Like a living organism, music has undergone a vast evolution through time. As new instruments are invented and sound preferences change, music branches out and distinct genres and sub-genres develop like new species in a phylogenetic tree. Therefore the question arises: can musical diversity be quantified and classified to find common ancestors as in ecology? Music is mathematical – sound vibrations are wave functions – its numerous attributes such as rhythm, patterns, and time signature are all measurable. Why settle on qualitative explanations to musical diversity when, similar to traits in a genome, these attributes can be used to find common structural links to determine musical ancestry and mutations over time?
Qualitative, attribute-based music classification is inconsistent and subjective. The different types and depth of class labels can vary from individual to individual depending on musical preferences. For example, someone who listens to a variety of electronic dance music might distinguish between Drum & Bass and UK Garage, while someone who prefers pop might not. Genres can also become unclear if an artist crosses between them from song to song – one album does not necessarily fall under a single genre. Although Beyoncé’s Lemonade is officially (by Apple Inc. standards) considered pop music, I would argue that “Daddy Lessons” unmistakably falls under the country umbrella.
Several advances have been made in an attempt to ameliorate this confusion and normalize the data for music information retrieval. At the forefront of this movement was the Music Genome Project (MGP), a research group started in 2000 and led by Will Glaser, Tim Westergren and Jon Kraft with the aim of organizing songs using a mathematical algorithm. So far, the MGP has subdivided the musical Genome into five secondary genomes – Pop/Rock, Hip-Hop/Electronica, Jazz, World Music, and Classical Music.
In response to this development, researchers at the School of Electrical and Computer Engineering at Georgia Tech University published a paper in 2007 describing a new, more consistent music labeling procedure. This new set of labels disposed of commonly-known genres and instead used descriptors from the Music Genome Project that were based on acoustic features such as rhythm, timbre, tonality, and song structure. As a result, they obtained a list of 375 words that can be used to describe and classify music according to acoustic attributes. Although this was a noble contribution to the characterization of music, “extensive vamping” and “triple note feel” seem difficult to translate into an evolutionary tree.
Thankfully Dr. Matthias Mauch of Queen Mary University in London took a simpler approach to measure musical evolution in his 2015 paper “The Evolution of Popular Music: USA 1960-2010”. Expanding on the work of the researchers at Georgia Tech, he used statistical tools to quantify some of the aforementioned musical attributes – chords, harmony, timbre, and instrumentation – in songs from fifty years of Hot 100 Billboard Charts in the United States, ranging from 1962 to 2010. Faced with a much smaller sample of music history, he then grouped similar sounds into clusters with distinguishing descriptors. For example, the cluster titled “drums, aggressive, percussive” characterizes music popular in the 1980’s such as the Pet Shop Boys or Robert Palmer. When plotted graphically over time, the frequency of this descriptor being applicable to popular music peaks between 1980 and 1990, correlating exactly with the heyday of drum machine samplers and synthesizers.
Although there is no evidence of natural selection to drive the fluctuations in music diversity (maybe apart from Hot 100 Billboard Charts), Dr. Mauch’s method clearly illustrated significant peaks in musical evolution. The data showed faster rates of sound diversity increase in the years of 1964, 1983, and 1991, or the peak years of the British Invasion, the 80’s, and the rise of hip-hop. It is therefore reasonable to assume that measures of ecological evolution, although not ideal, can be used to find common structural links in determining the history of music.