voice dictation for linux

Talk. It types.

Voxtty is a private, local-first voice keyboard for Linux. Press a key, speak, and your words land in whatever app you're in — transcribed on your own machine.

Ubuntu / Wayland · press Alt+D to dictate

Wispr Flow has no Linux build

Voxtty is the one it can't be: native to Linux, private by default, and free to run — your audio never has to touch a server.

How it works

Three steps, zero friction

No window to focus, no button to find. Voxtty runs in your tray and drops text wherever your cursor already is.

Trigger

Press Alt+D from anywhere. Voxtty starts listening in the background.

Speak

Talk naturally. Whisper transcribes on-device — no upload, no round trip, works offline.

It types

Cleaned-up text is typed straight into the focused app — Slack, your editor, the browser, anything.

What's inside

Local by default. Smart when you want it.

Everything you need runs on your machine. The cloud is strictly opt-in.

// on-device

Local transcription

Whisper runs on your CPU. Your voice never leaves the machine.

// background

Runs in your tray

Starts on login and waits quietly. One keypress to dictate, one to stop.

// anywhere

Types into any app

If you can type there, you can dictate there. No integrations required.

// offline

Works with no internet

On a plane, off the grid — dictation keeps working. No account needed.

// optional

AI cleanup

Opt in to send text (never audio) to Claude for punctuation and polish.

// yours

Custom dictionary

Teach it your names, jargon, and spellings. Applied instantly, offline.

Privacy, not as a setting — as the default

Your voice stays on your machine

Transcription is 100% local. Nothing is uploaded, logged, or sent to a server. The only time text leaves your computer is if you turn on the optional AI cleanup — and even then it's the transcript, never the audio. Turn it off and Voxtty is fully offline, forever.

Install

Up and running in a minute

Clone it, run the setup script, log back in. Voxtty registers as a background service with a tray icon and a hotkey.

bashUbuntu / Wayland
$ git clone https://github.com/PominausGH/voxtty.git
$ cd voxtty
$ ./setup.sh        # installs deps + the voxtty service

First run downloads the Whisper model. Then press Alt+D and start talking.

Open-core

Free forever. Pro when you want more.

The core app is open source and does real work on its own. Pro adds the polish and the cross-platform builds.

Free · open source
$0

Everything you need to dictate on Linux.

  • Local Whisper transcription
  • Global hotkey (Alt+D)
  • Types into any app
  • Offline rule-based cleanup
  • Custom word dictionary
Pro
$5 /mo · BYO key

or $50/year — 2 months free. The smart features and the polish.

  • AI cleanup — cloud & local LLM
  • Per-app tone & formatting
  • Auto-learning dictionary
  • GUI settings + auto-update
  • Windows & macOS builds

vs Wispr Flow

The one Wispr Flow can't ship

Wispr Flow is genuinely good — on Mac and Windows, in the cloud. On Linux, and when privacy isn't negotiable, that's where Voxtty lives.

VoxttyWispr Flow
LinuxNativeNot available
TranscriptionOn your machineCloud
Works offlineYesNo
Audio leaves your deviceNeverAlways
PriceFree · Pro $5/mo (BYO key)$15 / month
Open sourceYesNo
AI cleanupOptionalBuilt-in

Questions

Good to know

Does my voice go to the cloud?

No. Transcription runs entirely on your machine. The only thing that can leave is text — and only if you opt into AI cleanup. Leave it off and Voxtty is fully offline.

Does it work without internet?

Yes. Everything except the optional AI cleanup runs locally, so dictation works on a plane or off the grid.

Which apps can I dictate into?

Any of them. If you can type in it, Voxtty can type in it — editors, browsers, chat, terminals.

Is it really free?

The core app is free and open source. Pro (AI cleanup, per-app formatting, a GUI, cross-platform builds) is $5/month or $50/year. AI cleanup runs on your own API key, so your usage stays yours and costs pennies.

Windows or macOS?

Linux first. Native Windows and macOS builds are on the roadmap for Pro.