Inspiration
We kept hearing from blind/low-vision readers, neurodivergent learners, and folks with limited mobility that traditional screen readers feel rigid. They wanted a calmer “tell me what’s important” companion that also understands plain voice commands. That feedback pushed us to combine AI summaries with conversational controls wrapped in a soothing UI that mirrors the AURA logo.
What it does
- Extracts the main article on any page and either reads it verbatim or sends it through an AI summary mode (if you provide an API key).
- Injects an overlay with live status plus Pause / Resume / Stop controls.
- Accepts hands-free commands for playback (“pause”), navigation (“scroll down 30%”, “go right”), zoom (“zoom in 15%”), and opening sites—even unknown ones via AI URL resolution.
- Offers a polished popup/settings experience with saved speech rate/pitch/voice and AI config.
How we built it
- Chrome extension (Manifest V3) using vanilla JS, HTML, CSS, and shared theme tokens (
src/common/theme.css). contentScriptorchestrates the DOM analyzer, overlay, andpageReader(Web Speech API).- The service worker relays commands, stores settings via
chrome.storage, and optionally proxies AI calls through a Node/Express server. - Voice commands run inside the popup using
SpeechRecognition; they map transcripts to structured commands before messaging the active tab.
Challenges we ran into
- Chrome auto-closes the popup when it loses focus, so keeping voice recognition alive required UX guidance and sparks future architectural work.
- Designing a glassmorphic look without sacrificing accessibility: contrast, dark mode, reduced motion, and focus indicators needed special care.
- Safely interpreting “open <site>” phrases: we had to combine whitelist matching, AI resolution, and sanitized search fallbacks to avoid misnavigation.
Accomplishments that we're proud of
- A cohesive brand experience—gradient typography, feature cards, overlay, popup, and settings all feel like one product.
- Comprehensive voice UX: scroll/zoom/site commands in addition to media control, all with status feedback.
- AI features reuse the same configuration flow for summaries and smart navigation, making advanced capabilities approachable.
What we learned
- Voice-first UX lives or dies on transparency; documenting limitations (e.g., popup focus) builds trust until deeper fixes ship.
- Accessibility guidelines can coexist with modern aesthetics if tokenized and enforced centrally.
- Extension teams benefit from server stubs (our Express proxy) to iterate on AI features without exposing keys during demos.
What's next for AURA - Adaptive Universal Reading Assistant
- Move voice listening outside the popup (service worker or content-side recognizer) so commands keep working while users browse.
- “Social assist” snippets plus tone controls to help users participate in communities more comfortably.
- Richer personalization (e.g., favorite voices per mode, multilingual summaries) and exporting transcripts for note-taking.
Built With
- chrome.runtime
- chrome.storage
- chrome.tabs
- css
- css.-web-apis:-speechsynthesis-and-speechrecognition.-chrome-apis:-chrome.action
- express.js
- html
- javascript
- node.js
- openai
- speechrecognition)-chrome-apis-(chrome.action
- using-javascript
Log in or sign up for Devpost to join the conversation.