We were inspired to increase the public's access to information stored in lengthy videos. From personal experiences, we found it difficult to find what we were looking for in long video lectures, news reports, or earnings calls. This inspired us to built snipped, a better, faster, and more efficient way to access the information we were searching for.

What it does

The product is a web application that summarizes long videos based on the videos' captions. Snipped finds the key subjects and topics discussed in a video, and also pinpoints their location within the video, allowing viewers to jump to the topics they are interested in. It also visualizes data about the video's keywords and displays graphs providing a sentiment based analysis of the video.

How I built it

Our strategy was to use a video API to extract the captions from YouTube videos and then to use IBM's Alchemy API to develop a subject ranking algorithm for the topics in the video. We also used the Wikipedia API to display a short summary of each key word. We also linked the user to the specific time in the YouTube video where the certain key word was discussed.

Challenges I ran into

Our biggest challenges was trying to learn how to implement a variety of APIs into a cohesive, well-organized, summarization algorithm.

Accomplishments that I'm proud of

We're proud of the educational application of our product. If a student doesn't have time to watch long video lectures, the student could use our product to study only the concepts they find most difficult. Our product also gives users the opportunity to learn more about how important certain concepts are to a lecture and to research more information about those concepts.

What I learned

We learned how to use and integrate the various APIs we relied upon. We also got better at quickly finding bugs and learned skills applicable to general about web development.

What's next for Snipped

We hope we can continue to develop and refine Snipped so it becomes an extremely useful resource for individuals to acquire information from lengthy videos.

(Our product currently only works with youtube videos with closed captions)

Share this project: