YouTube Summarizer for Obsidian

Automatically fetch YouTube video transcripts, generate AI-powered summaries, and create organized notes in your Obsidian vault.

Features

✅ Transcript Extraction - Automatically fetches transcripts from YouTube videos
✅ AI Summaries - Generates comprehensive summaries using Claude
✅ Key Points - Extracts 5-7 main takeaways from each video
✅ Smart Tagging - Auto-generates relevant tags for categorization
✅ Obsidian Integration - Creates beautifully formatted Markdown notes
✅ Dashboard Generation - Automatically organizes videos by tags and channels
✅ Batch Processing - Handle multiple videos from Watch Later playlists
✅ Channel Organization - Groups videos by YouTube channel

Installation

1. Clone or Download

git clone <repository-url>
cd youtube-summarizer

2. Install Dependencies

pip install -r requirements.txt

Required packages:

• youtube-transcript-api - Fetches video transcripts
• anthropic - Generates AI summaries with Claude

3. Set Up API Key

Get your Anthropic API key from https://console.anthropic.com/

Option A: Environment Variable (Recommended)

export ANTHROPIC_API_KEY="your-api-key-here"

Option B: .env File Create a .env file:

ANTHROPIC_API_KEY=your-api-key-here

4. Configure Vault Path

Edit youtube_summarizer_enhanced.py or create config.py:

VAULT_PATH = "/Users/stanley/ObsidianVault"  # Update this

Usage

Command Line Interface (Recommended)

Process a Single Video

python cli.py -u "https://www.youtube.com/watch?v=VIDEO_ID"

Process Multiple Videos from File

Create a text file watch_later.txt:

https://www.youtube.com/watch?v=VIDEO_ID_1
https://www.youtube.com/watch?v=VIDEO_ID_2
https://www.youtube.com/watch?v=VIDEO_ID_3

Then run:

python cli.py -f watch_later.txt

Interactive Mode

python cli.py -i

Enter URLs one at a time, type done when finished.

Generate Dashboard Only

python cli.py --dashboard

Python API

from youtube_summarizer_enhanced import process_youtube_url

# Process a single video
url = "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
note_path = process_youtube_url(url)

# Process multiple videos
from youtube_summarizer_enhanced import process_watch_later_playlist

urls = [
    "https://www.youtube.com/watch?v=VIDEO_1",
    "https://www.youtube.com/watch?v=VIDEO_2",
]
process_watch_later_playlist(urls)

Generated Note Structure

Each video note includes:

# Video Title

## Metadata
- **Video ID:** abc123
- **URL:** https://youtube.com/watch?v=abc123
- **Channel:** [[Channel Name]]
- **Created:** 2024-12-23
- **Tags:** #youtube #security #malware

## Summary
AI-generated comprehensive summary (2-3 paragraphs)

## Key Points
- Main takeaway 1
- Main takeaway 2
- Main takeaway 3
...

## Transcript

[00:15] First line of transcript [00:32] Second line of transcript ...

Dashboard Organization

The script automatically generates dashboards:

Weekly Dashboard

• Groups videos by tags
• Groups videos by channels
• Shows chronological list

Monthly Dashboard (Coming Soon)

• Links to all weekly dashboards
• Monthly statistics

Directory Structure

ObsidianVault/
├── YouTube Notes/
│   ├── Video Title 1.md
│   ├── Video Title 2.md
│   └── ...
├── Channels/
│   └── (Future: channel-specific pages)
└── Dashboards/
    ├── Week_2024-12-23.md
    ├── Week_2024-12-16.md
    └── ...

How It Works

1. Extract Video ID - Parses YouTube URL to get video ID
2. Fetch Transcript - Uses YouTube Transcript API to get captions
3. Generate Summary - Sends transcript to Claude for analysis
4. Extract Metadata - Parses video title, channel, tags
5. Create Note - Generates formatted Markdown in Obsidian vault
6. Update Dashboard - Organizes videos by category

Configuration Options

Edit config_template.py and save as config.py:

# Vault Settings
VAULT_PATH = "~/ObsidianVault"
VIDEOS_FOLDER = "YouTube Notes"

# Summary Settings
SUMMARY_MODEL = "claude-sonnet-4-20250514"
MAX_TRANSCRIPT_LENGTH = 8000

# Transcript Settings
PREFERRED_LANGUAGES = ['en', 'en-US', 'en-GB']

# Dashboard Settings
WEEKLY_LOOKBACK_DAYS = 7
DEFAULT_TAG = "#youtube"

Troubleshooting

No Transcript Available

Some videos don't have transcripts (disabled by creator or auto-captions unavailable). The script will create a basic note with metadata only.

API Rate Limits

If processing many videos, you may hit Anthropic API rate limits. The script handles this gracefully and continues with remaining videos.

File Permission Errors

Ensure your Obsidian vault path is correct and you have write permissions.

Module Not Found

Make sure you've installed all dependencies:

pip install -r requirements.txt

Advanced Features

Custom Note Templates

Modify _build_note_content() in ObsidianNoteCreator class to customize note format.

Tag Customization

The AI automatically generates relevant tags, but you can add your own in the DEFAULT_TAG setting.

Batch Processing

Process large playlists efficiently:

# In youtube_summarizer_enhanced.py
BATCH_SIZE = 10  # Process 10 videos at a time
PAUSE_BETWEEN_BATCHES = 5  # 5 second pause between batches

Getting YouTube Watch Later URLs

Method 1: Manual Export

1. Go to https://www.youtube.com/playlist?list=WL
2. Copy each video URL
3. Paste into a text file

Method 2: Browser Extension

Use a browser extension like "Copy All URLs" to export all URLs at once.

Method 3: YouTube Data API (Advanced)

Use Google's YouTube Data API to programmatically fetch your Watch Later playlist.

Examples

Process Security Conference Talks

# Create a file with DefCon talks
cat > defcon_talks.txt << EOF
https://www.youtube.com/watch?v=TALK_1
https://www.youtube.com/watch?v=TALK_2
https://www.youtube.com/watch?v=TALK_3
EOF

# Process them
python cli.py -f defcon_talks.txt

Daily Research Workflow

# 1. Export today's Watch Later videos to file
# 2. Process them
python cli.py -f todays_videos.txt

# 3. Generate dashboard
python cli.py --dashboard

# 4. Open Obsidian and review your notes!

Contributing

Improvements welcome! Some ideas:

• YouTube Data API integration for automatic Watch Later sync
• Support for playlists
• Video thumbnail download
• Speaker identification in transcripts
• Custom summary prompts
• Export to other note-taking apps

License

MIT License - feel free to use and modify!

Credits

• Built with youtube-transcript-api
• Powered by Anthropic Claude
• Designed for Obsidian

Support

For issues or questions:

1. Check the Troubleshooting section
2. Review the examples
3. Open an issue on GitHub

---

Happy note-taking! 📝