SAM - Sarcasm Authentication Model

Inspiration

One of our members has Autism Spectrum Disorder and as someone who lived with Autism, social cues came pretty hard to him. One of the biggest struggles he faced was in regards to understanding sarcasm from his peers and teachers, which sometimes required him to ask follow-up questions that ended up annoying those around him. His desire to create a tool that could help him with this challenge and help those like him is what gave inspiration for this project.

What it does

Detects sarcasm in users' conversations by analyzing real-time speech-to-text and facial expression recognition and provides a confidence score for its accuracy.

How we built it

We developed the project using Python, leveraging the Tkinter library for the user interface, the ChatGPT API for natural language processing, and Hugging Face for AI model training and testing.

Challenges we ran into

Machine learning is challenging... no kidding.

Initially, one of our team members developed a custom machine learning model that trained on a Twitter dataset containing sarcastic and non-sarcastic comments. The goal was to create a predictor capable of identifying sarcasm based on the data. However, we faced difficulties integrating this model into the larger project, as an result of this, we turned to OpenAI for further exploration.

What we learned

To thoroughly plan the project and create a more detailed timeline and strategy.

What's next for SAM - Sarcasm Authentication Model

Our goals for SAM is far from over!

Plans for the future is to develop a facial recognition model created through a trained machine learning system that can read based off of a trained model of sarcastic faces and non-sarcastic facial expressions simultaneously with a tone analysis model and our working speech-to-text system to ultimately have a sarcasm detector that utilizes multi-modal with context.