jev-voice-control
macOS menu-bar Swift app that turns on-device transcribed speech into Jev typed decisions and macOS actions, confirming hard-to-undo steps.
Overview
Jev Voice (chris-wozniczek) is a macOS menu-bar app that turns spoken commands into actions using TypeSafe's Jev. Hold Option-Space, speak and release; Jev decides what to do for commands like open chrome and go to google.com, quit spotify, type hello world, set volume to 30 or maximize chrome. App names support common aliases, and the app can speak short replies and optionally ask for voice confirmation before running commands.
Mixed commands combine deterministic local actions with Jev-guided UI work, for example opening an app locally and then letting Jev guide a click, keeping typed decisions in order and asking for confirmation before actions that sound hard to undo. Speech is transcribed on-device (SpeechAnalyzer streaming on macOS 26, Whisper or the classic recogniser), and only the transcript plus minimal app context such as the frontmost app and installed app names is sent to Jev. Typing verifies the focused field and falls back to a Unicode keyboard path if needed, and custom vocabulary improves recognition. Version 0.9.1 is a public reliability release with a documentation site.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which action does this spoken command require?(inferred) | — |
| Choice | Which UI element should be clicked for the Jev-guided part of the task?(inferred) | — |
How it works
- Transcribe speech on-device
- Send transcript and minimal app context to Jev
- Execute local actions and Jev-guided UI steps in order
- Confirm hard-to-undo actions; verify typed text
Strengths
- On-device transcription
- Confirmation for risky actions
- Typing verification with fallback
- Public release with docs site
Caveats
- macOS only
- Transcript and app names are sent to Jev
At a glance
- Type
- App
- Status
- beta
- Language
- Swift
- Runtime
- macOS menu-bar app
- Has code
- yes
- Stars
- 6
Use cases and integrations
How curated lists describe it
Control your Mac by voice. Speech → Jev (TypeSafe AI System One model) typed decisions → macOS actions. Menu-bar Swift app.
wh000wh000/awesome-jev-liveControl your Mac by voice. Speech → Jev (TypeSafe AI System One model) typed decisions → macOS actions. Menu-bar Swift app.
hellogumbo/awesome-jev
Frequently asked questions
What is sent to Jev?
The transcript and a small amount of app context such as the frontmost app and installed app names.
Does it confirm risky commands?
Yes, it asks before actions that sound hard to undo, and optionally before every command.
Which speech engines are supported?
Apple SpeechAnalyzer streaming on macOS 26, Whisper and the classic Apple recogniser.
Related projects
Voice assistant daemon for Hyprland Linux desktops that acts through tool modules, with Jev gating which requests and actions proceed.
ProjectbetaPythonTalk to your Mac: local whisper.cpp transcribes, one Jev request picks the action and typed arguments, and code executes macOS automation.
ProjectprototypePythonFloating macOS bar for voice or text commands where Jev picks the next action from live accessibility controls in a loop.
ProjectprototypeSwiftmacOS voice and typed computer-use app where Jev picks the next on-screen action from the Accessibility tree, with no screenshots.
AppprototypeSwiftPre-alpha Rust and Tauri voice control layer for macOS that reads the accessibility tree; Jev is a fallback rather than the router.
ProjectprototypeRustElectron app for controlling Windows 10/11 by Turkish or English speech, using OpenAI Realtime, Jev decisions, UI Automation and Playwright.
ProjectprototypeTypeScript