Individual word translation feature
Recording functionality for voice
Translation feature
Chat Screen
Multiple chats supported

About LingoBuddy

Inspiration

Traditional language learning apps like Duolingo focus on memorizing words, but this approach has limitations. Most people learn languages for travel where they need speaking skills, not reading... You can always use Google Translate for text. LingoBuddy bridges the gap between memorization and real conversation practice.

What it does

A conversational AI that lets you have real voice conversations in your target language. All chats are stored so you can review them later, translate words you missed, and track your progress over time.

How we built it

Backend: AWS serverless with Lambda functions, Deepgram for speech-to-text, OpenAI for responses, ElevenLabs for text-to-speech, and Supabase for data storage.

Frontend: React with Vite, TypeScript, and Tailwind CSS for a modern, responsive interface.

Challenges we ran into

Getting the API flow right was difficult - Deepgram wasn't cooperating initially. Finding the right ElevenLabs voices took time since many had accents. Hooking up the frontend audio recorder was challenging as it was my first time working with audio APIs.

Accomplishments that we're proud of

Created a fully functional app with user accounts, conversation storage, and seamless voice interaction. Successfully implemented technologies I've been wanting to try, and built a scalable serverless architecture.

What we learned

Hosting online is easier than expected - AWS serverless framework gets your backend running quickly. Working with multiple APIs taught us about error handling and fallback strategies.