Image

Al Gorithm and His Digital Divas. Are They The Next Chart-Toppers?

Dec. 20, 2011
Bad musicians rejoice – fame and fortune could be yours. Forget natural talent, years of hard practice, sex appeal, on-stage charisma and of course well-financed media promotion, all you need now to generate a chart-topping hit song is a software package developed at Bristol University in the UK. Hendrix would turn in his grave; Elvis would have viewed it with a very suspicious mind and Lennon and McCartney would see it as just a cheap ticket to ride. So what is this software, how was it created, and how will it mass-produce the X Factor wannabes of tomorrow?  Following that I'll tell you why it will never succeed. A team of researchers led by Dr Tijl de Bieat at the University of Bristol’s Intelligent Systems Laboratory in the Faculty of Engineering studied data from the United Kingdom's top 40 music charts from the past 50 years. The idea was to differentiate the hits from the flops. Parameters included musical attributes such as tempo, rhythm, duration and loudness. The researchers also computed more detailed information about the music such as harmonics and chord sequences. From this data an algorithm was created that could rate a song relative to its audio characteristics. By studying all the UK hits and measuring their audio features the researchers were able to calculate how important each of the 23 audio parameters was. By doing so they found they could classify a song into a ‘hit’ or ‘not hit’ based on its score, with an accuracy rate of 60% as to whether a song will make it to the top five, or if it will never reach above position 30 on the UK top 40 singles chart. Additionally, the research highlighted some interesting musical trends. Apparently, before the 1980s, the danceability of a song was not very relevant to its hit potential, an assumption that the Rolling Stones would certainly challenge given the rhythms of some of their early hits. Also, in the 1980s slower musical styles such as ballads, were more likely to become a hit. Really, try telling

Bad musicians rejoice – fame and fortune could be yours. Forget natural talent, years of hard practice, sex appeal, on-stage charisma and of course well-financed media promotion, all you need now to generate a chart-topping hit song is a software package developed at Bristol University in the UK.

Hendrix would turn in his grave; Elvis would have viewed it with a very suspicious mind and Lennon and McCartney would see it as just a cheap ticket to ride.

So what is this software, how was it created, and how will it mass-produce the X Factor wannabes of tomorrow?  Following that I'll tell you why it will never succeed.

A team of researchers led by Dr Tijl de Bieat at the University of Bristol’s Intelligent Systems Laboratory in the Faculty of Engineering studied data from the United Kingdom's top 40 music charts from the past 50 years. The idea was to differentiate the hits from the flops.

Parameters included musical attributes such as tempo, rhythm, duration and loudness. The researchers also computed more detailed information about the music such as harmonics and chord sequences.

From this data an algorithm was created that could rate a song relative to its audio characteristics. By studying all the UK hits and measuring their audio features the researchers were able to calculate how important each of the 23 audio parameters was. By doing so they found they could classify a song into a ‘hit’ or ‘not hit’ based on its score, with an accuracy rate of 60% as to whether a song will make it to the top five, or if it will never reach above position 30 on the UK top 40 singles chart.

Additionally, the research highlighted some interesting musical trends.

Apparently, before the 1980s, the danceability of a song was not very relevant to its hit potential, an assumption that the Rolling Stones would certainly challenge given the rhythms of some of their early hits.

Also, in the 1980s slower musical styles such as ballads, were more likely to become a hit. Really, try telling that to Technotronic and their hit track, Pump Up The Jam.

The researchers of this "X-factory" type software do admit that the predictive accuracy of their hit potential software varies. They found it particularly difficult to predict hits around 1980. The software performed best in the first half of the nineties and from the year 2000. According to the university team this suggests that the late seventies and early eighties were particularly creative and innovative periods of pop music.  So, it would seem that the software can deal with most things except musical talent.

The research team found that until the early nineties, chart hits were typically harmonically simpler than other songs of the era. On the other hand, from the nineties onward hits more commonly had simpler binary rhythms such as four beats to the bar.

And what about current music trends? The Bristol University researchers concluded that on average all songs on today's charts are becoming louder and that hit tracks are relatively louder than the songs that languish at the bottom of the charts.

So why wont Al Gorithm and His Digital Divas be chart toppers?

Fundamentally this software has accumulated, absorbed and analyzed a huge amount of musical data but it can never simulate the elements I mentioned in my first paragraph that make particular songs massive hits. Fortunately, the vagaries of the human musical appetite requires a musical menu with imagination stretching beyond that of logic-based creations emanating from binary circuits.

As Queen and Freddie Mercury would have put it: " It's A Kind of Magic."

About the Author

Paul Whytock Blog | European Editor

Paul Whytock is European Editor for Penton Media's Electronics Division. From his base in London, England, he covers press conferences and industry events throughout the EU for Penton publications and its Engineering TV and Radio services Qualified to HNC Full Technological Certificate standard, Whytock trained as an automotive design engineer with Ford Motor Company prior to entering technical journalism.

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!