Linux and Windows desktop dictation

Hold a key. Speak. Text appears where you work.

gdictate is a compact desktop app for live dictation, speaker transcription, sequential paste, and local audio/video transcription.

Download latest View source

2hold-to-talk channels

6release package formats

0speech API keys required

1shared settings core

Built for fast transcription workflows

Live dictation

Use microphone mode for your own speech and speaker mode for calls, meetings, videos, or system output.

Tiny live popup

Always-on-top, transparent, click-through capable, and designed to stay out of the way.

Sequential paste

Final recognition chunks can be inserted while dictating, with fallback paste on release.

Daemon and tray

The desktop shell supervises the Python daemon, exposes tray actions, and keeps the app running in the background.

OS diagnostics

Preflight reports explain Chrome, audio, paste, hotkey, popup, and platform-specific setup state.

File transcription

Optional local faster-whisper and diarization pipeline exports TXT, SRT, VTT, and JSON.

Install from release packages

Pick the latest package that matches your system. Linux builds include AppImage, deb, rpm, and Arch pacman package. Windows builds include setup exe and MSI.

Linux portableAppImage

Debian / Ubuntudeb package

Fedora / RPMrpm package

Archpacman package

Windows setupsetup exe

Windows MSIMSI package

git clone https://github.com/bigidulka/gdictate.git
cd gdictate
./install.sh
.venv/bin/python gdictate.py --daemon --no-ui

npm run tauri:dev

Platform notes

Linux

Primary target. PipeWire/Pulse speaker routing, wl-copy paste, ydotool/wtype key injection, evdev hold hotkeys.

Windows

GUI package and microphone path are supported. Speaker transcription needs a loopback recording input such as Stereo Mix or VB-CABLE.

Chrome bridge

Chrome/Chromium/Edge runs as a small hidden Web Speech bridge. No Google Cloud account or paid key is needed.