Skip to content

README.md

Latest commit

 

History

History
98 lines (71 loc) · 3.03 KB

README.md

File metadata and controls

98 lines (71 loc) · 3.03 KB

YouTube Transcript Analyzer

YouTube Transcript Analyzer Logo

A modern web application that analyzes YouTube playlist content using AI. Extract transcripts, generate summaries, and get AI-powered insights to quickly determine if videos are worth watching.

Features

  • 🎬 Playlist Processing: Analyze entire YouTube playlists with one click
  • 🤖 AI Analysis: Generate summaries, key takeaways, and topic classifications
  • 📝 Content Evaluation: Get "worth watching" verdicts with justifications
  • 📊 Structured Insights: Understand video organization and key points
  • 🏷️ Smart Categorization: Reliable category tagging with JSON-based AI responses
  • 📱 Responsive Design: Modern interface that works on all devices

Screenshots

Home Page Video Detail

Tech Stack

Frontend: React, TypeScript, Vite, React Router, Axios, React Markdown
Backend: FastAPI, SQLAlchemy, SQLite, YouTube API, Gemini AI

Quick Setup

Prerequisites

  • Node.js 22+ and npm
  • Python 3.13+ with uv
  • Google Cloud Project with YouTube Data API
  • Gemini API key

Backend Setup

cd backend
uv sync
# Create client_secrets.json and .env files (see backend README)
python authenticate_youtube.py
uvicorn app.main:app --reload

Frontend Setup

cd frontend
npm install
npm run dev

Access the app at http://localhost:5173

Project Structure

.
├── frontend/             # React frontend
├── backend/              # FastAPI backend
│   ├── app/              # Application code
│   ├── authenticate_youtube.py  # YouTube auth script
│   └── token.pickle      # Generated auth token
└── screenshots/          # Application screenshots

Key Features Explained

AI-Powered Analysis

The application uses Google's Gemini AI to analyze video transcripts and generate structured insights:

  • JSON-Based Responses: AI outputs are structured as JSON for reliable parsing and consistent results
  • Smart Category Extraction: Multiple fallback systems ensure videos always have relevant categories
  • Markdown Support: All AI-generated content is rendered with full Markdown formatting

Real-Time Processing

  • Processing status updates in real-time
  • Videos are automatically refreshed when analysis completes
  • Progress indicators show exactly what's happening

Documentation

For detailed setup instructions and development guides:

Troubleshooting

  • YouTube Authentication Issues: Delete token.pickle and run python authenticate_youtube.py again
  • CORS Errors: Ensure both frontend and backend servers are running
  • Missing Analysis: Check your Gemini API key and quota limits

License

MIT