Inspiration
We've all been there. You're on a call with friends planning where you want to go for the weekend, or meeting up with your classmates for a group project. You think everything is settled, but the day of, nobody remembers when or where to meet up. Nobody is taking notes on what everyone is saying or the key points of the conversation. Thus, combining speech to text apis and auto tl;dr algorithms, we created an automated solution to note taking during a group conference call.
How does it work
The basic flow is this: a user starts a video conference with a unique room code that they think of. Then, they can invite their friends through AT&T's SMS API, and our application will automatically format an invite message to a given phone number. Once everyone is in the call, they can begin talking while a transcript of their chat is generated in real time in a sidebar. When they are ready, they may generate a short summary of the transcript as a conclusion to their conference call.
Additionally, we have added a "vote" command. When someone says "vote," the application will keep track of the words everybody says and count them as a vote for a certain item. It will then tally up the votes and display the top 3 items on the transcript.
How we built it
We used React.js as a front end framework and Express.js and socket.io on our backend. We utilized the Web Speech API for speech to text and webrtc for our video call. We also used AT&T's SMS API to allow users to easily invite their friends to their conference call through a text message in the browser. We deployed our webapp through Microsoft Azure.
What's next for AutoScribe
We're looking to add more features to our application that can be triggered through voice commands like the "vote" trigger, for example, a "lookup events" trigger that will show any fun events or activities for a group to do when planning out a fun weekend.
Log in or sign up for Devpost to join the conversation.