Inspiration
Every quarter, students here get an email asking if we would like to take notes for students who are physically unable to. We wanted to build a service that does this automatically without the need for student volunteers, and this is how Scribe was born.
What it does
Scribe takes in a video file, usually of a lecture, and outputs a summary in the form of an audio and text file.
How we built it
We used HTML and bootstrap for the front of the website, and Flask and Python for the backend. We used Google Compute Engine's speech-to-text feature to convert the uploaded video file to a text file and the Intellexer API to summarize the text.
Challenges we ran into
The speech-to-text tool is not completely accurate, so the text we get from the video is not optimal. Since the Intellexer API summarizes that text, it has a snowball effect and doesn't result in the most accurate summaries.
What's next for Scribe
We definitely want to improve the frontend of the service, as we didn't get to spend too much time on that aspect. Also, we'd like to experiment with different speech-to-text solutions to see which one is the most accurate.
Log in or sign up for Devpost to join the conversation.