Inspiration

Dialects often differ from their base languages in unique words and sentence structures. We wanted to help people communicate naturally, even across dialects they don’t use every day.

What It Does

Dialink offers speech-to-text and text-to-text translation between dialects. You choose a source and target dialect, and it outputs a version that sounds natural to native speakers.

How We Built It

We used React for the frontend, FastAPI for the backend, and a fine-tuned transformer model for translation between dialects.

Challenges We Ran Into

Tuning the transformer to handle subtle dialect differences was tough, especially around language filtering and mismatched translations.

Accomplishments We’re Proud Of

We achieved reliable, natural translations between dialects of our native language — good enough that we’d actually use it daily.

What We Learned

We deepened our understanding of language models, fine-tuning, and built stronger skills in React and FastAPI.

What’s Next

We plan to integrate Dialink into video call apps like Google Meet or Zoom for live, real-time translation.

Built With

Share this project:

Updates