Harmony and Form in Pop/Rock Music (the previous version of Computational Music Analysis) was a hybrid (in-person and online), interdisciplinary, vertically integrated (faculty, graduate students, and undergraduate students working together), project-based course.
The course explored harmonic and formal structures in pop/rock music of the late 20th and early 21st centuries, with the overarching goal of discovering, creating, and disseminating new knowledge about this music through a collaborative research project. The course also provided supporting instruction in identifying harmonic and formal structures by ear, representing those structures using the standard symbols and terms of music theory, encoding those structures digitally, performing computational analysis of those digitally encoded structures, and using digital tools for online collaboration.
Following are core resources from the 2014 course:
At the end of the semester, students wrote up summaries of the work that we did creating software and using it to analyze music. If you’re interested in exploring the tools we developed or reading the results of our work, check out the Billboard Corpus Analysis, the MatLab Chord Recognition, or the Sonic Visualizer/Chordino repositories on GitHub, or read the following posts:
- The Parser, Andrew Mahan & Diego Escalante: A description of the software written to parse the McGill Billboard data set, with a discussion of some of the musical complexities and problems encountered in the data and the parsing process.
- Analysis of Phrase & Harmony Results, Brianne Borden & Tyler Honsel: A preliminary explanation of the analysis of phrase position on the probability of occurrence of particular chords and chord progressions.
- Clustering, Chris Rooney: A description of the process followed to perform a cluster analysis on the McGill Billboard Corpus, in order to find possible (sub-)styles and discrete harmonic practices within the Billboard pop/rock corpus.
- Understanding Cluster Data, Kyle Rooney: A preliminary analysis of one finding from the cluster analysis, with instructions on how to read the data from the other analyses.
- Viterbi Tagging Results, Aaron Davis: “Given a sequence of harmonic roots (roman numerals), can we automatically determine what formal module these roots belong to?” (using a Viterbi, Hidden Markov Model algorithm)
- Using DSP (Digital Signal Processing) for Chord Recognition in Pop/Rock Music, Alex Uribe, Erik Kierstead & J.R. Souders: A comparative analysis of existing tools for audio signal processing, with results and discussion of an in-depth analysis of a small test corpus using Sonic Visualizer & Chordino.
- Automated chord recognition using Machine Learning techniques: Part 1 | Part 2, Esther Vasiete: A comparative analysis of several MatLab-based tools for audio signal processing, with results and discussion of an in-depth analysis of a small test corpus using several high-performing toolkits.
Unless otherwise stated, all results materials are the sole property of the authors, copyright 2014, reproduced here by permission.