NOTE: Sci is usually at SciAm Blogs on Monday, but they're doing some updates to the site, and I thought I would stay over here for today, just in case.
I have an officemate (and I know they’re reading, everyone say hi to my awesome officemate!) who likes to listen to pop music in the lab. One time, the latest David Guetta (and various artists) was on, and we were working away like good little labrats. And then I made my officemate stop the music. I made them go back, and skip through some of the songs, playing about the first 20 seconds of each.
…they all sounded awfully alike. If you didn’t know what track number you were on, you’d never be able to tell whether you were about to hear Niki Minaj or Snoop Dogg within the first 10 seconds. The opening beats were similar, the opening chords were similar.
But we kept on chair-dancing along. I mean, it’s fun music.
But it did make me think. Is it just me, or does a lot of modern pop music really sound…the same? Or more alike, anyway, than in the past?
And luckily for me, a study came out that answers my question. It’s not just me and being old (get offa my lawn!) pop music really is becoming more…the same.
Serra et al. “Measuring the Evolution of Contemporary Western Popular Music” Scientific Reports, 2012.
It’s not just pop music, most country music sounds like other country music. And it’s not just limited to American music, a lot of other pop music from other countries sounds very similar as well. Most K-Pop sounds like most other K-Pop, and Indian pop music sounds a lot like other Indian Pop music. Heck, even a lot of indie music sounds similar.
But here's the question: what MAKES them similar? Is pop music really becoming more similar over time, and if so, how exactly? What is different about it from older music?
To look at this question, the authors of this study went to a huge database of music, the million song dataset. The dataset contains songs and descriptions from the past 55 years (more than 1,200 DAYS of listening, I really hope the authors didn't have to do that...). The music is of various genres (pop, rock, metal, hip hop, etc), and each is decribed in terms of three specific areas: loudness, pitch, and timbre. Loudness is not the ACTUAL loudness (I hear we can adjust that these days), but rather, subjective loudness, a PERCEPTION of sound amplitude. For example, this song:
Will be a lot louder in our perception than this one:
No matter what the volume.
Pitch and timbre are easier to quantify. Pitch is the tonal structure of the song, what key it's in, how the chords progress, what the melody is like, the basic "sound" of the piece. Timbre, on the other hand, is is the "color" of the piece, what it FEELS like, the difference between electric and acoustic guitar, between autotune and basic recording.
The analysis of these three metrics, looking at thousands of songs over time, is pretty complicated. The authors built up a "vocabulary" of musical elements, different types of pitch, timbre, and loudness. The combinations of these three elements will change beat to beat within a song, so they analyzed over 1,000,000 of the vocabulary phrases that were consecutive across several beats in the song (meaning more consistent than something occurring just once). They looked for patterns within the groupings of vocabulary phrases for each song.
And what they found was a lot of similarity. It turns out that there are specific combinations of loudness, pitch, and timbre which are REALLY persistent across time. There are certain combinations that clearly just sound "good" to us. And when you think about it, this makes a lot of sense, in many different genres. Some types of phrases and some aspects of music will persist over time, from classical to rock.
But what about the way music has changed? Well it turns out, the more things change, the more they stay the same!
What you can see above are various measures of the metrics used. The first shows similarity in pitch distribution from 1965, 1985, and 2005. The z score on the bottom is a measure of homogeneity, or how similar the pitches are in the songs, and you can see that the similarity of pitch has become more common over time. This help up for loudness, pitch, and timbre.
So what the authors concluded was that some aspects of music stay the same over time, but other aspects change, and it's these aspects that make music sound "new". But what's weird is that the aspects that change...well they actually become more similar over time. Music gets louder, the tunes get more restricted, and the timbre becomes the same old mix of electric guitar and synth. This homogenization, the increasing sameness, is actually what makes the music sound "new". And the authors argue that a song from 1955, if it had the right loudness, pitch, and timbre, could pass for a brand new shiny song, if it fulfilled the right requirement: sounding exactly the same as other current songs.
I should point out, they did study ONLY the pop songs in this database. I do wonder if the same holds true for other genres (and I strongly suspect that it does).
And it makes me wonder. Will we ever approach the pop music singularity? Does this mean that the one most repetitive, with the least variety of pitch progression, the most loud, the most boring, the most PERFECT pop song really could exist?
…is it this one?
Joan Serrà, Álvaro Corral, Marián Boguñá, Martín Haro, & Josep Lluis Arcos (2012). Measuring the evolution of contemporary western popular music Scientific Reports 2, 521 (2012) arXiv: 1205.5651v1