ai-video-pipeline

End-to-end AI video pipeline: TTS voice synthesis, FFmpeg video assembly, automated YouTube upload via OAuth 2.0 & Data API.

Scripts

Prerequisites

• Python 3.9+
• FFmpeg installed and on PATH: ffmpeg -version
• Google Cloud project with YouTube Data API v3 enabled
• OAuth 2.0 client credentials (client_secrets.json) — stored outside repo
• OpenAI API key (for script generation) or Gemini API key

Setup

pip install -r requirements.txt
cp .env.example .env   # fill in OPENAI_API_KEY, YOUTUBE_CLIENT_ID, YOUTUBE_CLIENT_SECRET

Place client_secrets.json in the project root (it is in .gitignore).

Usage

# Generate and upload a single video
python auto_video_bot.py --topic "Top 5 Python tips 2025"

# Batch generate from a topics file
python batch_video_generator.py --topics topics.txt

# Generate a YouTube Short
python shorts_video_bot.py --topic "Quick Python tip"

# Add captions to existing clips
python process_clips_with_captions.py --input clips/ --output output/

First run will open a browser for YouTube OAuth authentication. Token is cached for future runs.

Notes

• YouTube Data API quota: 10,000 units/day on free tier. Each video upload costs ~1,600 units.
• FFmpeg must support the codec used by your source clips (H.264 recommended).

Built with

Python · FFmpeg · OpenAI API · YouTube Data API v3 · OAuth 2.0
AI-assisted development (Claude, GitHub Copilot) — architecture, requirements, QA validation and debugging by me.