mp-web3

mp-web3 / jarvis-v3

Public

Fully local voice interface for Claude Code on Apple Silicon. Parakeet STT + Kokoro TTS + SmartTurn EOU + dual VAD.

apple-silicon local-llm mlx speech-to-text text-to-speech
17
0
100% credibility
Found Mar 19, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Jarvis v3 provides a fully local voice interface for interacting with command-line AI coding tools like Claude Code using speech-to-text and text-to-speech on Apple Silicon Macs.

How It Works

1
🔍 Discover Jarvis

You stumble upon Jarvis, a fun voice helper that lets you talk to your coding AI instead of typing, perfect for Apple Mac users.

2
💻 Get it ready on your Mac

Download the helper and install it quickly so everything is set up on your computer.

3
🎤 Pick your microphone and speakers

Choose which mic to speak into and speakers to hear replies from, like your headphones or built-in ones.

4
🪟 Open your AI chat window

Start a simple chat session in a special screen where your coding AI is running and ready.

5
🚀 Launch and start talking

Turn on Jarvis with one command, speak naturally, and watch it type your words and read back smart replies.

6
🗣️ Have hands-free conversations

Ask coding questions or give instructions by voice, feeling like chatting with a helpful friend.

🎉 Voice magic for coding

Now you code faster and easier, talking to your AI assistant anytime without touching the keyboard.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is jarvis-v3?

Jarvis-v3 turns your Apple Silicon Mac into a fully local voice assistant for CLI tools like Claude Code. Speak into the mic and it transcribes speech offline, polishes the text, injects it into a tmux pane, then reads back the response via speakers—all without cloud APIs. Built in Python with MLX for Metal acceleration, it offers CLI commands like `jarvis start` for terminal mode or `jarvis web` for browser-based echo cancellation.

Why is it gaining traction?

In a sea of cloud-dependent voice UIs, jarvis-v3 stands out as a fully local AI solution with smart end-of-utterance detection that handles natural pauses, barge-in to interrupt responses, and transcript polishing to cut filler words. Developers dig the tmux integration for any CLI workflow, plus dual VAD for noise filtering—making hands-free local AI feel responsive without latency or privacy leaks.

Who should use this?

Apple Silicon devs running local LLMs in tmux, like Claude Code users tired of typing prompts. Ideal for fully local voice assistant setups in homelab projects, from fully local security camera controls to fully remote jobs scripting CLI agents. Skip if you're not on Metal or need multi-speaker support.

Verdict

Promising early experiment (17 stars, 1.0% credibility score) with clear docs and easy uv setup, but lacks tests and echo cancellation without headphones. Grab it for holomat v3 jarvis-style fully local voice if you're prototyping offline AI—no better quickstart for tmux voice I/O on Mac.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.