Explainroo Lets AI Coding Agents Create Local Explainer Videos

Summary

Explainroo is a free, open-source tool that lets coding agents create narrated explainer videos and product demos from a topic, product, or website. The agent writes a Markdown script and a JavaScript scene file, while Explainroo renders the visuals and synchronizes them with speech. Its local pipeline uses the Kokoro open voice model for narration, Whisper through Transformers.js for word timing, Chrome for frame rendering, Rough.js and Lucide for drawings and icons, and ffmpeg for the final MP4. It also creates background music and sound effects, and provides still frames, contact sheets, layout checks, and speech checks so an agent can inspect its output. Videos can use five visual styles, several aspect ratios for platforms such as YouTube, TikTok, Reels, Instagram, and LinkedIn, word-by-word captions in selected formats, and adjustable playback pace. For product demos, an agent can reconstruct screens from application code or a website and simulate pointer clicks and text entry. The project requires Node.js 20.11 or newer, ffmpeg, and Chrome or Chromium; its initial voice and timing downloads total about 400 MB and do not require a GPU, account, or API key. Linux is the primary tested platform, while macOS and Windows have received less testing. Optional illustrations can be generated through OpenRouter at an additional per-image cost, and generated videos include an Explainroo watermark by default.