HearMe: Classifying Emotions from Audio

• Model: CNN trained on 300+ audio samples for classifying 5 emotional categories using MFE features.
• Toolchain: Used Edge Impulse for dataset collection, model training, and model deployment.
• App: Integrated the model into a Streamlit web app for real-time audio prediction with < 2-second latency.
• Tech Integration: Parsed serial audio data, labeled emotions, and updated interface on the fly.