vincentsch/explainroo

Explainer videos and product demos made by your AI agent. Free and open source: a local voice (Kokoro), word timing (Whisper) and a canvas renderer turn a script into a narrated MP4.

What it solves

It automates the creation of short explainer videos and product demos. Instead of manually animating and editing, a coding agent can generate the script, visual scenes, and final MP4 file entirely on a local machine.

How it works

An AI agent writes a script (script.md) and a JavaScript file (scenes.js) to define the drawings and timing. The system then processes these files using several local tools:

  • Voice: Uses the Kokoro open voice model for text-to-speech.
  • Timing: Uses OpenAI's Whisper to align spoken words with visual elements.
  • Pictures: Uses a headless Chrome browser to render frames, utilizing Rough.js for hand-drawn styles and Lucide for icons.
  • Sound: Generates custom background music and sound effects.
  • File: Combines all elements into an MP4 using ffmpeg.

To ensure quality, the tool provides the agent with scene stills, layout checks for overlapping text, and speech checks for pronunciation errors.

Who it’s for

Developers and content creators who use coding agents (like Claude Code or Codex) to quickly produce educational videos or software product demonstrations without needing professional video editing software.

Highlights

  • Local Execution: Voice, timing, and rendering run on the user's computer for privacy and zero cost.
  • Agent-Driven: Designed specifically to be operated by AI coding agents.
  • Multiple Aesthetics: Offers five visual styles (paper, clean, chalk, blueprint, and midnight).
  • Flexible Formatting: Supports various aspect ratios for YouTube, TikTok, Reels, Instagram, and LinkedIn.
  • Product Demos: Can rebuild app screens from code or websites to create simulated user interactions.

Related

  • Project
    UllrAI/CineGen-ShortDrama

    CineGen AI Director is a browser‑based production tool that turns a story outline into AI‑generated manga‑style videos. It uses Google Gemini for script and image creation and Veo for short video clips, letting users define start/end keyframes for each shot to retain visual control.

  • Project
    aleximmer/Laplace

    Laplace is a PyTorch library that turns a trained neural network into a Bayesian model using the Laplace approximation, offering posterior estimation, marginal‑likelihood computation, and uncertainty‑aware predictions.

  • Project
    win4r/openclaw-a2a-gateway

    OpenClaw A2A Gateway – a Node.js plugin that implements Google’s Agent‑to‑Agent v0.3.0 protocol for OpenClaw agents. Provides zero‑config installation, automatic peer discovery (DNS‑SD/mDNS), multi‑transport fallback (JSON‑RPC, REST, gRPC), bio‑inspired routing, circuit‑breaker resilience, bearer‑token auth, and file‑transfer tooling.

  • Project
    lemma-work/lemma-platform

    Lemma Platform is an open‑source framework that lets teams create AI‑powered “teammates” (pods) with shared tables, markdown memory, agents, and workflows. The teammate can be accessed from Slack, Teams, WhatsApp, Telegram, email, or a custom web app, and runs on the LLM subscription you already have (Claude Code, Codex, etc.). It provides fine‑grained permissions, low‑code pages/apps, and self‑improving harnesses that turn corrections into reusable workflows. Install via Lemma Cloud or locally, then scaffold a pod with `lemma pod create …`. The project includes Python and TypeScript SDKs and several ready‑made pod templates.