TRU: To Remember U

homepage of webapp
mp3 and live transcript
metadata of the user
edit transcript
output of the final generated biography

Inspiration

The digital age has given us countless tools to capture moments, but we often forget the deeper stories, the wisdom passed from generation to generation. Our inspiration for Tru was the realization that while we can take photos and videos, there's something irreplaceably magical about written legacies. We wanted to create a platform where memories, values, and life stories could be preserved in their purest form: words.

What it does

Tru acts as a digital biographer, capturing conversations with loved ones, and then transforming those moments into beautifully written legacies. Through voice-to-text transformation, emotion mapping, and biography customization, Tru ensures that the essence of every tale and sentiment is preserved for future generations.

How we built it

We began with extensive research into transcription algorithms and large-text interpretation to accurately capture conversations. Building on a robust back-end framework, we integrated GPT API to pick up on the nuances of each story. A user-friendly interface was then designed to allow for biography customization.

Challenges we ran into

LLM wise, the processing and handling extremely large prompts posed a significant challenge. Since an extended conversation easily exceeds the token limit of GPT models, we had to design our architecture around that issue. Backend development also proved to be challenging, especially since we wanted to store all prior data and add on to existing generated text.

Accomplishments that we're proud of

We're immensely proud of Tru's ability to capture the essence of personal stories in a written form from only an audio file. We believe that this tool will help people keep memories of their loved ones even after they leave their sides.

What we learned

We delved deeper into the intricacies of voice recognition and natural language processing. We honed our skills in working with LLM APIs, and we have gotten better at framework design for both frontend and backend development.

As a first-time hacking team, we also learned invaluable lessons in cooperation and team development. We each extended our strong suits and worked together seamlessly.

What's next for Tru: To Remember U

We hope to bring Tru to senior homes and give the elderly a chance to record their own tales. We believe that each person's story is one that is worthy of recording, and with Tru we can give more people the chance to do so. Technically, we envision expanding Tru's capabilities to include multimedia elements, like embedding audio snippets or integrating with platforms for an immersive experience. Additionally, we're looking to fine tune our base LLM models with professional writers and biographers to offer an even richer narrative for users.