AutoNoteTaker(ANT)

About AutoNoteTaker:

AutoNoteTaker is a speech-to-text program that transcribes ALL your lectures (and any other videos) for you automatically!

Our Inspiration:

This program was developed for the AssemblyAI challenge at MacHacks 2, and our main inspiration behind it was to create something that could help make the lives of all students that have online lectures easier and facilitate their learning.

What it does:

It takes in any mp4 or mp3 file, such as those of your online lectures, and produces a written script of everything said in the video. This program uses the power of AssemblyAI, an artificial intelligence API to do the raw transcription for us. It then exports this transcription into an easy-to-use Word document, complete with formatting, an auto-summary, a list of keywords/concepts (called ‘entities’), and the main topic of the discussion - all deduced using machine learning

How we built it:

It was built using python with a couple of libraries including docx, requests, systems, etc.

Challenges we ran into:

In this hackathon, we have faced and captured a lot of challenges. One example is when we tried to use an advanced feature from AssemblyAI, which can scan the whole transcript and determine its topic automatically by using AI technology. The data we required was only some small parts of a JSON file that was sent from the AssemblyAI. And we had to use data analysis techniques to extract the data we wanted, which was time-consuming.

What we learned:

In this hackathon we learned:

how to work with the AssemblyAI API and implement their features into python by using libraries, functions and File I/O.
how to export text to a Word document instead of a text file, using the Python docx library.
how to serialize/deserialize JSON files into Python files
how to work as a team efficiently under pressure by dividing the workload and accomplishing the task.

What's next for AutoNoteTaker(ANT) project:

Some next steps we have to further develop this program would be to add real-time transcription with more autonomous features like Sentiment Analysis, Content Moderation, and PII Redaction. We would also make it easier for users to transcribe directly from links rather than downloading it and inputting the path of the file.