Yes. MIT-licensed, no paid plans, no subscriptions, no watermarks. Optional third-party services (Pexels, Pixabay) may have their own terms and rate limits.
No. Openverse provides 600M+ CC-licensed images with no API key. For video stock, a free Pexels API key is recommended but not required.
Yes. It uses Voicebox in-repo TTS (Kokoro/Chatterbox) for realistic local voice synthesis with Edge-TTS as a free fallback, deterministic media selection, and ffmpeg/Remotion rendering. Optional AI visual verification uses Ollama or Gemini Vision.
No watermark is added by this project.
Yes. Place files in input/visuals/ and reference them with [Visual: filename.mp4] or [Visual: filename.jpg] in your script.
Yes. Use portrait orientation for 9:16 vertical video.
Yes. Run it locally, in Docker, or deploy behind your own infrastructure.
The project bundles ffmpeg-static and ffprobe-static, so most users don’t need a separate FFmpeg install. A global FFmpeg is only needed as a fallback on some machines.
The desktop app tries multiple voice paths: bundled Edge-TTS first, then Windows offline speech fallback.
Python 3.8 or later.
Node.js 18 or later.
Yes. Use the CLI (npm run generate) or the MCP server for headless operation.
Yes. The web portal exposes HTTP endpoints at http://localhost:3001/.