Hip-hop artists, record execs, fans and critics all have some notion, no matter how vague, of what it takes to make a hit song, a loose criteria that can reveal a lot about the listener. Sometimes it means riding a certain trend or donning a specific style. Other times it’s as easy as being yourself. It’s what we would call an inexact science, that is until now.
Anthony Abraham, Nikhita Koul and Joe Morales (students at UC Berkeley) created an algorithm that among other things will determine whether a song, according to Billboard standards, will become a hit or not. Dubbed the Rap Analysis Project (or R.A.P. for short) the program pulls from a database of lyrics from 1980 to 2015 and then detects notable trends, ones that can help producers understand the momentum of change with a bit more clarity.
The algorithm closely analyzes specific themes, references and word choices to make its conclusion. For example it can scan its database to see the frequency of how many times say New York is mentioned in a song. It then deciphers the data determining the number of chart topping songs that have referenced it. It also allows the user to filter the information through certain eras, so you can see if a Drake song would have been a hit in the ’80s. The creators have also assigned visuals for their data. In one example they combed the database to see what name brands were mentioned most in a hit song. Examples include Bentley, Apple, Maybach, Chevy and Twitter.
The goal of R.A.P. is to predict if a song will appear on the Billboard Top 100. The highest success rate came when they filtered the information through three categories: song topic, vulgarity and release date. And just like that the algorithm was able to predict with a 71% success rate if a song would be a hit or not.
It’s not an entirely new trend, analytics have dominated sports for years now, a field that has for so long relied on gut instincts and hunches. Rap is just the next arena to be infiltrated. The information will give execs something to be excited about, a way to get ahead of the curve. But they’ll most likely resort back to the old way, which means they’ll wait for an artist to blow up on their own and then copy it a thousand times over until someone slightly different comes along.