The aim of this project is to automate the production of songs within the Eurobeat genre. This will save Eurobeat artists roughly 2000 hours of work per year, without any noticeable drop in song quality.

This project depends on Beautiful Soup 4, Requests and the Natural Language Toolkit To install all dependencies, simply type into a terminal: pip install beautifulsoup4 requests nltk

Runs all of the modules below in sequential order. This will generate a new song if everything goes well.

LyricScraper scrapes the eurobeat-prime.com lyrics database and stores information in SongData objects. A list of all SongData objects is returned for further processing. If the module is called in a stand-along fashion, then the array gets pickled with the default python pickling library and stored to songs.pkl.

LyricParse deals with measuring the frequency of grammar components of the input songs. It measures the number of paragraphs, lines per paragraph and words per line in each song. Also, it determines the frequency of a given part of speech following another, the frequency of a part of speech starting a line, and the frequency of a word being a specific part of speech. If this module is run alone then these results are stored in grammar.pkl.

LyricGenerate takes the previously generated grammar object and uses it to create the lyrics of a new song. It randomly samples multiple distributions to do so. The distributions determine number of paragraphs, lines, and words as well as what part of speech and what word to include in the song. This doesn't produce a very good result, and will require further work to produce acceptable lyrics.

Other files are documented in their docstrings.