GitHub - T0mSIlver/localvoxtral: Talk to your coding agents by voice. Realtime, fully local macOS dictation that streams words as you speak and grounds LLM polishing in the exact Claude Code session under your cursor — Ghostty, iTerm2, Terminal.app, even a herdr pane. 100% on-device on Apple Silicon.

3 min read Original article ↗

localvoxtral app icon

Talk to your coding agents. Keep every word on your Mac.
Realtime, fully local dictation for the menu bar. Press a key and speak — your words appear while you're still talking.

Install · Documentation · Coding agents · Contributing

GitHub stars   Latest release   License

demo.mp4

Unlike tools that transcribe after you stop speaking, localvoxtral streams text as the audio arrives, powered by Mistral AI's Voxtral Mini 4B Realtime running on your own Apple Silicon. It is built first for prompting coding agents by voice, and it stays a solid general dictation app everywhere else. Everything runs on-device — no account, no subscription, nothing leaving your Mac.

Install

curl -fsSL https://raw.githubusercontent.com/T0mSIlver/localvoxtral/main/scripts/install.sh | bash

Or download the latest .dmg from Releases. Requires an Apple Silicon Mac on macOS 15+. A first-launch wizard handles permissions and the engine download; if Gatekeeper complains about a hand-installed DMG, see the install guide.

Features

  • Built for coding agents — dictate prompts straight into any CLI agent (opencode gets its own plugin), in any terminal — Warp, WezTerm, kitty, Alacritty, and more; polishing understands developer speech: "dash dash force" → --force, "use auth dot t s" → useAuth.ts (details)
  • Claude Code aware — dictation joins the exact session under your cursor — Ghostty, iTerm2, Terminal.app, a single herdr or cmux pane, over SSH, or a claude.ai/code Remote Control tab in your browser — and grounds polishing in its screen, your last prompt, the files Claude just touched, and that repo's vocabulary (details)
  • One key, two modes — tap for a reviewable overlay with optional LLM polishing, hold to stream words live into the focused app (shortcuts)
  • Private by default — audio, transcription, and polishing are local processes; no telemetry, no account, no cloud fallback (how it works)
  • Menu bar native — instant popover with dictation status at a glance, microphone picker, auto-copy of the final text, and the raw transcript one click away after a polished commit
  • Bring your own server — dictation and polishing can each point at any OpenAI-compatible endpoint instead of the built-in local engines
  • Multilingual — dictate in English, French, or any language Voxtral understands; polishing answers in the language you spoke (English and French are covered by the test suite)

Tip

If localvoxtral is useful to you, a ⭐ on this repo helps others find it.

Documentation

License

MIT