The biggest hit songs have increasingly simple and repetitive melodies

They no longer make songs like they used to. And this is not just being said out of nostalgia by the musical critics who lived through the musical revolutions of the 1960s and 1970s. A study that analyzed the biggest musical hits in the United States, between 1950 and 2022, has confirmed this, using scientific evidence. The researchers concluded that the melodies of the most popular songs are increasingly simple and repetitive. However, they clarify that this does not mean that today’s musical hits are worse or poorer musically, but rather that the complexity and richness in elements other than the melody are less evident.

For the public, melodies are what stand out most in pop music, but science has done little research into the ingredients that determine its success. Madeline Hamilton, lead author of new research published Thursday in the journal Scientific Reports, wanted to devote her doctoral dissertation to this issue. “I would argue that the melody, especially the lead vocal melody, is the most significant dimension of a song. It’s usually what we remember about it and what you sing or whistle when you’re listening to it — not the bass line or the drum beat,” says Hamilton. So she thought she should start with the melody when she set out to investigate “why we like the music we like, and how we might predict whether a person is going to enjoy a particular piece of music.”

This led to a “happy accident,” says the researcher. When she began to explore the melodies and analyze them, she first observed that over the years the number of musical notes per second was increasing. And then she discovered something she wasn’t looking for: she saw that the parameters she used to measure the complexity of melodies had been decreasing since the 1950s. In that decade, songs such as Domenico Modugno’s Nel blu dipinto di blu (Volare), which topped the chart in the United States, were triumphing.

Hamilton wanted to confirm her chance discovery and carried out a more in-depth analysis, together with Marcus Pearce, her supervisor at the music cognition laboratory at Queen Mary University of London. The first big challenge was to build their own catalog of high-quality and detailed musical transcriptions. Until then, there had not been one large enough, and that greatly limited the quantitative study of the melodies.

Most previous research in computer science applied to music had been done directly from audio cuts of songs, which reduced the study to very basic elements, such as the tempo of the songs or the timbre of the instruments. The analysis of these characteristics is now fairly automated and, with the help of artificial intelligence, is used by streaming services — such as Spotify, Tidal or Apple Music — to suggest songs similar to the ones we listen to.

A very musical lockdown

In order to go further and scientifically study something more abstract, such as melodies, Hamilton had to carry out the meticulous task of manually transcribing the vocal part of the songs. She did this during the first year of the Covid-19 pandemic, coinciding with the lockdown in London. “For two months, I spent about 10 hours a day to this task that requires intense concentration and attention to detail. At that time, it’s not like I had anything better to do than sit in my room transcribing music,” the researcher recalls.

After six months, she finally finished recording the notes and rhythmic structure of the most popular melodies of the last seven decades. Her catalog includes the five biggest hits of each year, according to the American Billboard Hot 100, which is based on record sales, radio station plays and streaming. She then polished and expanded the catalog, which is available online for public use and now covers from 1950 to 2022. It contains a total of 366 songs, with their melodic fragments encoded in more than 1,000 digital files in MIDI format.

Hamilton and Pearce subjected all this material to a statistical analysis with algorithms for the detection of change points. They settled on eight parameters that allow the melodies to be described with data. And through sophisticated computer calculations, they were able to confirm their initial discovery that melodies have become increasingly simple over the last seven decades. What’s more, the researchers observed that — unlike what has been seen in previously studied elements, such as the harmony and timbre of the instruments used — in the evolution of melodies, there are no cycles in which trends come and ago. Instead, the fall in complexity has been a constant.

In their article, the researchers also point out that there were two very pronounced drops in melodic complexity around 1975 and 2000. The authors attribute this to the influence of new styles such as disco music and hip hop, respectively. The researchers also found that in the last two decades, there has been a significant increase in melodic repetitions within songs, which they relate to loops. These sound loops were first characteristic of rap, and today they have become widespread in pop music.

But the researchers stress that the scientific evidence does not mean that today’s greatest hits are worse than the hits of the past. When explaining this shift, the study argues: “The decrease in the complexity of melodies may be associated with aspects of the modern predicament.”

In other words, it may be an evolutionary adaptation to a musical world in which faster songs — more notes per second —, with multiple vocal layers and instruments, and high-quality sound, are topping the charts. In order to avoid overwhelming the listeners, musicians may be choosing to simplify the melodies. “For example, the melody features small pitch intervals, a limited range of pitches, and lots of repetition,” the study states. This is evident in songs like Bad Guy, by Billie Eilish, which reached No. 1 in the U.S. in 2019.

However, the researchers recognize that more than just the top five hits of each year need to be analyzed. Hamilton is already working on expanding their catalog of the most successful melodies and is confident that, with the current momentum in generative artificial intelligence, “in a few years, we will be able to make automatic transcriptions of the melodies while maintaining high quality.” With the help of AI, she could provide more complete scientific interpretations of the evolution of music and, perhaps, return to her original idea of discovering why we like the music we like.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition