Meeting Buddy

Dashboard
Record Meeting
Sending Data to Server
Response from server with deepgram output
Meeting Archive
Detail of a older meeting

Inspiration

The inspiration of this project is to reduce the human effort in summarising the meetings. It also reduces human errors such as forgetfulness as well as impartiality in decision making.

What it does

This web app collects the audio from the meeting presentation and generates the summary of the meeting eventually sending the generated summary to all the attendees of the meeting through the mail.

How we built it

We followed the below steps to build this application:-

We used React-mice to record audio of the meeting conversation and recorded audio is processed on the Node.js server with Deepgram API to convert it to text format
Then Node.js is passing that generated text to python API.In the generated if there is any error there is an option with that user can edit that
In Python API we are using Hugging Face Transformers to summarize that text. 4 Then we are using Twilio API to mail that summarize text to all attendees of the meeting automatically. ## Challenges we ran into
Implementing Deepgram with Node.js was a bit challenging but rich documentation and workshop made it easy for us.
To deploy a model in google-colab using fastapi was also bit challenging for us. ## Accomplishments that we're proud of We can say that we implemented our thoughts in form of a web application in the given period of time. This task was challenging as well as exciting to accomplish for the team ## What we learned
We learned how to work in a team as well as time management during this hackathon.
We comprehended the ability to work under pressure effectively. ## What's next for Meeting Buddy In next we will do the Sentimental Analysis of the recorded audio during the meetings. We also wish to use the Zoom and Google meet API to fetch the audio transcripts to summarise the meetings.In future we plan to add authentication so multiple teams can use this web app parallel.