Introducing Voice Mirror
By Nathan
Introducing Voice Mirror
We’re excited to announce the alpha release of Voice Mirror — a voice-controlled AI agent overlay that lives on your desktop.
The Problem We’re Solving
AI coding assistants are powerful, but they’re trapped inside text boxes. You have to switch windows, copy-paste context, and type everything. What if your AI could just hear you, see your screen, and act on your computer?
Our Approach: Eyes + Ears + Voice
Voice Mirror is a transparent desktop overlay that combines:
- Voice Control — Wake word activation, push-to-talk, or continuous conversation
- App Preview — Build an app by voice and watch it render live; the AI sees and drives the running app
- Screen Awareness — Captures your display and understands what you’re looking at
- Terminal Power — Spawns Claude Code internally for real command execution
- Browser Automation — Full Chrome control via CDP
- Persistent Memory — Three-tier memory system across sessions
The result? A voice-native IDE where your AI builds, sees, and drives what you make.
What’s Included
Voice Mirror ships with 45 MCP tools across 5 groups, support for 75+ AI models via Claude Code, OpenCode, and Ollama, and a lightweight voice stack that runs entirely on CPU. It’s Windows-first during alpha, with macOS and Linux on the way.
Alpha Access
Voice Mirror is free and open source. We’re looking for early adopters who want to push voice-controlled AI to the next level.