Inspiration

We wanted to build a program to help improve the user's reading comprehension, not only when study is difficult and distracting but also when there is no time to study, such as when one is driving or cleaning their space. Learning requires time and effort; our project comes from a desire to facilitate the effort and make the most of the necessary time.

What it does

The app will read a passage to the user using Gemini's vocal features for the user to intake the text from a database. At the end of the passage, it will question the user based on the contents of the text and use speech-to-text to take in their response, comparing the desired AI-generated answer with the vocal user input.

How we built it

We used GitHub, app.diagrams, Procreate, Python, Java and Android Studios. We split into a backend and frontend team, the frontend team creating assets and XML files, while the backend created middleware and server programming. We integrate with huggingface API to get data from a Project Gutenberg dataset, which is then passed to Gemini for AI transcription, and further analysis.

Challenges we ran into

Getting the AI to ask questions from the passage and exclude questions regarding the passages credits. As well as creating the buttons in Android Studio. We struggled to get the image assets to work across devices.

Accomplishments that we're proud of

Using experimental text-to-speech generation from Gemini, we were able to successfully pass a .wav file to Android to be played aloud for the user.

What we learned

How to format the app using Android Studios. How to avoid GitHub merge conflicts while simultaneously pushing to the same repository. Flask server programming and orchestration using Docker Compose.

What's next for Recall

Making the app compatible with Apple CarPlay and Android Auto, as well as creating different users in order to keep track of user improvement to display diagnostics.

Built With

Share this project:

Updates

Submission history