Inspiration

As many more professors find themselves adopting the online blended classroom, many of the lectures can quickly become long, boring, and tedious. This problem is what drove us to create SparkSummary, a machine-learning tool that makes it easier and quicker to take blended classroom course work.

What it does

SparkSummary uses IBM's Watson and Bluemix APIs to listen to the audio files of lectures, transcribes them to a legible text-based format, and finally uses algorithms to determine the key phrases and critical sentences in the lecture. Our program then stores the summaries on a virtual flask server. Users can then retrieve those summaries by either visiting the web page, or they can more conveniently use an IoT device (like an AWS IoT Button that we use) to deliver today's summary straight to the user's mobile device.

How we built it

We started with IBM's machine-learning APIs and worked from there. By building a virtual flask with python, it enabled us to store data to our server and allow users to retrieve the summaries. We also used the AWS IoT platform to connect an IoT Button to our project giving our users more functionality.

Challenges we ran into

LOTS OF THEM... none of our team was familiar with Python, so building a server with Python was rather difficult as we were all learning on the fly. IBM's Watson APIs also had a distinct set of challenges the main of which was that it was difficult to parse sentence based text without any punctuation. The punctuation was critical for the development of the summaries, and playing around with these APIs was not trivial.

Accomplishments that we're proud of

IoT Integration, Audio Transcription, Text Summarization, Back end development, Python-Based Server

What we learned

So.Much.Python. HTTP Get/POST requests were also fundamental for our project, and we had to learn these.

What's next for SparkSummary

Better Speech-to-text and Text-to-summary integration... we also want to add a better file uploading and sharing protocol. If universities start blocking our service from their Wi-Fi, then we take that as a compliment.

Share this project:

Updates