A Computational Approach to Rap Lyrics Generation

DeepBeat, an algorithm for rap lyrics generation, outperforms top human rappers by 21 % in terms of length and frequency of the rhymes in the lyrics.

 

The rapping algorithm was quantitatively evaluated with two measures. First, the prediction performance was evaluated by measuring how well the algorithm predicts the next line of an existing rap song. This it can do surprisingly well.

 

- An 82% accuracy was achieved for separating the true next line from a randomly chosen line, says researcher Eric Malmi.

 

Second, a rhyme density measure was  introduced  and  showed that DeepBeat, an algorithm for rap lyrics generation, outperforms the top human rappers by 21 % in terms of length and frequency of the rhymes in the produced lyrics.

The validity of the rhyme density measure was assessed by conducting a human experiment which showed that the measure correlates with a rapper's own notion of technically skilled lyrics.

 

Researchers Eric Malmi and Pyry Takala, professors Tapani Raiko and Aristides Gionis from the Department of Computer Science at the Aalto University and professor Hannu Toivonen from the University of Helsinki and HIIT developed a machine-learning algorithm that learned to rap. To train the machine-learning algorithm, the researchers began with a database of over 10,000 songs from more than 100 rap artists.

 

DopeLearning: A Computational Approach to Rap Lyrics Generation

 

Aalto CS news: Researchers developed a machine-learning algorithm that learned to rap