Inspiration

I noticed traditional language apps are often disconnected from real-world nuances. I built LinguaLearn to adapt to a learner's immediate needs by generating custom lessons on the fly for specific scenarios like business trips or interviews.

What it does

LinguaLearn is an AI-powered app that moves beyond static textbooks:

  • Generates bespoke lessons from user descriptions.
  • Includes high-fidelity Text-to-Speech for 10 languages.
  • Allows learners to save words to a personal Vocabulary Vault.

How we built it

I engineered the platform using a modern, performance-focused stack: Frontend: React 19 and TypeScript. Build: Vite. AI: Gemini 3.0 Flash for structured data and Gemini 2.5 Flash for audio. Styling: Tailwind CSS for a minimalist "Ink & Paper" design system.

Challenges we ran into

The primary challenge was stabilizing experimental TTS models. I engineered a recovery pipeline featuring auto-sanitization, tiered prompts, and exponential backoff to handle server errors and rate limits.

Accomplishments that we're proud of

  • Created a seamless user experience from scenario input to lesson generation.
  • Implemented robust support for 10 languages, including complex scripts like Japanese and Korean.

What we learned

I learned to manage experimental AI APIs and refined my skills in creating focused educational environments using minimalist UI/UX principles.

What's next for LanguaLearn

  • User accounts for cloud synchronization.
  • Gamification features including streaks and daily goals.
  • Real-time AI conversation mode for spoken practice.
  • Image-to-lesson capabilities for translating menus or signs.

Built With

Share this project:

Updates