William Gao

Predicting song popularity from lyrical content using machine learning

Started Feb. 6, 2021

Abstract or project description

In this project, William will source a dataset of song lyrics and popularity metrics for 100-10,000 songs (the more the better). He will write an algorithm to parse this data and convert it into machine-interpretable form. He will then implement, compare, and contrast various machine learning models in a regression task using lyrical content to predict popularity. Finally, William will evaluate performance of the ML models and implement some creative application of his choice, like song generation, transfer learning to a different domain, etc.