Do you spend your spare time penning lyrics and dreaming of hitting it big? If so, you may have more than humans to compete with in the future. Researchers at the University of Aalto in Finland have created a machine-learning algorithm that can both analyze existing rap lyrics and then spit out its own…and not just weak rhyming verses, either. The resulting lyrics are said to be on the same topic and in the same style as existing lyrics it has analyzed, and with a complexity on par with human-sourced lyrics. Training for the algorithm involved more than 10,000 songs from in excess of 100 different rappers.
The machine-learning algorithm is called DeepBeat, and as a test of its ability to produce adequate rap lyrics, the researchers had it analyze lines from existing rap songs, then choose what the next line should be from a list of random lines (as well as the proper line). The result was an 82-percent accuracy rating when choosing the correct lyric.
In addition, the algorithm can also generate rap songs by choosing existing lines from a database, assembling together into complete song. Below is one example of a song it produced using such a method (lines taken from different songs and cobbled together):
For a chance at romance I would love to enhance
But everything I love has turned to a tedious task
One day we gonna have to leave our love in the past
I love my fans but no one ever puts a grasp
I love you momma I love my momma – I love you momma
And I would love to have a thing like you on my team you take care
I love it when it’s sunny Sonny girl you could be my Cher
I’m in a love affair I can’t share it ain’t fair
Haha I’m just playin’ ladies you know I love you.
I know my love is true and I know you love me too
Girl I’m down for whatever cause my love is true
This one goes to my man old dirty one love we be swigging brew
My brother I love you Be encouraged man And just know
When you done let me know cause my love make you be like WHOA
If I can’t do it for the love then do it I won’t
All I know is I love you too much to walk away though
According to the researchers, the algorithm is also able to produce songs that have a higher “rhyming density” that what human rappers tend to come up with — by about 21-percent. It’s no replacement for humans yet, of course — the algorithm can assemble raps based on structure, but it can’t analyze the meaning to produce something solid. Yet.
SOURCE: MIT Technology Review