Clear Voice Extensions

ClearVoice

Inspiration

Many students don’t struggle because they lack knowledge, instead they struggle because they don’t understand what a question is asking.
This is especially true with dense academic language, math-heavy problems, PDFs, and content written in a second language.

We noticed that existing tools either generate too much text, solve problems directly (which isn’t always helpful), or don’t work well with PDFs, equations, or accessibility needs. We wanted to build something that helps students understand first, before they start solving.

That idea became ClearVoice.

What it does

ClearVoice is a student-first Chrome extension that turns confusing text, questions, and equations into short, clear explanations.

Students can:

Highlight text on any website or
Use Snip Screenshot to capture PDFs, images, or math equations

ClearVoice then provides focused tools:

What is this asking? – explains the goal of a question in one sentence
Break into steps – breaks tasks into simple steps without solving
Plain English – rewrites complex text in very simple language
Which concept? – identifies the key concept or formula involved

It also supports multiple languages:

English
French
Spanish
Mandarin
Punjabi

Finally, ClearVoice can read explanations out loud, helping with accessibility, focus, and language learning.

How we built it

ClearVoice is built as a Chrome Extension with a clean side panel UI.

Frontend:
- JavaScript, HTML, CSS
- Custom snip overlay for screenshot selection
- KaTeX for rendering math cleanly
Backend:
- Cloudflare Workers for a fast, serverless API
- Google Gemini API for text understanding, OCR, and explanations
- ElevenLabs for natural text-to-speech
Architecture:
- Text highlight + OCR fallback
- Short, mode-specific prompts to avoid over-explaining
- Multi-language UI and responses

Challenges we ran into

Making text selection work reliably across websites, inputs, and PDFs
Handling OCR rate limits gracefully without breaking the user experience
Ensuring math equations display clearly and are also readable out loud
Keeping explanations short and helpful, instead of overwhelming

Each challenge pushed us to simplify further and focus on student usability.

What we learned

Less output often leads to better understanding
Accessibility features (voice, language support) help everyone, not just edge cases
Serverless tools like Cloudflare Workers are incredibly powerful for rapid iteration
Clear UX matters just as much as strong AI models