Control Claude Code by voice while you work hands-free

A developer built a local spoken-command setup around Claude Code and used it to fix a WhatsApp integration without touching the keyboard.

Done with Claude Code· Advanced· Feb 2026

What they did
They put together a voice pipeline on an M4 Pro over the course of an afternoon. Parakeet v3 TDT converted speech into text, Pocket TTS spoke Claude’s replies, and a custom MCP connection linked the pieces to Claude Code. To make it usable, they rigged AirPods stem clicks so a tap could stop the current response and start recording again, then mirrored the session to a TV with AirPlay.
How it went
That setup let them work through a custom WhatsApp integration while brushing their teeth. They called the experience fun and easy on the back. The build involved some hacking to capture the AirPods click, and performance depended on local model speed on the M4 Pro.
Tips
Expect some hardware tweaking if you want a headset gesture to act like push-to-talk. It is worth testing the speech loop on a small debugging task before depending on it.

Try it yourself with Claude Code

Help me build a spoken control setup for Claude Code on my [computer type] machine. I want [speech-to-text tool] to turn my voice into text, [text-to-speech tool] to read Claude’s replies out loud, and an [MCP bridge/custom connector] to join them together. Guide me through the setup in small steps, and after each stage ask me to confirm before moving on. If a step installs software, changes audio or device settings, starts a debugging session, sends data, deletes something, or submits anything, show it clearly and wait for my approval first. When everything works, I should be able to speak a task, hear the response, and use the setup to troubleshoot [specific project].

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.