Integration of a music generator and a song lyrics generator to create Spanish popular songs

Abstract

[EN]The automatic generation of music is an emerging field of research that has attracted wide attention in Computer Science. However, most works are centered in classical music. This work develops ETHNO-MUSIC, an intelligent system that generates melodies based on popular music. ETHNO-MUSIC generates melodies with Markov models, which learns from a corpus of Spanish popular music. Then, given the importance of the lyrics in this context, ETHNO-MUSIC was integrated with Tra-La-Lyrics, an existing system that generates lyrics following a melody, which has been specifically adapted to suit this purpose. Several experiments were carried out to evaluate the quality of the results, based on human opinions towards generated pieces of music and lyrics. Overall, results are positive. Briefly, they reflect that, on the one hand, the melodies transmit a feeling of Spanish popular music, and on the other hand, the text of the lyrics is related to the topics analyzed, and the rhythm follows the melodic aspects of the music.