Free · Windows Portable

typwrtr

Cross-platform desktop dictation app. Press a global hotkey, speak into any focused app, get clean transcribed text pasted in. Whisper runs in-process with GPU acceleration — no cloud, no API keys, no audio ever leaves your machine.

Local-only Whisper.cpp CUDA / Metal Tauri + Rust Self-learning

Windows · NVIDIA GPU CUDA · .exe · 23 MB Windows · CPU only .exe · 8 MB View source

v0.1.1 · Windows 10/11 · macOS build from source · Pick CUDA if you have an NVIDIA GPU, otherwise grab CPU.

Need an MSI for silent install / Group Policy? CUDA .msi (35 MB) · CPU .msi (21 MB)

What it does

Dictate anywhere

Global hotkey records mic audio, whisper.cpp transcribes in-process, the cleaned text is synthesised-pasted into whatever app has focus.

Fully local

No cloud. No API keys. No daemons. The only network call is the one-time Whisper model download.

Self-learning

Correct a transcription once and the replacement table biases future inferences toward your jargon, names, and homophones.

Per-app profiles

VS Code gets the technical-vocabulary prompt; Slack gets default. Postprocess modes: default / markdown / plain / code.

Voice commands

Inline phrases like new line, period, cap that, code mode, clipboard instead.

Clipboard-safe paste

Whatever was on your clipboard before dictating is restored ~120 ms after the synthesised paste.

Default hotkeys

Action	Windows	macOS
Toggle recording	Ctrl + Shift + Space	Cmd + Shift + Space
Push-to-talk	Ctrl + Shift + Enter	Cmd + Shift + Enter
Fix-up last transcription	Ctrl + Shift + ;	Cmd + Shift + ;

First launch

1. Unzip typwrtr-windows-portable.zip
2. Run typwrtr.exe (no install required)
3. Pick your microphone in Settings
4. Download a Whisper model from the app (medium.en is a good default)
5. Press Ctrl+Shift+Space anywhere and start speaking