In plain words: A computer program learns to pick the true next line of a rap from a set of candidate lines, then stitches lines from existing songs together so the result rhymes and still makes sense. Its lyrics pack 21% more rhyme than top human rappers.
Abstract
Writing rap lyrics requires both creativity to construct a meaningful, interesting story and lyrical skills to produce complex rhyme patterns, which form the cornerstone of good flow. We present a rap lyrics generation method that captures both of these aspects. First, we develop a prediction model to identify the next line of existing lyrics from a set of candidate next lines. This model is based on two machine-learning techniques: the RankSVM algorithm and a deep neural network model with a novel structure. Results show that the prediction model can identify the true next line among 299 randomly selected lines with an accuracy of 17%, i.e., over 50 times more likely than by random. Second, we employ the prediction model to combine lines from existing songs, producing lyrics with rhyme and a meaning. An evaluation of the produced lyrics shows that in terms of quantitative rhyme density, the method outperforms the best human rappers by 21%. The rap lyrics generator has been deployed as an online tool called DeepBeat, and the performance of the tool has been assessed by analyzing its usage logs. This analysis shows that machine-learned rankings correlate with user preferences.
Eric Malmi, Pyry Takala, Hannu Toivonen, Tapani Raiko, Aristides Gionis
arXiv:1505.04771 · cs.LG, cs.AI, cs.CL, cs.NE · submitted May 18, 2015 · updated Jun 9, 2016
abstract · pdf · html · This is a pre-print of an article appearing at KDD'16
