Computational Music Analysis

MUSC 5151, CSCI 7000, CSCI 4830
CU–Boulder, May 2016
Kris Shaffer, Ph.D. – instructor

Getting started

Welcome to Computational Music Analysis! Please fill out this pre-course survey and read the following in preparation for our first meeting on Monday, May 9.

You also may be interested in seeing the materials and project results from a previous offering of this course in May 2014.

May 9: Introductions

Just what is computational musicology? How does it relate to the digital humanities? to data science? Today, we'll get a broad overview of the discipline (slides here), as well as get to know each other and some of the tools we'll be using. We'll also work through the following reading together:

May 10: Building and breaking

Are the digital humanities about "building stuff" or "breaking stuff"? Hack or yack? big data? sharing? social justice? We'll read about and unpack these questions as we explore the nature of (and some of the debates about) the digital humanities.

Before class please read/watch the following, and then add a comment in our Slack discussion responding to the question, "What is/are the digital humanities?":

May 11: Musical building blocks

What are the basic building blocks in music? We'll explore some of the basics: rhythm and meter, pitches, and chords. No in-depth music theory is necessary, but a basic familiarity will help a lot. We'll also look at one study in computational musicology that explores chords and chord progressions in pop/rock songs.

Before class please read/watch the following, and then add a comment or question in our Slack discussion responding to one of the readings/videos:

May 12: Musical data as musical style

Musical style is a cultural practice, but it can also (to an extent) be represented via data. What is the relationship between data, statistics, and computation on one hand, and musical style on the other?

Before class please read/watch the resources below. Then leave two comments in Slack: one (in #readingresponses) with a question/comment/summation of the readings, and one (in #projectideas) with an idea for a project to follow-up on deClercq/Temperley.

May 13: The McGill Billboard Data Set

We'll end our first week looking at two more advanced studies involving the harmonic structures of songs in the McGill Billboard Data Set.

Before class please read/watch the following, then leave a question or insight for discussion in #readingresponses on Slack:

  • Christopher William White, gloss on Hidden Markov Modeling in musical studies
  • Ashley Buygoyne, Stochastic processes and database-driven musicology, Chapters 4–5, but only the following pages
    • p. 123 through the end of the first partial paragraph on p. 146
    • last paragraph of p. 147 through the end of the first partial paragraph on p. 148
    • last paragraph of p. 152 through Table 4.3 on p. 153
    • first full paragraph on p. 155 through the end of the first partial paragraph on p. 164
    • last partial paragraph on p. 172 through the first partial paragraph on p. 179
    • last paragraph of p. 185 through p. 191

May 16: Classical and folk music

Today, we'll focus on computational studies of classical and folk music, as well as the relationship between musical structures and musical expectations.

Before class please read/watch the following, then leave a question or insight for discussion in #readingresponses on Slack:

May 17: Music21, MSD, and project decisions

We finish our exploration of existing projects with an introduction to music21 and the Million Song Dataset, followed by a discussion in which we decide on our collaborative project(s).

Before class please install Python 3 and music21 on your own computers, and read/watch the following, then leave a question or insight for discussion in #readingresponses on Slack:

  • Bertin-Mahieux, et al., "The Million Song Dataset" (in our shared folder on Google Drive)