Hype Classifier

A PyTorch-based binary classification model that predicts whether a song is "hype" or not based on Spotify's audio features. This model was trained on Spotify audio feature data recorded April 2019 and prior: https://www.kaggle.com/datasets/tomigelo/spotify-audio-features/data

Overview

This model makes use of Spotify's audio features (energy, danceability, valence, and tempo) to classify songs into two categories:

1. Hype: High-energy, danceable tracks
2. Not Hype: Lower-energy, more relaxed tracks

Features

• Built with PyTorch for flexible model architecture
• Binary classification with sigmoid output
• Input: 9 normalized audio features
• Hidden layers with ReLU activation
• Dropout for regularization
• Sigmoid output for binary classification

Audio Features Used

Audio features used and how Spotify describes them:

1. Energy: Intensity and power of the track
2. Danceability: How suitable the track is for dancing
3. Valence: Musical positivity/happiness
4. Tempo: Speed of the track (BPM)