We thought of making an application which can summarize videos. This way we can automate note-taking.

What it does

First the user enters the video file and then the website returns transcript and the summary of the video.

How we built it

For the backend, we used ffmpeg to get the audio from the video and used the Azure Speech SDK to get the transcript. Then we used extractive text summarisation using BERT to obtain the a summary of the timestamps. HTML, CSS and JavaScript was used for the front-end. The website accept the video file from the user and displays the transcript and the summary. The Flask API was used for the middle end.

Challenges we ran into

  1. Package Management (External Packages would not install properly)
  2. Issues with the Flask API. (It would often return errors with no explanation.)

Accomplishments that we're proud of

  1. The backend process of converting the video file into a transcript and summary.
  2. Getting the Flask API to work.
  3. Package management.
  4. Fixing Errors.

What we learned

  1. General Front and Backend programming.
  2. Implementing web development with python.

What's next for Video Summarizer

  1. We could develop the application further to accept URLs and audio files.
  2. Figuring out what the timestamp offset format is and displaying it as minutes:seconds
Share this project: