AccSense

Inspiration

Several past research studies have shown that individuals are judged and perceived negatively for the accents they carry. Individuals are further led to feed into these projections when they’re exposed to the plethora of materials online that “guide” them into getting rid of their accent. We built AccSense to challenge these ideals because we believe accents should be normalized rather than shunned. Through our platform we hope to spread and share culture

What it does

AccSense is a communication aid designed to bridge language and accent barriers. It provides tools like speech-to-text, text-to-audio, and accent-aware analysis to make conversations clearer and more inclusive across diverse cultures and dialects.

How we built it

It utilizes a pipeline consisting of fine-tuned speech recognition, an emotional analysis layer, and a speech simplification layer to make conversations clearer and more inclusive while preserving the cultural nuances and dialects.

Challenges we ran into

LLM vs STT Model. Using an LLM for STT is an easy way of implementing our end goal, but fine tuning a STT model can return greater results for our use case of accent specific recognition.

Accomplishments that we're proud of

Creation of a Full Stack Application. As beginner and less experienced hackers, it was a first for each of us building a Full Stack Application.

What we learned

Tech Stack: We each tackled a new set of tools in order to complete this project. Learning and creating at the same time. Time Management: Staying on track with our code and progress is hard to manage in such a short amount of time. We are coming away with greater awareness of what it means to build a project.