EmotionalCounselor

Inspiration

Frustration about not finding parking anywhere near the HackNC place led us to think of ways to detect strong emotions and have a system to offer calming support. We also wanted to learn about natural language processing, so this seemed like a good idea.

What it does

It takes several recordings of a person throughout the day and performs emotion recognition with NLP techniques. Then it can provide some basic statistics about the user's mood during the day.

How we built it

We started piecing together different parts. One of us handled the speech to text part using Assembly AI's API. Then the other wrote the part about detecting emotion from text and putting it all together.

Challenges we ran into

We did not manage to finish a basic prototype. We wanted to do emotion detection on the audio waves (to detect things like tone, etc) as well as the text using our own custom-trained machine learning model. However, due to time constraints we decided to use off-the-shelve models that can provide reasonable emotion detection. Additionally, we are both inexperienced in front end stuff, so we struggled with coming up with a GUI.

Accomplishments that we're proud of

Getting some rudimentary type of emotion detection working at a decent level some of the time.

What we learned

How to use tools like Assembly AI's speech to text system and the basics of natural language processing.

What's next for EmotionalCounselor

Do emotion recognition with an actual classifier that has been trained on a larger amount of data. Also to provide a front end with all the user-facing functions we wanted to offer in the first round.