Voice Dictation on Linux: Best Offline Apps & Complete Setup Guide

Why voice dictation on Linux was historically hard

Until recently, Linux users had few real options for voice dictation:

• Google Chrome's Web Speech API — browser-only, requires internet, no system-wide injection
• Julius / CMU Sphinx — open-source but outdated accuracy, complex setup, no continuous improvement
• Kaldi-based tools — research-grade accuracy but no polished user interface
• Cloud services (Google, AWS) — excellent accuracy but audio leaves your device

The commercial leaders (Superwhisper, Dragon, Wispr Flow) simply didn't support Linux. None of the desktop environments filled the gap either — GNOME and KDE offer accessibility and input features, but nothing provides the "hold a key, speak, text appears anywhere" experience that macOS and Windows users take for granted.

How Whisper AI changed Linux voice dictation

OpenAI's open-source Whisper model changed the equation. Released in 2022 and continuously improved through Whisper Large V3, Whisper delivers:

• Near-human accuracy — competitive with cloud services even for accented speech
• 100+ language support — multilingual transcription from a single model
• CPU-runnable — no GPU required, works on any modern Linux machine
• Fully open-source — can be run completely offline, self-hosted, no licensing fees

The result: a wave of Linux-compatible voice dictation tools built on top of Whisper, from polished paid apps to developer-focused scripts. For a deeper look at how the engines themselves work, read our offline speech recognition explainer.

Full comparison: best Linux voice dictation tools in 2026

AirTypes — best polished offline voice dictation for Linux

AirTypes is the only actively-developed, paid, offline voice dictation app that currently ships on Linux. It fills the gap that Superwhisper and Wispr Flow leave — a polished, maintained tool you don't have to configure manually. It ships as a native package with Whisper built in, no Python environment required, and supports system-wide text injection on both X11 and Wayland.

What you get

• Global hotkey (Ctrl+Shift+Space by default) — hold to record, release to transcribe and inject
• 6 Whisper model tiers — from Swift (~40 MB, <1 second) to Deep (Large V3, ~1.6 GB, maximum accuracy)
• Filler word removal — strips um, uh, like, you know automatically
• Writing tones — Formal, Professional, Neutral, Casual, Friendly
• 100+ language support via Whisper's multilingual models
• My Agent — voice to BYOK AI, response typed at cursor (GPT-4, Claude, Ollama, any OpenAI-compatible)
• Transcription history — searchable log of all past dictations

System requirements (Linux)

• Ubuntu 20.04+, Fedora 36+, Arch Linux, Debian 11+, or any distribution with glibc 2.31+
• 4 GB RAM minimum (8 GB recommended for Large models)
• x86_64 architecture — no GPU required, runs entirely on CPU
• Microphone: any PulseAudio or PipeWire-compatible audio device
• Display server: X11 (full support) or Wayland (clipboard injection mode)

The free options, honestly assessed

Vocalinux — best free Linux voice dictation

Vocalinux is an open-source, offline voice dictation tool built on Whisper for Linux. It handles system-wide text injection and works in most desktop environments (GNOME, KDE, XFCE). No cost, no subscription.

Best for: users comfortable with a manual setup who don't need filler word removal, AI routing, or ongoing updates. Good "it works" solution once configured.

Limitations: no Windows or macOS. Manual model management. No tone adjustment. Community-supported only.

OpenWhispr — best for self-hosting / cross-platform

OpenWhispr is a cross-platform open-source wrapper around Whisper that runs on Linux, macOS, and Windows. It handles model management and injection. More polished than raw whisper.cpp but still requires terminal setup.

Best for: developers who want cross-platform offline transcription with full control and no subscription.

Nerd Dictation — lightweight Vosk option

Nerd Dictation is free, open source, and scriptable, using Vosk for offline recognition. It's lightweight and good for simple dictation on X11.

Limitations: command-line setup, lower accuracy than Whisper Large, and clunky on Wayland.

Whisper, the DIY way

You can run Whisper (or whisper.cpp / faster-whisper) directly and script transcription yourself. You get state-of-the-art accuracy for free — but you maintain the glue: capturing the mic, triggering on a hotkey, and injecting text into the focused window.

Setup on Ubuntu / Debian

AirTypes ships a .deb package for Ubuntu 20.04+ and Debian-based distributions.

1. Download the .deb file from the download page
2. Install with:

   sudo dpkg -i airtypes_*.deb
   sudo apt-get install -f  # resolve any dependencies

3. Launch AirTypes from your application menu or run airtypes from terminal
4. Grant microphone permission when prompted
5. Open the AirTypes settings and download your preferred Whisper model

AirTypes will appear in your system tray. Hold Ctrl+Shift+Space in any application to start recording.

