100% offline · Runs on your Mac

Dictation that
never leaves
your Mac

Press a hotkey, speak, done. Your words appear at the cursor — transcribed entirely on your Mac. No cloud, no account, no data collected.

Download .dmg View Source
Your voice stays
on your hardware

No servers. No API keys. No analytics. Audio goes from your microphone to your cursor — then it's gone. Nothing is ever written to disk.

🔒

No network requests

After the initial model download, LocalWhisper never makes a single network call. Verify it yourself — it's open source.

💾

RAM-only processing

Audio is captured as float32 in memory, transcribed, and immediately discarded. Nothing is ever saved to disk — not even temporarily.

🔓

Fully open source

Every line of code is on GitHub. Audit the source, build it yourself, verify every claim. Transparency is not optional — it's the architecture.

Three steps.
That's it.
1

Tap or hold your key

Double-tap Ctrl to toggle, or hold Right ⌘ to push-to-talk — your choice. Recording starts instantly with a floating overlay.

Ctrl+Ctrl
2

Do your thing

Speak naturally while you keep working — switch apps, browse, type. The recording runs in the background.

3

Done — text appears

Double-tap again (or release the hold key) to stop. Your speech is transcribed and pasted wherever your cursor is. Not in a text field? No problem — it's saved in the menu bar for you to copy later.

A lock that listens.
Animated, alive.

The lock + waveform icon reflects the current state with live animations. Adapts to light and dark mode automatically.

Idle
Static lock + waveform
Ready for input
Recording
Bars pulse with red glow
Active voice capture
Processing
Spinner replaces bars
Transcribing with Whisper
Д
Translating
Dual arcs spinning
Translating to English
Simple and powerful.
Everything you need.
🌐

12 languages

Speak in German, French, Spanish, Japanese, Chinese, and 7 more languages. Auto-detect available too.

🔮

Translate to English

Speak in any language and get English output — translation happens on-device, no extra download needed.

Clean up text

Toggle on “Clean up” to automatically fix grammar, remove filler words like “um” and “uh”, and cut false starts.

5 model sizes

From Tiny (75 MB, near-instant) to Large (3 GB, best accuracy). Pick your speed vs. quality preference in Settings.

Customizable shortcut

Double-tap (Ctrl, Cmd, Shift) or hold (Right ⌘, Right ⌥) — choose your style. Press Escape to cancel.

🎨

Animated overlay

A floating lock icon shows recording, processing, and translating states with live animations. Works in light and dark mode.

From raw speech
to clean text

Toggle “Clean up” and the app removes filler words, fixes punctuation, and cuts false starts — keeping your exact words and sentence structure. All on-device.

Before

“so yeah i went to the uh store and like bought some stuff um and then i went home”

After cleanup

“I went to the store and bought some stuff, and then I went home.”
Up and running in
under a minute

Download the app

Grab the .dmg, drag to Applications. On first launch, go to System Settings → Privacy & Security → Open Anyway (app is unsigned). Then grant Microphone & Accessibility permissions.

Download from GitHub Releases
Or build from source
# Clone & install brew install ffmpeg git clone https://github.com/gnlcsvn/local-whisper.git cd local-whisper python -m venv venv && source venv/bin/activate pip install -r requirements.txt # Run directly python main.py # Or build .app pip install pyinstaller pyinstaller LocalWhisper.spec --noconfirm

Requires a Mac with Apple Silicon (M1 or later) · macOS 13+