MoodBot

Inspiration

During a time like Covid, I tend to call my friends and loved ones often. However, during calls there are always misinterpreted messages and emotions. I wanted to better analyze these emotions to better support my friends during a time where mental health is at it's lowest. I believe that knowing your closed one's emotions can better benefit both of you when it comes down to solving the mental health pandemic on top of the Covid pandemic!

What it does

An audio recording is analyzed using sentiment analysis to determine how the speaker(s) from the audio recording are feeling. Data such as sentiment, opinions, and emotions is displayed for the user to determine the feelings of the speaker(s).

How I built it

When a user uploads an audio recording, the audio recording is stored in our AWS S3 Bucket. From there, once the submit button is clicked, it is then run through 3 different APIs. First, Azure's Speech to Text API converts the audio to text. Once the audio is converted to text, the text is then run through Azure's Sentiment Analysis API and also IBM Watson's Tone Analyzer. This gives us responses containing data about sentiment and emotions, which is then served to the frontend React server.