TextiFy

Translated Text Page
Digitalized Text Page
Home Page

Inspiration

The inspiration for this project came from our work with the Department of English at our university. Our supervisor has been working on digital humanities projects for years, including converting Ruskin manuscripts—a process that previously took hours to transform into XML and then into HTML. We wanted to streamline and modernize this workflow.

What it Does

Texti-Fy converts old manuscripts in Greek, Old English, Latin, and Sanskrit into digital text and provides English translations. We leverage the Classical Language Toolkit (CLTK) for lexical understanding, and we also integrate text-to-speech functionality to make these texts more accessible.

How We Built It

For the backend, we used FastAPI along with Google Cloud Console for OCR and image processing. The frontend was built with Next.js and Material UI for a responsive and modern user interface. This project was made possible through strong collaboration among our team members.

Challenges We Ran Into

One of the main challenges was fine-tuning our image processing with Google Vision. Initially, we achieved only about 60% accuracy. By incorporating advanced computer vision techniques, contextualization, bias detection, and using Gemini, we were able to significantly improve the accuracy. Rate limiting on Gemini, however, presented additional hurdles in retrieving accurate text efficiently.

Accomplishments We're Proud Of

We are proud to have contributed to a project in digital humanities, a field often overlooked despite its importance in preserving culture and tradition. We were able to achieve up to 96% accuracy on some manuscripts—a significant improvement from traditional methods.

What We Learned

Through this project, we gained valuable experience in teamwork and collaboration, managing tight deadlines, networking, and integrating AI technologies like Gemini. We also learned how to fine-tune these tools to fit the specific needs of digital humanities.