Press caps lock. Talk to any app.
nui.cat is a native speech to text daemon written in Rust.
Try it free - early access alpha
curl -fsSL https://nui.cat/install.sh | sudo sh Debian 13+, Ubuntu 24.04+, Fedora 40+, Arch (AUR), or AppImage.
nui.cat is simple
Install = done. Push to talk or Toggle mode for longer dictation. Private on-device or cloud. It just works. Linux Wayland and X11. GNOME, KDE, Hyprland, Sway, COSMIC...
Features
- 100 min free monthly.
- Private on-device or cloud
- Vulkan GPU acceleration
- Push-to-talk or toggle modes
- Multilingual — works with 99 mixed-language speech
- Unlimited monthly, yearly, lifetime plans available
Pricing
nui.cat is in alpha, built by a solo dev. Only get a paid plan if you actively want to support development. Things won't always work perfectly.
Upgrade from the app — run nui settings
FAQ
Is my audio private?
With the local engine, audio never leaves your machine. Transcriptions run privately on-device. The cloud engine sends audio to our cloud transcription provider. Audio is never stored in either mode.
Do I need a GPU?
No. nui.cat auto-detects your hardware on first run. If you have a Vulkan-capable GPU (NVIDIA, AMD, or Intel), local transcription uses it automatically. Otherwise, it defaults to the cloud engine.
What languages are supported?
99 languages, including mixed-language speech in the same sentence. The multilingual model handles code-switching naturally — no need to pick a language first.
How much does it cost?
Free tier gives you 100 minutes per month. Unlimited plans start at $9.90/month, with yearly and lifetime options.
What desktops are supported?
GNOME, KDE Plasma, COSMIC, Hyprland, Sway, and any wlroots-based compositor on Wayland. X11 is also fully supported.