgXcQ

Requirements

• Node.js 18+

No npm install needed. No dependencies.

How It Works

1. Tries YouTube's Innertube API first (Android client - faster)
2. Falls back to web page scraping if needed
3. Extracts caption track URL from response
4. Fetches XML captions
5. Parses segments with timestamps
6. Returns clean text

API Keys

None required. Uses YouTube's public APIs.

Output Formats

Flag	Output
(default)	Plain text, all segments joined
--timestamps	[0:00] First line\n[0:05] Second line...
--json	[{text, offset, duration, lang}, ...]

Limitations

• Only works for videos with captions (manual or auto-generated)
• Private or age-restricted videos will fail
• YouTube may require CAPTCHA if too many requests from same IP