WhisperType documentation
Complete setup guide, troubleshooting, and reference for WhisperType — local AI voice dictation for Windows and Mac.
Quick start
System requirements
- Windows 10 or 11 (64-bit, Intel or AMD), or macOS (public beta): Apple Silicon Mac: version 1.4.1, macOS 14 or later · Intel Mac: version 1.3.2, macOS 11 or later
- GPU: optional. An NVIDIA card is used when the CUDA 12 libraries are present; Apple Silicon uses Metal
- 75 MB to 3 GB of disk space for the model you choose on first run
- No GPU? WhisperType runs on the CPU — just slower
Download and install
- Download WhisperType from the Download page
- Run the installer (
WhisperType-Setup-…exeon Windows, the.dmgon macOS) - Follow the installation wizard
Windows SmartScreen Warning
On first install, Windows SmartScreen may show a warning. Click "More info" → "Run anyway." WhisperType is not code-signed yet.
macOS: Open Anyway, then three permissions
Open the .dmg and drag WhisperType to Applications. The first launch is blocked: open System Settings → Privacy & Security, scroll to the message about WhisperType and click Open Anyway.
WhisperType then asks for three permissions: Microphone to hear you, Accessibility to type the text, and Input Monitoring to see the F8 key. Updating from 1.3 asks for the macOS permissions one last time; later updates keep them.
First run (model download and trial)
On first launch, setup has you pick a model and start your 14-day free trial:
- Model: you pick one of four, from Base (~75 MB) to Large v3 (~3 GB); Small (~475 MB) is preselected. How long it takes depends on your connection
- Trial: enter an email address to start. No credit card required.
- A progress bar shows the download status. Don't close the app.
- WhisperType then lives in the system tray on Windows and in the menu bar on macOS.
This happens once. After it, transcription needs no connection; the app still goes online for license checks, updates and usage statistics.
Your first transcription
- Open any app with a text field (Discord, VS Code, Word, etc.)
- Place your cursor where you want text to appear
- Press F8 (default hotkey)
- Speak clearly
- Press F8 again to stop
- Text appears at your cursor position
Using WhisperType
The F8 shortcut (and how to change it)
Default: F8 starts/stops recording
To change: open Settings → Hotkey, click Record new hotkey, press the new key or combination and save.
The floating widget
While you record, a small floating bar appears:
- Status: "Listening" while it records, "Transcribing…" after
- Waveform: Audio level visualization
- Drag: Move widget anywhere on screen
What's included
Every trial and every subscription gets the same app — there's no reduced tier. That includes:
| Feature | Included |
|---|---|
| Model | Four Whisper models, Base to Large v3 — you choose |
| Speed | Depends on your hardware and the model you choose |
| GPU Acceleration | Automatic: NVIDIA CUDA on Windows, Metal on Apple Silicon, CPU otherwise |
| Price | $9/month · 14-day free trial, no card |
Supported languages
On Auto, WhisperType chooses between English and Spanish. For another language, pick it in Settings → General → Primary language, which offers:
- English, Spanish, French, German, Portuguese, Italian
- Japanese, Chinese, Korean, Russian, Arabic, Hindi
Troubleshooting
App doesn't start
- On macOS: if the launch is blocked, use Open Anyway in System Settings → Privacy & Security
- On Windows: check Event Viewer for error details
- Try running as Administrator
- Check antivirus/firewall isn't blocking the app
No audio detected
- Check microphone is connected and enabled
- Test the microphone in your system's sound settings; on macOS, check System Settings → Privacy & Security → Microphone
- Ensure no other app is exclusively using the microphone
- Try a different microphone
Text not appearing in app
- Ensure cursor is in a text field
- Try in a different app (Notepad, Word, etc.)
- Check if target app requires elevated privileges
- On macOS: check that Accessibility is granted in System Settings → Privacy & Security
GPU not being used
- Ensure NVIDIA drivers are up to date
- The NVIDIA path needs the CUDA 12 libraries; without them WhisperType uses the CPU
- There is no GPU switch: WhisperType uses an NVIDIA GPU on Windows or Apple Silicon on a Mac when it can, and the CPU otherwise. AMD and Intel graphics are not used
Common Windows 11 issues
- Microphone privacy settings: Allow apps to access microphone
- Real-time protection: Add WhisperType to Windows Defender exclusions
- Power settings: Ensure high-performance power plan
License
Starting your trial
- Download and install WhisperType
- On first launch, enter an email address to start your 14-day free trial — no credit card
- The full app is unlocked for the length of the trial
Subscribing
- Subscriptions are not open yet, and nobody is charged. Get an email the day they open
- When they open: you subscribe from inside the app for $9/month, pay through Stripe, receive a license key by email, and paste it in Settings → License (at the top of the General tab)
What your subscription includes
- All four models, Base to Large v3
- Every feature, on one active machine at a time
- All future updates for as long as you're subscribed
There's no reduced free tier anymore: if your trial or subscription lapses, the app stops running until you activate again.
Moving to a new machine
- Deactivate the license in Settings → License on the old machine
- Activate it on the new machine with the same key
- Lost access to the old machine? Email hello@whispertype.ai and we'll reset the activation
Cancelling and refunds
Subscriptions are not open yet, so nobody is charged and there is nothing to cancel. See the Refund & Cancellation Policy for how billing, cancellation, and refunds work.
System requirements
- OS: Windows 10 / 11 (64-bit, Intel or AMD) · Apple Silicon Mac: version 1.4.1, macOS 14 or later · Intel Mac: version 1.3.2, macOS 11 or later (public beta)
- GPU: optional · NVIDIA with the CUDA 12 libraries on Windows · Metal on Apple Silicon
- No GPU: Any CPU — WhisperType falls back to CPU transcription, just slower
- Disk space: 75 MB to 3 GB for the model you choose on first run
Everything runs locally — no internet connection is needed for transcription after the model download.