A group of Finnish programmers have come up with “DeepBeat”, an algorithmic program that learns how to rhyme based on the analysis of nearly 10,000 songs.
It’s unlikely to generate the words of Shakespeare or Rakim, but DeepBeat could give you some “rhymespiration”. The machine-learning algorithm has been cultivated by a team of computer scientists from Finland and what makes it particularly unique is its focus on assonance rather than perfect rhyming (basically, repetition of vowel sounds like “the cat sat on the mat” as opposed to “we’re Sampleface/we run the place/from here to space/with poise and grace“). This was to align itself better with the rap which doesn’t use perfect rhymes as much as it used to. DeepBeat has also been used to rank the rhyming abilities of the rappers who inadvertently “tutored” it. The likes of Inspector Deck and Rakim came out on top while shockingly, Eminem didn’t do so well. But according to an interview with Technology Review, this was down to Eminem often “achieving his rhymes by ‘bending’ words, a trick that this technique does not allow for”. Phew, crisis adverted.
(via Mental Floss)