Music and I go way back. As a toddler, I loved singing along to popular songs using my pink brush as a "microphone". My voice ended up being more off-key than my 4-year-old self presumed, but my musical journey didn't end there: I started playing the flute and the violin at 7, joined an orchestra and choir for several years, and picked up the piano later on.
I don't play much (or particularly well) anymore, but I still love music. I listen to it every day and have felt its power to deeply affect my mood. However, I'm not the only one. Music is one of those rare things that transcend social, geographical and economical barriers: everyone seems to like it, even if we all have different preferences regarding tempo, style, and so on, and it has become embedded in the culture, history and daily life of modern society.
I decided to combine my passion for analytics with my enthusiasm for music to analyze how the latter has changed over the past seven decades and how its evolution is connected to social trends and developments.
I owe a debt of gratitude to Spotify and Genius, whose APIs I used to gather all the data! I took the songs from Spotify's "All Out" playlists, which feature the 150 "biggest songs" for every decade from the 1950s to the 2010s, continued using the Spotify API to gather track features, and moved on to the Genius API to obtain the lyrics for every song. In the process, I found some bizarre results - instead of lyrics, some tracks came out with excerpts from Game of Thrones scenes, Arabian Nights chapters, and the like, prompting a protracted yet somewhat amusing data cleaning operation.
Song Features
Songs have a lot of attributes we generally don't think about. These include features like loudness, valence (how positive a song is) and acousticness. How have they changed over time? What do these changes tell us about societal and cultural trends? These are the questions I aim to answer on the first part of the analysis.
Loudness
Music has been getting louder, particularly since the 1990s: the average loudness went from -11.1 dB in the 1950s to -8.4 in the 1990s to -5.9 in the 2010s. From the 90s to now, the average song has therefore almost doubled its sound energy. In terms of hearable loudness, this represents an increase of about 20%.
Note: every dot represents a track and the fuchsia and pale pink lines show, respectively, the yearly and decade averages.
According to a study published in the Journal of General Internal Medicine, this increase is related to the progressive improvements in audio engineering, as they have allowed for quieter spots within tracks, therefore making them louder. Technology has also changed the way we listen to music, and improvements in hardware and software have made devices like stereos and headphones readily available to millions of users.
Prolonged exposure to loud music, in turn, is a risk factor for noise-induced loss hearing (NIHL), which currently has no cure. The World Health Organization estimates that 1.1 billion young people worldwide could be at risk of NIHL due to unsafe listening practices.
Valence
Valence, defined as the musical positiveness conveyed by a track, has decreased over time, going from an average value of 0.61 in the 1950s to 0.49 in the 2010s. While there are many plausible reasons behind this, experts have been unable to come up with any conclusive explanations.
Duration
Track length increased until about the 1990s and has been decreasing ever since. The average song in the 2010s is 44.6 seconds shorter than its 1990s counterpart, which is attributed to two factors: the reduction of the human attention span and the rise to power of streaming platforms.
A strong suspect behind the reduction of our attention span is the overabundance of information of our day and age: having so much information constantly bombarding us exhausts our attention and causes us to continuously look for newness because we fear missing out (the so-called FOMO). As for popular music, so much content being generated through so many channels has led to the emergence of a "skipping culture", and the music industry has responded by making songs shorter.
Additionally, the emergence of players like Apple Music and Spotify has rewritten the rules of commercial music. In the past, artists used to get paid for physical sales of albums or singles but today streaming platforms pay on a per-stream basis. However, consumers don't have to stay tuned the entire time: a stream counts after users hit the 30 second mark listening to the track. This new model has incentivized artists to make shorter songs to increase their chances of keeping consumers' attention and obtaining higher revenues.
Acousticness
Acousticness represents how much sound in a song is produced through acoustic means (for example, with a violin or a guitar) as opposed to electronic means. Music is becoming less acoustic and turning more electronic instead.
According to the Echo Nest, a music intelligence data platform for developers and media companies, this trend mirrors technology's integration into the greater society. The Echo Nest has also found that songs are becoming more rhythmically precise and artificial sounding.
Song Features From the 1950s to Now
So louder, shorter, less positive, and less acoustic - anything else? Actually, yes. In the next part, I explore additional song attributes (like song popularity and artist) and their evolution and relationship to social and cultural developments.
Comments