🕑 Real-Time Whisper Voice Recognition with vosk model feedback 🎙

New logo: Hopefully everyone likes it.
vosper 2.0: The codebase was rewritten and more customization was added!
Pip is coming: A proper, easier installation and update alternative will be launched soon.
vosper 2.1: The codebase is being refactorized for highly-optimized implementation.

• Minimal approach
• Easy installation
• Easy modification
• Fast text feedback thanks to vosk
• Detects human voices (It records audio for Whisper only when needed)

git clone https://github.com/appvoid/vosper.git && 
cd vosper && 
chmod +x install.sh &&
./install.sh

python3 main.py # It's pretty minimal...

paypal donation or patreon support

• Vosk Real-Time inference and Whisper VAD support
• Class-Based implementation
• Easier way to choose a whisper model
• Improved code quality, comments, readability, etc...
• Verbosity switch
• Customizable settings
• Custom VAD model support
• Python's pip installation method
• Keyboard support
• Documentation

Real-Time usage scenarios (like a voice assistant for example) requires a GPU with at least 2-4~ gb of vram. The more the vram, the largest the model you can load, the better the transcription and the slower it gets.