In today’s day and age, we’re seeing more crossover than ever between musical artists of different genres. This project builds a model which predicts a song's genre based solely on its lyrical content.

A full description of the project can be found at saisenberg.com.

• Python (suggested install through Anaconda)

• R

• Python:

  • bs4, numpy, pandas, re, requests, sklearn, string, warnings (all installed with Anaconda)
  • json (!pip install json)
  • nltk (!pip install nltk)
  • xgboost (!pip install xgboost)

• R:

lib <- c('dplyr', 'geniusR', 'jsonlite', 'lubridate', 'stringr')
install_packages(lib)

This code scrapes Billboard, Ranker, and TheTopTens for artists of different genres. Any duplicate artists are removed as appropriate.

The output of /python/artist_collection.ipynb can also be found at /data/json_genres.json.

This program scrapes and cleans lyrics from Genius, categorizing results by genre. Visit Genius to view or obtain a Genius client access token.

The output of /r/genius_scraper.R can also be found at /data/lyrics.csv.

This code preprocesses all lyrics for modeling, and runs Naïve Bayes, support vector machine, and gradient boosting models to predict a song's genre from its lyrics.

• Sam Isenberg - saisenberg.com | github.com/saisenberg

This project is licensed under the MIT License - see the LICENSE.md file for details.

• Billboard
• Ewen Henderson
• Genius
• Ranker
• TheTopTens