Rap Lyrics Generation Using Vowel GAN

Tomoya Miyano, Hiroaki Saito

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Despite the success of recent rap and poetry generations using neural models, many of them do not consider vowels of the entire lyrics. Also, in many cases it is virtually impossible to generate completely new lyrics, because only existing rap lyrics are used as data sets. This paper proposes a new method of rap lyrics generation using a large amount of text such as novels in addition to rap lyrics. We divided the generation of rap lyrics into two steps; first, Generative Adversalial Net (GAN) generates rhymes and flows. Second, sequence-to-sequence converts them into rap lyrics. In addition, this method refers to the generation style of rap songs. In other words, they determine the music and rhythm first and apply the words second. We evaluated our method based on BLEU that can be measured mechanically.

Original languageEnglish
Title of host publicationComputational Linguistics - 16th International Conference of the Pacific Association for Computational Linguistics, PACLING 2019, Revised Selected Papers
EditorsLe-Minh Nguyen, Satoshi Tojo, Xuan-Hieu Phan, Kôiti Hasida
PublisherSpringer
Pages307-318
Number of pages12
ISBN (Print)9789811561672
DOIs
Publication statusPublished - 2020
Event16th International Conference of the Pacific Association for Computational Linguistics, PACLING 2019 - Hanoi, Viet Nam
Duration: 2019 Oct 112019 Oct 13

Publication series

NameCommunications in Computer and Information Science
Volume1215 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

Conference16th International Conference of the Pacific Association for Computational Linguistics, PACLING 2019
CountryViet Nam
CityHanoi
Period19/10/1119/10/13

Keywords

  • GAN
  • Rap song
  • Sequence-to-sequence

ASJC Scopus subject areas

  • Computer Science(all)
  • Mathematics(all)

Fingerprint Dive into the research topics of 'Rap Lyrics Generation Using Vowel GAN'. Together they form a unique fingerprint.

Cite this