Vox2
a tiny two-way translator for your desktop
Type, paste, speak or snip text in any app and read it, or hear it, in another language. Two boxes, live translation both ways, and a few global shortcuts. A 2 MB download, about 5 MB installed.
What it does
- Live, both ways. Type in either box and the other translates as you go. Words show grey until the engine settles, then go dark.
- From any app. Select text anywhere and press a shortcut; the translation appears in a small bubble by your cursor, with a match score. Drag it anywhere, and it stays put like a sticky note.
- Snip & translate. Draw a box around text on screen (images, video subtitles, apps that block copying) and read it in your language.
- Dictation. Tap or hold a key and talk. Speech is transcribed on your machine by OpenAI's open Whisper models (up to Large v3 Turbo, the most accurate), and can be typed straight into the app you're using.
- Read aloud. Natural male and female voices in about 75 languages. Each word lights up as it is spoken; slow, normal or fast.
- Conversation mode. Two people take turns talking; each phrase is translated and spoken to the other person.
- Engines. Google Translate (free), or Claude, ChatGPT or Gemini with your own API key, with tone and "who it's for" so pronouns come out right.
- Back-translation with a match score. See your translation turned back into your language, scored by how much of your meaning survived. It compares meaning, not exact words, on your computer. How it works, and how well.
- In your language. Every menu, button and message in any of 133 languages, following your computer's language to start. Right-to-left languages flip the layout. Machine-translated, then checked by translating each string back; Vietnamese read in full by a person.
- Small things. Detect language, pinned languages, pronunciation for non-Latin scripts, 30-item history, always on top, 37 themes.
- Windows and Mac. The same app on both. On Mac the shortcuts follow Mac conventions, and a permissions screen walks you through what macOS asks for.
Shortcuts
| Windows | Mac | |
| Ctrl Shift Space | ⌃ ⌥ V | show or hide Vox2, ready to type |
| Ctrl Alt T | tap Right ⌘ | translate the selected text in any app |
| Ctrl Alt S | ⌃ ⌥ S | snip an area of the screen and translate it |
| Right Ctrl | Right ⌥ Option | dictate: tap to start (stops when you go quiet), or hold to talk |
| Esc | Esc | stop reading aloud or dictating, from any app |
| Ctrl P | ⌘ P | pin on top (while Vox2 is in front) |
All of them can be changed in settings.
Install on Windows
- Download
Vox2-setup.exeand run it. It installs for your user only; no admin needed. - Windows will say "unknown publisher" because the installer isn't code-signed. Choose More info → Run anyway.
- Vox2 updates itself from GitHub releases. You can switch that to "ask me" or off in settings.
Windows 10/11, 64-bit.
Install on Mac (beta)
- Download
Vox2-mac.dmg, open it and drag Vox2 into Applications. - The first time you open it, macOS says it can't verify the developer (Vox2 isn't notarized by Apple yet). Go to System Settings → Privacy & Security and click Open Anyway. Once is enough.
- A permissions screen lists what macOS needs you to allow: Input Monitoring (shortcuts from other apps), Accessibility (typing dictation, copying a selection), Screen Recording (snip) and Microphone (dictation). Click allow on each; they tick off as you go.
- Vox2 updates itself from GitHub releases and keeps its permissions across updates.
Apple Silicon Macs (M1 and newer). More about Vox2 on Mac.
How it works
- Shell: Tauri 2. A small Rust binary hosts the UI in the system web view (WebView2 on Windows, WebKit on Mac), so there's no bundled browser.
- UI: plain HTML, CSS and JavaScript. No framework, no build step.
- Native side (Rust): a global keyboard hook for the shortcuts (
rdevon Windows, a Core Graphics event tap on Mac), simulated input (enigo) to copy selections and type dictation into other apps, screen capture (xcap), the tray, and signed self-updates. - Speech to text: OpenAI's Whisper model running locally through transformers.js and ONNX Runtime, on the GPU via WebGPU when available. The model downloads once (41–250 MB) and is released from memory when idle.
- Screen text: Apple's built-in text recognizer on Mac; Tesseract.js on Windows, also local, with language data fetched on first use.
- Voices: Microsoft's neural voices (the ones Edge's Read Aloud uses), or OpenAI's, or the system's.
- Translation: Google Translate's public web endpoint, or the Anthropic, OpenAI and Gemini APIs, called directly from your machine.
Privacy
- No account, no server of ours, no analytics.
- Text goes only to the translation engine you pick. Your voice and screen snips are processed on your computer.
- Settings and history are stored locally in your user folder. API keys are locked in Windows Credential Manager (the Keychain on Mac), not in a settings file.
- The free Google Translate and Microsoft voice services are unofficial endpoints and could stop working; the API engines are the official route.
Build from source
git clone https://github.com/chrisqtruong/vox2 cd vox2 cargo install tauri-cli --version "^2" --locked cargo tauri build
Needs Rust and, on Windows, the Visual Studio C++ build tools. The UI lives in resources/, the native code in src-tauri/src/.