youtube-search-tool-using-subtitles 0.1.0

Search and query YouTube videos using AI embeddings

YouTube Search Tool

Search and query YouTube videos using AI embeddings. Add individual videos or entire channels, then search through their content using natural language.

Features

• 🎥 Process individual YouTube videos or entire channels
• 🔍 Semantic search through video transcripts
• 🤖 Multiple AI providers (OpenAI, Google, Local models)
• 🌐 Web interface with Gradio
• ⚡ Fast batch processing with concurrent downloads

Installation

pip install youtube-search-tool

Quick Start

1. Set up environment

Create a .env file:

# Choose your AI provider
USE_EMBEDDING_PROVIDER=openai  # or 'google' or 'local'
USE_LLM_PROVIDER=openai

# API Keys (if using cloud providers)
OPENAI_API_KEY=your_openai_key
GEMINI_API_KEY=your_google_key

# Optional: Customize processing
MAX_WORKERS=4
MAX_CHANNEL_VIDEOS=100

2. Add videos

# Add a single video
yt-search add "https://www.youtube.com/watch?v=QPM0WNqwJBc"

# Add an entire channel
yt-search add "https://www.youtube.com/@Lamediainglesa"

3. Search

# Search through your videos
yt-search search "machine learning basics"
yt-search search "python tutorial"

4. Web Interface

# Launch the web interface
yt-search web

Requirements

• Python 3.8+
• Chrome/Chromium browser (for channel scraping)
• API keys for cloud providers (optional, can use local models)

Supported AI Providers

• OpenAI: text-embedding-ada-002 (requires API key)
• Google: text-embedding-004 (requires API key)
• Local: all-MiniLM-L6-v2 (no API key needed)

License

MIT License