Video Summarizer

Generated Transcript (using Google Cloud Video Intelligence API)
Generated Timestamp

Inspiration

When learning from online content and recording of live lectures, it often takes a long time to thoroughly review the content. One can put the video on 2x speed, but is there a more efficient way to cover the curriculum? In efforts to create a better way to learn video content, we employed AI to learn from and take away the main ideas of a lecture, allowing the user to focus on core topics and cut-down on video-watching time.

What it does

This program takes in a YouTube video link to create and return a summarized transcript of the YouTube video's content to the user. It also returns a list of important words along with the timestamps relevant to those keywords, enabling the user to directly jump to that particular timestamp when they are watching the video.

How we built it

This program starts by putting YouTube videos into the Google Cloud API to create a text transcript of the video. This transcript is then sent to the OpenAI API to summarize. The summary along with the timestamp is then displayed on the screen for the end user.

Challenges we ran into

To use the Google Cloud API, we needed to learn how to convert the YouTube videos to an object storable in the Cloud so that it could be converted to text. The current version of Google Cloud Video Intelligence is built in such a way that it supports videos that are present in a cloud bucket.

Accomplishments that we're proud of

We are proud of pulling together a large program in a small amount of time, especially considering this is our first time participating in a hackathon.

What we learned

We learned how to quickly scan Python library documentation to identify what can help us create the program that we need, as well as figuring out how to delegate tasks properly to get our project done in the allotted time frame.

What's next for Video Summarizer

The next steps of the project include, but are not limited to, creating an animated video of the created transcripts and also providing of concatenating the original video based on the summary and timestamps.