Dictate into any text field — offline, private, and instant. Powered by on-device Whisper + Chrome AI.
VoiceFill is a Chrome extension that leverages on-device machine learning to provide fast, private speech-to-text transcription directly in your browser. It integrates smoothly into any webpage using injected content scripts and an easy-to-use side panel.
- On-Device ML: Uses Hugging Face Transformers for local, privacy-first transcription.
- Offline Capable: Works without an internet connection after the initial model download.
- Instant Integration: Dictate directly into any active text field on web pages.
- Modern UI/UX: Built with React, Tailwind CSS, and Zustand.
- Framework: React 19 + TypeScript
- Build Tool: Vite + CRXJS (for manifest V3 generation)
- Machine Learning:
@huggingface/transformers(running in an Offscreen Document) - Styling: Tailwind CSS
- State Management: Zustand
- Messaging:
webext-bridgefor seamless communication between Extension contexts.
src/background/: Background service worker managing sessions and the offscreen document.src/content/: Content scripts injected into web pages to handle UI interactions and input manipulation.src/offscreen/: Offscreen document housing the heavy Whisper model Web Worker for audio processing.src/sidepanel/: Side panel UI providing control over the transcription process and settings.manifest.config.ts: Chrome Extension Manifest configuration powered by CRXJS.
npm installnpm run devOpen Chrome and navigate to chrome://extensions/. Enable "Developer mode" and click "Load unpacked", then select the generated dist folder.
npm run build