Inspiration

One common problem faced by all of us was the large amount of information present in textbooks and lectures. Sometimes the professor might not post their slides and they might sometimes have hand writing that isn't the easiest to read. Thats what caused us to think about a solution that is friendly for users while still being easy to use.

What it does

ETA takes in textbooks or notes from a class, and from that you can ask it to generate practice problems or summaries for the current content being covered in the course. We also have different modes so that you can get exactly what you need whether it be exam review, lecture summaries or a quiz mode.

How we built it

The frontend was built with React and Three.js for the 3d models. We also used DynamoDB to store the provided context from the user so that ETA can always give relevant course content. Furthermore, we also used ElevenLabs to generate a friendly sounding voice so that the user doesn't feel overwhelmed by blocks of text.

Challenges we ran into

A challenge that we ran into was deciding how to optimize our token efficiency whether it be towards writing to the database, generating prompts or generating a voice for the virtual assistant. We accomplished this by storing the vocal files in the cache instead of writing them to the database as well as being particular on how much context we are storing at any given time.

Accomplishments that we're proud of

An accomplishment that we are proud of is getting all the models and animations working properly as we have no experience incorporating 3D models in a frontend design.

What we learned

We learned how to incorporate and utilize auth0 and DynamoDB.

What's next for ETA

We are looking forward to add speech to speech communication between the user and the virtual assistant as well as adding a weekly plan to speed up user's learning. We would also like to enable the user to add their own models and voices so that they can further personalize their learning.

Share this project:

Updates