2023-06-20 21:30:00
Flowers by Miley Cyrus, Kill Bill by SZA, Escapism by Raye: they are all huge hits of 2023. Great songs, but what really makes it a hit? Is there some kind of magic or lottery that causes that, or is it somehow measurable when a certain number becomes a gigahhit? Well, at least it can be predicted, so hits apparently have something in common. Machine learning can almost always predict exactly which songs will storm the charts.
Machinelearning
A lot of music is released every day, especially now that AI can also make music (which is not allowed to win prizes). How do you deal with that as a radio station? Or as a streaming service? What are you pushing forward to be even more of a hit? AI was once developed to help predict that, but it only got it right on half of the hits. However, thanks to machine learning, there is still a success story to tell.
Researchers at Claremont Graduate University have linked neurological data to machine learning, which made it possible to predict hits very well. The researchers put sensors on people and then had them listen to 24 songs and write down their preferences. As they listened and judged, their brain responses were watched by the scientists, and specifically the part that controls mood was where machine learning could work its magic. Successfully. Without machine learning, it was possible to predict a hit in 69 percent of the cases based on that mood part of the brain, while that was much higher with machine learning. In a whopping 97 percent of cases, machine learning was right.
High expectations
The researcher says that it takes neural activity from just 33 people to estimate whether millions of people will listen to a particular song. Even if the machine learning only heard 1 minute of the song, it still correctly predicted whether the song will be a hit 82 percent of the time. Very handsome. And it also says something about our brain: so we can probably figure out quite quickly whether a song is ‘our thing’ or not. It is undoubtedly not for nothing that many people know after hearing one tone of a song which song it is. Music does something special to our brain: that’s why it has such a good effect on people with Alzheimer’s.
Imagine Spotify investing in this kind of machine learning? Then you no longer have to spend the entire New Music Friday to discover music that you like. On the other hand, that is also a shame: that box of bonbons that Spotify serves you every Friday, it also has something. We are also curious to what extent there are differences between different people, or whether we are really not all that special and still really all go for the same hits. But in general, machine learning can predict it very accurately, so who knows where this will go. We can’t predict it, but machine learning might.
Laura Jenny
When she’s not tapping, she’s traveling around the wonderful world of entertainment or some cool place in the real world. Mario is the man of her life,…
#Machine #learning #predict #monster #hit