🎙️ Podcast Transcription & Chat App

Transform YouTube podcasts into interactive conversations using local Whisper transcription and GPT-4o chat.

Features

• 📺 Download audio from YouTube URLs
• 🎯 Local Whisper transcription (no API costs)
• 💬 Interactive chat about podcast content
• 📄 Export transcripts (TXT, SRT, JSON)
• 💾 Export chat conversations (Markdown)
• 🔒 Privacy-focused (local processing)

Setup

1. Clone Repository

git clone <repository-url>
cd social_media_vlog_chat

2. Create Virtual Environment

Windows:

python -m venv .venv
.venv\Scripts\activate

Mac/Linux:

python -m venv .venv
source .venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

Note for Python 3.13 users: If you encounter audioop module errors, also install:

pip install audioop-lts

4. Set OpenAI API Key

Windows:

set OPENAI_API_KEY=your_api_key_here

Mac/Linux:

export OPENAI_API_KEY=your_api_key_here

5. Install FFmpeg

Windows:

• Download from https://ffmpeg.org/download.html
• Add to PATH

Mac:

brew install ffmpeg

Linux:

sudo apt install ffmpeg

Usage

Quick Start

python run_app.py

This starts both the backend and frontend automatically.

Manual Start

1. Backend: python backend.py
2. Frontend: streamlit run frontend.py

Using the App

1. Transcribe: Enter YouTube URL → Download → Transcribe
2. Chat: Ask questions about the podcast content
3. Export: Download transcripts and chat conversations

File Structure

podcast-transcription-app/
├── backend.py          # FastAPI server
├── frontend.py         # Streamlit interface
├── config.py          # Configuration settings
├── run_app.py         # Launcher script
├── requirements.txt   # Dependencies
├── downloads/         # Audio files
├── transcripts/       # JSON transcripts
└── exports/          # Exported files

System Requirements

• RAM: 8GB+ (16GB recommended for medium model)
• Storage: 2GB+ free space
• Python: 3.8+
• Internet: For downloading videos and API calls

Costs

• Transcription: Free (local Whisper)
• Chat: ~$0.03 per 1000 tokens (GPT-4o)
• Storage: Local files only

Troubleshooting

Backend Won't Start

• Check Python version (3.8+)
• Install missing dependencies: pip install -r requirements.txt
• Verify FFmpeg installation: ffmpeg -version

Whisper Model Loading Slow

• First run downloads ~1.5GB model
• Subsequent runs are faster
• Model stored in ~/.cache/whisper/

OpenAI API Errors

• Verify API key is set correctly
• Check API quota/billing
• Ensure stable internet connection

Model Information

Whisper Medium:

• Size: ~1.5GB
• Languages: 99 languages
• Accuracy: Good for most podcasts
• Speed: ~1:4 ratio (1 min audio = 4 min processing)

Privacy

• Audio files stored locally in downloads/
• Transcripts stored locally in transcripts/
• Only chat messages sent to OpenAI API
• No data uploaded to external services (except OpenAI chat)