Install with 1 command:
curl -fsSL https://raw.githubusercontent.com/MagnetosphereLabs/Verbatim/main/install.sh | bash
Uninstall with 1 command:
curl -fsSL https://raw.githubusercontent.com/MagnetosphereLabs/Verbatim/main/uninstall.sh | bash -s -- --purge-system
Local AI voice dictation for Linux. Tested on Pop!_OS, but should also work on common Linux distros like Ubuntu or Linux Mint.
Click a text input, press Super+V, speak naturally, pause for a moment, and Verbatim types what you said into the focused app. It works with browsers and other desktop apps.
- NVIDIA GPU acceleration through faster-whisper and CTranslate2
- AMD and other Vulkan GPU support through whisper.cpp
- CPU fallback only when GPU acceleration is not available
- Automatically detects when WiVRN audio devices are active and connected to your headset.
- Switches your microphone input to the WiVRN headset mic when VR mode is ready.
- Switches your audio output to the WiVRN headset speakers for seamless VR audio.
- Remembers your previous desktop input and output before switching to WiVRN.
- Automatically restores your original audio devices when WiVRN closes or disconnects.
- Runs in the background, so audio switching works even when the overlay is not open.