Can big data really predict what makes a song popular?

Music is part of our lives in many ways. I hear it on my commute and it echoes in shopping centers. Some of us seek live music at concerts, festivals and shows, and rely on music to set the tone and mood of our days.

While you may understand what genres and songs you like, it’s not exactly clear why certain songs are more appealing and popular. Perhaps the lyrics speak of experiences? Perhaps energy makes it attractive? These questions are important to music industry professionals, and analyzing data is an important part of it.

At Carleton University, a group of data science researchers sought to answer the following questions:

music industry income

Revenue in the music industry comes from two sources that are influenced by many factors: live music and recorded music. During the pandemic, cancellations of in-person performances reduced live music revenues, but streaming revenues increased.

With the growth of digital platforms such as Spotify and TikTok, the majority of music revenue now comes from digital media, primarily music streaming. How and how this revenue reaches singers and songwriters in general is another matter.

Popularity on digital platforms

A song’s popularity on digital platforms is considered a measure of the revenue that song generates.

So producers try to answer questions like, “How can I make a song more popular?” And “What are the characteristics of the songs that entered the top charts?”

YouTube, Twitter, TikTok, Spotify, Billboard (Billboard Hot-100, by data researchers We have created a systematic study that collects data from as “Billboard hot top” or “Billboard Top-100” in our work and others).

We linked datasets from different platforms with Spotify’s acoustic descriptive metrics, or “descriptive features” of songs. These features are derived from a dataset that generated categories for measuring and analyzing song quality. Spotify’s metrics capture descriptive features such as: acoustic, energy, Danceability When instrumentality (a collection of instruments and voices for a particular song).

We tried to find trends and analyze the relationship between song characteristics and popularity.

Weekly Billboard Hot-100 rankings are based on US sales, online streams and radio play.

Our analysis of Spotify and Billboard reveals useful insights for the music industry.

What predicts Billboard hits?

To perform this research, we used two different data sets on Billboard hits from the early 1940s to 2020 and Spotify data related to over 600,000 tracks and over 1 million artists.

Interestingly, no substantial correlation was found between the number of weeks a song remained on charts, a measure of popularity, and the sonic characteristics included in the study.

Our analysis found that newer songs tend to stay on the charts longer, and that a song’s popularity affects how long it stays on the charts.

In a related study, researchers collected data on Billboard’s Hot 100 from 1958 to 2013, tempo When Danceability They often get higher peak positions on the Billboard charts.

Predict Spotify song popularity

We also used song features to generate a machine learning model to predict song popularity on Spotify. Preliminary results concluded that the features were not linearly correlated. energy.

This shows the Spotify metrics we investigated. acoustic, Danceability, interval, energy, explicitness, instrumentality, vitality, Chatting (a measure of whether spoken language is present in a song), tempo When Release year —wasn’t a strong predictor of a song’s popularity.

The majority of songs in the Spotify dataset were not listed as explicit. instrumentality When Chattingand usually were recent songs.

While one might think that some inherent characteristics of a particular song make it popular, our research suggests that popularity can only be attributed to quantifiable acoustic factors. It became clear that it was nothing.

This means that song creators and consumers must consider other contextual factors beyond the music’s characteristics, captured by Spotify’s measurable data, that may contribute to a song’s success. means

Factors Affecting Changes in Popularity

Our research confirms that the factors that influence a song’s popularity change over time and need to be continuously investigated.

For example, among songs produced between 1985 and 2015 in the UK, songs produced by female artists were more successful.

Other aspects can contribute significantly to a song’s success. Data scientists suggest lyrical simplicity, advertising, and distribution plans as potential factors predicting a song’s popularity.

READ MORE: Beatles ‘Get Back’ documentary reveals creativity doesn’t occur naturally

Attachment listener

Many musicians and producers use popular events and marketing tactics to promote their songs. Such events create social engagement and audience engagement, connecting listeners to the song being played.

For the general public, following a long lockdown, live music events are an opportunity to reunite with friends and enjoy live artistry and entertainment.

While attending a music event or listening to a song, think about what you enjoy about that song.