Offline Voice Assistant for Linux

Private, on-device voice assistant built with local AI models. Runs entirely offline after initial setup—no cloud, no data sharing.

Features

• Real-time speech-to-text: Fast dictation using distilled Whisper.
• Local text generation: Contextual responses powered by Qwen2.5-0.5B-Instruct.
• Voice-controlled automation: Launch apps and execute workflows via spoken commands.
• Text summarization: Instantly summarize selected text with one hotkey.
• Seamless desktop integration: Auto-types output directly into active windows using xdotool.

Built for speed and privacy on Linux desktops. Lightweight models run efficiently on consumer hardware (CPU/GPU).

Experience fluid, Mac-like voice interaction—completely local and under your control.

Installation

pip install torch transformers sounddevice scipy pynput
sudo apt install xdotool xclip

Usage

Run with python voice_assistant.py. Models auto-download on first launch (internet required once).

Hotkeys:

• Alt+1: Dictation mode
• Alt+2: Generate response from speech
• Alt+4: Launch apps by voice
• Alt+5: Summarize selected text

Designed and built by a GenAI engineer focused on practical, private Linux tools.