🎙️ Podcast Chat

AI-Powered Podcast Conversations — Download any YouTube podcast, transcribe it with AI, and have intelligent conversations about the content.

---

🎨 Screenshots

Dark Mode	Light Mode

✨ Features

Feature	Description
YouTube Download	Paste any YouTube URL with live video preview
One-Click Transcribe	Single "Transcribe Video" button downloads and transcribes automatically
Progress Tracking	Beautiful animated progress bar showing download and transcription status
Intelligent Chat	Ask questions and get contextual answers from podcast content
20+ AI Features	Summary, flashcards, Q&A, notes, insights, and more — all powered by Ollama
History Management	Save and revisit your transcribed podcasts with History tab
Dark/Light Mode	Toggle between dark and light themes with persistent preference
User Authentication	Secure email-based login with per-user history
Multi-language	English, Spanish, French, German, Hindi, Portuguese, Italian
Smart Chunking	Dynamic audio chunking based on video length for optimal processing

---

🚀 Quick Start

Prerequisites

• macOS 10.15+, Windows 10/11, or Linux (Ubuntu 20.04+)
• Python 3.10+
• FFmpeg (auto-installed on macOS, manual on Windows/Linux)
• Ollama (auto-installed on macOS, manual on Windows/Linux)

Installation

# Clone the repository
git clone https://github.com/ShivashishY/Podcast_Chat.git
cd podcast-chat

# Create virtual environment
python3 -m venv .venv
source .venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Configure environment
cp .env.example .env
# Edit .env and add your SMALLEST_API_KEY

Get Your API Key

1. Go to smallest.ai and create an account
2. Navigate to API Keys section
3. Create a new key and add it to your .env file:

   SMALLEST_API_KEY=sk_your_api_key_here

Run the App

python app.py

Open http://localhost:5000 in your browser.

One-Click Start

Platform	Launcher	How to Run
macOS	Start Podcast Chat.command	Double-click the file
Windows	Start Podcast Chat.bat	Double-click the file
Linux	start-linux.sh	Run ./start-linux.sh in terminal

The launchers automatically:

• Install dependencies if missing
• Create virtual environment
• Download Ollama model (first run)
• Start the application

---

💡 How to Use

1️⃣ Paste & Preview

Paste a YouTube URL — a live video preview appears instantly

2️⃣ Transcribe

Click Transcribe Video — the progress bar shows download (0-50%) and transcription (50-100%)

3️⃣ Chat & AI Features

Ask any question about the podcast content, or use one of the 20+ AI features!

Example Questions:

• "What was the main topic discussed?"
• "Summarize this podcast in 3 bullet points"
• "What did the guest say about [topic]?"
• "When did they talk about [subject]?"

4️⃣ AI Features

After transcription, access the AI Features panel with collapsible categories:

Category	Features
Popular	Summary, Key Insights, Clean Transcript, Proper Notes
Basic Content	Micro Summary, Short Summary, Bullet Points, Notable Quotes
Analysis	Extract Ideas, Extract Insights, Extract Patterns, Extract Wisdom
Study & Education	Flashcards, Concept Map, Q&A, Outline Notes, Cornell Notes, Rapid Logging, T-Note Method, Charting Method, QEC Method, Q&A Split Page

---

⚙️ Configuration

Environment Variables

Variable	Required	Default	Description
SMALLEST_API_KEY	✅ Yes	-	API key from smallest.ai
FLASK_DEBUG	No	false	Enable debug mode
PORT	No	5000	Server port
OLLAMA_URL	No	localhost:11434	Ollama server URL
OLLAMA_MODEL	No	llama3.2	LLM model to use

Smart Audio Processing

The app handles any length video by:

• Compressing audio to mono 16kHz for efficient processing
• Splitting into 60-second chunks to stay under API limits
• Seamlessly stitching transcripts with accurate timestamps

---

🏗️ Project Structure

podcast_chat/
├── app.py                    # Main Flask application
├── auth.py                   # Authentication module
├── models.py                 # Database models
├── templates/                # HTML templates
│   ├── index.html            # Main app interface
│   ├── login.html            # Login page
│   └── signup.html           # Signup page
├── downloads/                # Downloaded audio (gitignored)
├── transcripts/              # Saved transcripts (gitignored)
├── docs/                     # Documentation
├── requirements.txt          # Python dependencies
├── .env.example              # Environment template
├── setup.sh                  # Setup script (macOS/Linux)
├── start.sh                  # Start script (macOS/Linux)
├── start-linux.sh            # One-click launcher (Linux)
├── Start Podcast Chat.command # One-click launcher (macOS)
└── Start Podcast Chat.bat    # One-click launcher (Windows)

---

🎨 Design

Modern two-column layout with dark/light mode support:

Element	Dark Mode	Light Mode
Primary Accent	#2B9D90 (Teal)	#2B9D90 (Teal)
Background	#0d0d0d	#f5f5f5
Cards	#161616	#ffffff
Text	#e4e4e7	#1f1f1f

UI Highlights:

• Dark/Light mode toggle with persistent preference
• YouTube video preview in left column
• Animated progress bar with step indicators
• Collapsible AI feature categories
• History tab for past transcriptions, which can be reused without processing again
• Chat + AI features in right column

---

🔒 Security Features

• ✅ Password hashing with Werkzeug
• ✅ Session management via Flask-Login
• ✅ Auto-generated secret key (no manual configuration)
• ✅ Per-user data isolation
• ✅ CSRF protection
• ✅ SQLite database for user storage

---

❓ Troubleshooting

"Cannot connect to Ollama"

# Start Ollama manually
ollama serve

"AI Model Not Found"

# Pull the required model
ollama pull llama3.2

"Transcription Failed"

• Verify your Smallest AI API key in .env
• Check your internet connection
• Long podcasts (>1hr) may take several minutes

"App Won't Start"

# Re-run setup
./setup.sh

# Or manually install dependencies
pip install -r requirements.txt

---

💾 System Requirements

Requirement	macOS	Windows	Linux
OS Version	10.15+	10/11	Ubuntu 20.04+
Python	3.10+	3.10+	3.10+
RAM	4GB (8GB recommended)	4GB (8GB recommended)	4GB (8GB recommended)
Storage	5GB	5GB	5GB
FFmpeg	Auto-installed	Manual install	apt install ffmpeg
Ollama	Auto-installed	Manual install	curl -fsSL https://ollama.ai/install.sh | sh

---

🤝 Contributing

Contributions are welcome! Please read our Contributing Guide for details on:

• Setting up the development environment
• Code style guidelines
• Submitting pull requests

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🙏 Powered By

• Smallest AI — Pulse STT for transcription (displayed as "Powered by Smallest AI" badge)
• Ollama — Local AI inference for chat & 20+ AI features
• yt-dlp — YouTube downloads
• Flask — Web framework

---

Made with ❤️ for podcast lovers

Report Bug · Request Feature