Innovative AI Solution: Browser-Based Voice Assistant
I’ve developed a groundbreaking voice assistant that operates entirely within the browser—no backend or API calls needed! This project explores the capabilities of modern browser-based AI, aiming for a full voice pipeline that runs client-side with minimal latency.
How it works:
- Speech to Text: Utilizes Whisper tiny en via WebAssembly.
- Language Model: Employs Qwen 2.5 0.5B via a Llama.cpp WASM port.
- Text to Speech: Implements the browser’s SpeechSynthesis API for real-time responses.
Key Features:
- Completely offline after initial load—ensuring privacy.
- Works seamlessly in Chrome or Edge 90+.
- Currently available in English, with potential for improvements.
Join me in shaping the future of local AI and browser technologies! I welcome feedback, especially from fellow innovators aiming to enhance performance or mobile support.
👉 Check out the demo and share your thoughts! Demo Link | GitHub Source
