Talk instead of type, in any app.
Tap a key, speak, and tap it again. Your words appear wherever you were typing.
Voice snippets, dictation and screen context · Illustrated app examples with the native Vox overlay
- Tap right Shift and start talking.
- Tap it again. Vox Transfer types what you said into the app you're in.
- Changed your mind? Click × in the recording overlay to cancel. Double-tap cancellation is available as an opt-in setting.
Shift still types capitals: only a tap on its own starts dictation. It works in email, chat, documents, code editors and terminals.
Vox Transfer runs in the background, with a microphone icon in the menu bar (macOS) or the system tray (Linux). Everything else is in its menu and its Settings window, from choosing another hotkey to searching what you've dictated.
On macOS or Linux/X11, a small recording overlay shows the microphone waveform and transcription status near the bottom of the target display. It is on by default and follows Reduce Motion. Turn it off in Settings → General → Show recording overlay. Overlay settings and platform coverage
On macOS or Linux, run:
curl -fsSL https://github.com/runsonmypc/vox/releases/latest/download/install.sh | bash- macOS needs uv first:
brew install uv. When macOS asks, allow Microphone, Accessibility and Input Monitoring, then quit Vox Transfer from its menu and open it again from Applications. Screen Recording is optional: it lets Vox Transfer read names on your screen so it spells them right. - Linux needs Ubuntu 24.04 or similar, in an X11 session: at the login screen, choose
Ubuntu on Xorg. The installer asks for your password to add a few system packages. There's
also a
.debon the releases page.
Vox Transfer then asks for an OpenAI API key, which you can create at platform.openai.com/api-keys (OpenAI bills API use separately from ChatGPT). Rather not? It can also transcribe on your own computer, for free.
Run the same command again to update. More install options
Using Claude Code, Codex, Cursor or another AI coding assistant? Paste this in:
Install Vox Transfer for me by following https://github.com/runsonmypc/vox/blob/main/docs/ai-setup.md
Then ask me what I'd like to customize (hotkey, vocabulary, snippets, language, offline
transcription) and set it up.
- Works wherever you type. It pastes into the app you're in, then puts your clipboard back.
- OpenAI or fully offline. Transcribe with OpenAI, or locally with whisper.cpp. Switch in Settings.
- Spells your words right. Add names and jargon to your vocabulary.
- Snippets. Say "dev server" on its own to insert
npm run dev -- --host 127.0.0.1 --port 4321. Save your own phrases and expansions in Settings; commands are pasted, not executed. - Searchable history of everything you've dictated, kept on your computer. Failed recordings can be retried locally or with OpenAI Batch.
- Private. No analytics and no telemetry. What leaves your computer
|
|
|
Voice snippets on macOS Full size · Dark |
Searchable history on Linux Full size · Dark |
Teach Vox Transfer your names and jargon · Full size (dark)
- Install: other ways to install, permissions, updating and uninstalling
- Using Vox Transfer: the hotkey, the menu, Settings and long recordings
- Settings: every option, sounds, and offline transcription
- Privacy: what is sent and what stays on your computer
- Troubleshooting
- Development
Copyright (C) 2026 Alex. Vox Transfer is free software: you can redistribute it and change it under the terms of the GNU General Public License, version 3. It comes with no warranty; see LICENSE for the full terms.