A team of British scientists reckon they can predict the success of a pop song with abut 60 percent accuracy using a machine learning algorithm.
The University of Bristol team looked at the official top 40 singles chart over the past 50 years, and evaluated musical features such as tempo, time signature, song duration and loudness. They also computed more detailed summaries of the songs such as harmonic simplicity, how simple the chord sequence is, and non-harmonicity – how ‘noisy’ the song is.
They then came up with a ‘hit potential’ equation that scores a song according to its audio features. They found they could classify a song as either a hit or a non-hit based on this score, with an accuracy rate of 60 per cent as to whether a song will make it to top five or remain below position 30.
“Musical tastes evolve, which means our hit potential equation needs to evolve as well,” says Dr Tijl De Bie, senior lecturer in artificial intelligence.
“Indeed, we have found the hit potential of a song depends on the era. This may be due to the varying dominant music style, culture and environment.”
Before the nineteen-eighties, for example,the danceability of a song wasn’t that relevant to its hit potential. Since then, though, danceable songs have been more likely to become a hit.
The team’s found its algorithm is more accurate for some eras than others.It’s been particularly difficult to predict hits around 1980, they say, with the equation performing best for the first half of the nineties and since the year 2000.
This suggests that the late seventies and early eighties were particularly creative and innovative periods of pop music, they say.
These days, though – and rather depressingly – it appears that people are looking for loudness above all else, with all songs on the chart becoming louder – especially those that make it to the top.