Skip to main content

Documentation Index

Fetch the complete documentation index at: https://vowen.mintlify.app/llms.txt

Use this file to discover all available pages before exploring further.

​
Supported Hardware

HardwareStatus
Windows x64Fully supported
Windows ARM64Experimental (slower due to emulation)
NVIDIA GPUOptional, enables acceleration

​
Key Differences from macOS

FeaturemacOSWindows
Default shortcutFnCtrl + Shift
AI shortcutAlt + ShiftAlt + Shift
”Option” keyOptionAlt
”Command” keyCommandWindows key
Fn key supportYesNo
PermissionsManual (Accessibility, Mic)Automatic
Code signingYes (notarized)Not yet
GPU accelerationNot needed (Neural Engine)NVIDIA CUDA

​
Windows-Specific Features

​
GPU Acceleration (NVIDIA)

Speed up local model transcription dramatically:
  1. Go to Settings > Models
  2. Scroll to “GPU Acceleration” section
  3. Click Download to get CUDA modules
  4. Restart Vowen (or restart system if needed)
With GPU acceleration, Large v3 Turbo responds in 1-2 seconds.
Only NVIDIA GPUs with CUDA support are currently compatible. AMD and Intel GPUs are not supported.

​
Start at Login

Enable automatic startup: Settings > General > Open at Login

​
Minimize to Tray

Start the app minimized: Settings > General > Start minimized to tray

​
System Tray

Right-click the Vowen icon in the system tray (bottom-right) for:
  • Model and microphone switching
  • Meeting notes controls
  • Copy last transcription
  • Open settings
  • Quit

​
Installation Notes

​
Windows Defender Warning

Windows shows “Windows protected your PC” because the app isn’t code-signed with an EV certificate. This is normal for indie software. To proceed: Click “More info” → “Run anyway”

​
Antivirus False Positives

Some antivirus software (QuickHeal, Kaspersky, Norton) may flag Vowen. Add the Vowen installation folder to your antivirus whitelist.

​
Performance Tips

Local models are slower on Windows than macOS (no Neural Engine). To get fast transcription:
  1. Best: Use a cloud model (Groq — free and fast)
  2. Good: Enable GPU acceleration (NVIDIA only)
  3. OK: Use Tiny or Base model (fast but less accurate)
Without GPU acceleration, the Medium and Large models can take 5-10+ seconds to transcribe on Windows. Cloud models are strongly recommended for Windows users without NVIDIA GPUs.
Have a Windows-specific question? See the Windows FAQ.