Serenity

Inspiration

Inspired by the prevalent need for products fighting mental health issues, paired with our group members having also experienced stress and anxiety within our own lives, we decided to build Serenity in hopes of mitigating this large problem.

What it does

Serenity allows a user to have a conversation with a mental health companion, otherwise known as Serenity. This AI Chat Bot is backed by data from scientific studies to map the user's mood, determined through conversation, to a background paired with audio customized solely based on the users mood. In addition to having a personalized user UI, users can use both the text to speech and voice to text features to simulate a real life conversation, through verbal contact.

How we built it

We used HTML & CSS for the frontend, while using GroqCloud's API for text to speech, and Open AI's API for our chatbot. Additionally, we utilized scientific studies to perfect the mapping system of a user's conversations to mood, to allow optimal color and composition to be displayed in the background. We deployed our website with Vercel.

Challenges we ran into

A challenge we ran into was the issue of a user losing all their data and conversations if they ever left the website, so in order to deal with this, we stored localized memory within the users device, encompassing a short summary of the prior conversation, to allow the AI to recall past messages if the website is opened again.

Accomplishments that we're proud of

An accomplishment we are proud of was the ability to come up with the unique idea of a changing background backed by data, as well as the ability to seamlessly integrate multiple APIs within our project.

What we learned

We learned quite a lot, and specifically gained a lot more experience within text to speech and voice to text features. We also gained more experience within HTML and CSS through this project.

What's next for Serenity

Our future goals with Serenity are to train our own LLM to allow for a stronger mapping system, increasing the accuracy for the detection of the mood of the user. We also hope to upscale this project and gain more input for our product through beta testing.