Local AI ยท No cloud dependency for speech

Speak naturally.
Get perfect text.

Hold Fn, say anything, release โ€” and polished text appears exactly where you need it. Powered by on-device Whisper and cloud AI.

Voice Input
Hold Fn to record
๐ŸŽ™๏ธ
๐Ÿ“
โœจ
โšก
Output
100% Private transcription
<2s End-to-end latency
3 AI writing modes
Any app Works system-wide

Four steps. Zero friction.

From pressing a key to perfectly formatted text appearing in your app โ€” all in seconds.

01

Hold & Speak

Press and hold Fn anywhere on your system โ€” no need to switch apps or click anything.

02

On-Device Transcription

Whisper AI runs locally on your machine โ€” your voice never leaves your computer. Fast, private, and accurate.

03

AI Refinement

Amazon Nova AI polishes your transcript โ€” rephrased for clarity, grammar-corrected, or turned into actionable guidance.

04

Auto-Paste

Refined text is automatically pasted into whatever app or field was active โ€” email, Slack, Google Docs, anywhere.

Built for speed & privacy

Everything you need to turn speech into perfect written words, without interrupting your flow.

System-Wide Hotkey

Works in any application without needing to focus the Voice Input window. Hold Fn and speak โ€” release to process.

Fn macOS & Windows

100% Private Transcription

Whisper AI runs entirely on your device. Your voice audio never touches a server. Local, fast, private.

Real-Time Status Feedback

A tiny floating overlay shows exactly which stage you're in โ€” recording, transcribing, rephrasing, or done โ€” at a glance.

Amazon Nova AI

Cloud AI refines your transcription for clarity, grammar, and style โ€” making rough spoken words into polished written text.

Smart Auto-Paste

Refined text is automatically injected into the active field โ€” no clipboard tricks needed. Falls back gracefully if permissions are limited.

Debug Log Panel

A collapsible pipeline log shows every step with timestamps and color-coded severity. Perfect for understanding exactly what happened.

Three ways to transform your voice

Choose the mode that fits your intent โ€” click each to see an example.

You said
"um so basically i was thinking we should maybe like push the deadline back a week because the team is kinda swamped right now you know"
Rephrase
Rephrased output
"I'd like to propose extending the project deadline by one week. The team is currently at capacity, and this additional time would help ensure quality delivery."
Rephrase mode restructures your spoken words into clear, professional, well-organized text. Ideal for emails, messages, and documentation.
You said
"the new feature gonna be ship next week, we needs to test it thoroghly before we release it to the customers"
Polish
Polished output
"The new feature will ship next week. We need to test it thoroughly before releasing it to customers."
Polish mode fixes grammar, spelling, and awkward phrasing while preserving your original voice and structure. Perfect for quick corrections.
You said
"how do I set up a GitHub Actions workflow that runs my tests automatically whenever I push to main"
Task Support
Actionable output
"1. Create .github/workflows/test.yml
2. Set trigger: on: push: branches: [main]
3. Add jobs with runs-on: ubuntu-latest
4. Steps: checkout โ†’ setup-node โ†’ npm install โ†’ npm test"
Task Support mode treats your speech as a question or task and provides step-by-step actionable guidance. Great for capturing quick how-to notes.

Always know what's happening

The floating overlay icon changes color and animates through each stage of the pipeline.

Idle
Ready to record
Recording
Capturing audio
Transcribing
On-device Whisper
Rephrasing
AI processing
Done
Text pasted!
Error
With recovery
Idle
๐ŸŽ™ Recording
๐Ÿ“ Transcribing
โœจ Rephrasing
โœ“ Done

Available for Mac & Windows

Free to download. Works out of the box.

iOS โ€” Coming Soon Android โ€” Coming Soon
macOS
macOS 12 Monterey or later ยท Apple Silicon & Intel
Universal Binary Notarized
Download for Mac .dmg ยท v1.0.0
Windows
Windows 10 or later ยท 64-bit
NSIS Installer Portable .exe
Voice audio is processed entirely on-device. Only the text transcript is sent to the cloud for AI refinement.