Inspiration
We've all had that experience the night before the test when we're trying to watch all the KhanAcademy or Crash Course: History videos that we can. We wanted to make it easier to go through YouTube videos and get the most important parts. That's how we decided to make Too Long; Didn't Watch (TL;DW)
What it does
TL;DW allows the user to input the URL of a YouTube video. Our product will then display the most important clips from the video alongside the important captions, skipping over parts of the video that were not deemed important. The user can click on captions to go to the part of the video where the caption occurred.
How we built it
Our backend is a server hosted on Microsoft Azure, in which we hosted our MongoDB database. We used node.js and a number of APIs to do the heavy lifting on summarization, such as Microsoft's Video Indexer and Text Analytics APIs.
Challenges we ran into
Most of us had very little experience with node.js and MongoDB, so it was a real learning process as we sorted out our bugs along the way.
Accomplishments that we're proud of
In the end, managing to leverage the APIs felt amazing, because a vision we had at the start of the hackathon had finally come to fruition.
What we learned
We learned a heck of a lot about node.js, MongoDB, and
What's next for TL;DW
Speeding up and improving summaries is next. It's something that we used APIs for this time around, so it would be difficult but worthwhile to improve
Log in or sign up for Devpost to join the conversation.