Inspiration
We noticed that many busy professionals and students often stop watching long tutorials or online courses halfway through. Our idea was to create a solution that highlights the most important parts of these videos, helping people quickly find and learn the key insights without getting overwhelmed or losing interest.
What it does
Our tool automatically transforms long educational videos into bite-sized summaries. It pulls the transcript, identifies the important sections, and presents them with clear summaries and time-stamped markers. That way, learners can jump straight to the topics they care about the most and absorb the content more efficiently.
How we built it
We integrated the YouTube Data API to obtain transcripts and timestamps. Then, we used Gemini to format the content into structured JSON. We analyzed the transcript to pinpoint the essential moments and synchronized those with timestamps to create short clips. On the development side, we combined Node.js, Vue.js, Firebase Hosting, Firestore, Gemini API, Apify, and Vercel for seamless functionality and deployment.
Challenges we ran into
We faced a major issue with YouTube not working as expected in production, which slowed down our video processing. Additionally, we had to overcome rate limits when using Gemini, which required optimizing how we handled our requests.
Accomplishments that we're proud of
We successfully tackled the rate limit problems, ensuring the tool could handle larger batches of video transcripts. We also integrated a feature that allows users to chat with the material for further insights, making the learning experience more interactive.
What we learned
Throughout the process, we discovered how to efficiently convert long-form content into short, focused highlights. We also deepened our understanding of using Gemini to produce clean, structured JSON outputs for better organization and data handling.
What's next for Shortok
Looking ahead, we plan to implement asynchronous requests to handle videos even more quickly and efficiently. We also aim to offer live Spanish audio versions of the highlights, making the learning experience accessible to a multilingual audience.
Log in or sign up for Devpost to join the conversation.