NOTE: Play this clip (http://www.voiptroubleshooter.com/open_speech/american/OSR_us_000_0014_8k.wav) in order to see the audio to effectively see the Speech to Text processes and to verify the audio in the transcript. Additionally, from 2:00 to 3:15, the Speech to Text will be running, and the video will include this one minute pause as the video runs and the transcript prints.
Inspiration
In the area of virtual learning, it takes a long time for students to be able to go through recordings of online lectures fo find the part that they need. It is often a waste of time to go through an entire video to find one thing you need.
What it does
With an input of an audio file, the app takes the audio and converts it into text so the entire transcript can be display. Additionally, for longer scripts, a summary is provided by taking keywords from the transcripts.
How I built it
Speech to text was done with the help of Google Cloud. Front end with HTML/CSS. NLP for summarization was done through the transformer library. Python and Flask to put it all together
Challenges I ran into
It was difficult to incorporate the nlp summarization at first, since it was hard to find an effective way to do this.
Accomplishments that I'm proud of
Working with flask for the first time.
What I learned
I learned how to work with flask.
What's next for SumAI
A chrome extension and timestamp search
Log in or sign up for Devpost to join the conversation.