Lost in Lecture

Lost in Lecture

Inspiration

We all know how hard it is to stay focused during long lectures, especially after a big meal. In today’s world, where we’re used to fast, bite-sized entertainment, staying engaged for 45 minutes can be tough. We wanted to create something that would help students stay attentive and make learning more interactive.

What it does

"Lost in Lecture" is a Chrome extension designed to keep students engaged during lectures. Once you start recording, the extension listens to the lecture in real-time, converts the speech to text, and sends it to the backend for analysis. It then generates quizzes based on the content being taught and pops them up at regular intervals or whenever the student chooses to engage.

At the end of the lecture, the extension summarizes the lecture, so students can easily review what they’ve learned and save the summary for later reference. All of this is stored in Firebase, and students can access it via a website built with Next.js and React.

How we built it

We used several technologies to make Lost in Lecture work seamlessly:

Vosk for real-time speech-to-text conversion.
Meta-Llama-3.1-8B-Instruct-fast AI to summarize the lecture and generate quizzes.
Python Flask to handle backend processes.
Firebase for storing lecture summaries and quizzes.
Next.js and React for the companion website, ensuring a smooth user experience across devices.

Challenges we ran into

Real-time processing: Converting speech to text in real-time and making sure quizzes pop up without delay was a challenge.
AI accuracy: Ensuring that the AI-generated summaries and quizzes were relevant and accurate to the lecture content.
Syncing across platforms: Making sure everything, from summaries to quizzes, synced seamlessly between the Chrome extension and the companion website.

Accomplishments that we’re proud of

The extension works in real-time, engaging students during lectures and making learning more interactive.
We were able to successfully integrate Vosk for speech-to-text, Meta-Llama for summaries, and Firebase for data storage in one smooth experience.
The website is functional, allowing students to access saved summaries and quizzes anytime, from any device.

What we learned

Real-time AI processing can be tricky, but with the right tools, it’s possible to make it seamless.
Handling data across multiple platforms requires careful syncing and testing to ensure a smooth experience for the user.
We learned a lot about integrating advanced technologies like Vosk and Meta-Llama in a user-friendly way.

What’s next for Lost in Lecture

We plan to improve the AI’s quiz generation capabilities to make them even more accurate and varied.
Expanding to support other types of content (like videos and images) to create an even richer learning experience.
Adding more features to the website, like study progress tracking and personalized quiz recommendations.