For Vocalinux, follow the installation guide in its project repository — it uses pip and requires PortAudio.

Setup on Fedora

AirTypes ships an .rpm package for Fedora 36+ and Red Hat-based distributions.

1. Download the .rpm file from the download page
2. Install with:

   sudo rpm -i airtypes_*.rpm
   # or
   sudo dnf install ./airtypes_*.rpm

3. Launch AirTypes and configure your microphone and Whisper model

Note for Fedora users: if you use PipeWire (default on Fedora 34+), AirTypes will automatically detect it. No additional configuration is required.

Setup on Arch Linux

AirTypes ships an AppImage that works on Arch and other rolling-release distributions.

1. Download the .AppImage from the download page
2. Make it executable:

   chmod +x AirTypes_*.AppImage

3. Run it:

   ./AirTypes_*.AppImage

4. To install system-wide, move it to /usr/local/bin/ or add a desktop entry

AUR package coming soon for streamlined Arch installation. For Vocalinux or OpenWhispr, AUR packages may be available — check the respective project repositories.

Wayland vs. Xorg: what to know

AirTypes supports both display servers, but with different text injection mechanisms:

Xorg (X11)

Full support. AirTypes uses keyboard-level injection via xdotool-style input, which types your transcribed text character by character at the cursor. Works in all applications including terminal emulators.

Wayland

Wayland's security model restricts direct keyboard injection from external applications. AirTypes handles this by automatically switching to clipboard mode: your transcribed text is copied to the clipboard and pasted via Ctrl+V. This works in virtually all Wayland applications (GNOME, KDE Plasma 6, Sway).

To set your injection mode, go to Settings → Behavior → Injection Mode → "Direct Paste" (Clipboard + Ctrl+V). Ubuntu defaults to Wayland on GNOME, so this matters for most Ubuntu users.

Tips for best results on Linux

• Use PipeWire — modern Linux distributions default to PipeWire for audio, which offers lower latency than PulseAudio. AirTypes detects and uses it automatically.
• Test your microphone first — run arecord -l to verify your mic is recognized, and use pavucontrol or the GNOME sound settings to check input levels.
• GNOME 45+ users — you may need to grant microphone permission via Settings → Privacy → Microphone if the app doesn't prompt you.
• Custom hotkey — if Ctrl+Shift+Space conflicts with an existing shortcut in your desktop environment, open AirTypes Settings → Hotkey to change it.
• Startup on login — add AirTypes to your startup applications via your desktop environment's settings, or add it to your ~/.config/autostart/ directory.

Final recommendation

The right Linux voice dictation tool depends on your priorities:

• Want install-and-go, polished UI, AI features: AirTypes ($3.99/mo, 7-day free trial)
• Want free and fully offline, OK with manual setup: Vocalinux
• Developer, want cross-platform + self-hosted: OpenWhispr
• Want maximum DIY control: whisper.cpp + xdotool script

For most Linux users who want something that "just works" — AirTypes is the only polished, maintained, offline option currently available.

FAQ

Does Ubuntu have built-in speech to text?

Only limited support via accessibility and input frameworks — nothing like the system-wide dictation macOS and Windows ship. For reliable voice typing on Ubuntu, use a dedicated tool: Nerd Dictation, a Whisper script, or a packaged app like AirTypes.

What is the best free speech-to-text for Linux?

Vocalinux, Nerd Dictation (Vosk), and Whisper-based scripts are the strongest free, open-source options — all need terminal setup. For a polished app with system-wide hotkey typing and no scripting, AirTypes has a 7-day free trial.

What is the best offline voice recognition for Linux?

A Whisper-based tool. AirTypes packages Whisper into a native Linux app (.deb, .rpm, AppImage) with offline transcription and system-wide injection on X11 and Wayland — one of the few turnkey offline options.

Does voice dictation on Linux work on Wayland?

Yes. Wayland restricts direct keyboard injection, so tools adapt. AirTypes automatically uses a clipboard-paste mode on Wayland (GNOME, KDE Plasma, Sway) so transcribed text still lands in your apps.

Do I need a GPU for offline voice dictation on Linux?

No. Whisper models run on the CPU. Any modern x86_64 processor with 4 GB RAM handles the smaller tiers; 8 GB is recommended for the largest, most accurate models.

Related: read how offline speech recognition actually works, compare the best offline speech recognition apps across all platforms, or start with the complete guide to offline voice-to-text.