PolyglotPlay

Llamas and Pandas Logo

Inspiration

Our team loves to travel and explore places. We all come from different ethnics and lived in different parts of the world. We realized that it was really hard to learn each others language and wanted an easier to learn. We knew chat bots were all type but we really wanted to learn how to speak it, we knew the best way to do it would be through speaking to another person but real life coaches are really expensive and tough. That's when we came up with the idea of PolyglotPlay!

What it does

PolyGlotPlay is an AI-powered language tool that helps you learn essential beginner language skills through a fun, choose-your-own-adventure experience. 🌍 4 Languages 🎭 4 Interactive Scenarios 🛤️ Endless Story Paths Powered by: 🎙️ ElevenLabs – Realistic voice imitation for lifelike conversations 🧠 Google Gemini – Smart, dynamic dialogue generation for natural flow Simply choose a language and a scenario. Then, jump into a real-time voice conversation with your AI teacher — practicing key terms, pronunciation, and real-world speaking skills as you continue the story with your voice. By the end, you're not just learning — you're ready to land, speak, and explore the new world.

How we built it

We started with the Frontend part using languages such as HTML and JavaScript. We wanted to make it easily accessible to everyone hence why we created the program in a website format. We wanted to have a cute and welcoming design, so we had the program have a cute and cartoony design. This made the user feel like that they were really in a story and in their own adventure. Later we worked on the Backend which was main entirely with Python

Challenges we ran into

🧠 Backend Integration Our biggest challenge was connecting the AI voice chatbot to our website. Since Python isn’t naturally web-friendly, we used Flask to bridge Python with JavaScript — but syncing the two came with constant errors. 🎙️ Voice Input Issues Capturing and processing user voice was tricky. We ended up: Recording the user's voice Saving it as an MP3 Sending it to the AI for response 🎨 UI & Animation Constraints Building the UI was manageable, but we struggled with alignment and layout bugs. We envisioned a vibrant, animated background with dynamic scenarios, but time constraints limited us to a simpler design.

Accomplishments that we're proud of

Being able to get the AI coach to really interact with the user and understand them. The AI coach was the hardest part of the project and it really acting close to a human coach is an insane achievement. Linking everything up to the frontend and backend took the most amount of time to do at our beginner level.

What we learned

We learned a lot on how to use AI oriented APIs since going into this none of us had any idea what they're full capability was. We learned a lot about web development integration with Flask. Flask was our main framework that helped us integrate Python into JavaScript. While being very simple, it still required a lot of effort to link up perfectly with the APIs and the website. This helped a lot with We learned a lot about each other strengths and how tiring/stressful it can be when working on a strict deadline. In school, we had a lot of times to work on small assignments but working on a big project in 24 hours is a big constraint.

What's next for PolyglotPlay

Increasing the capabilities of the AI coach like being able to recognize if the user said something correctly but not in the right format. (Ex. In Japanese, a person would have said a word in slang but should have been corrected to say it in a more formal way.)