A GAN based drum pattern generation UI prototype

Autoren Hamid Eghbal-zadeh
Richard Vogl
Gerhard Widmer
Peter Knees
Titel A GAN based drum pattern generation UI prototype
Buchtitel Proceedings of ISMIR2018 Late-Breaking Demos Session
Typ in Konferenzband
Monat September
Jahr 2018
Seiten https://drive.google.com/file/d/12uktQ77d7MNyuQdOUfNKrT6zNIaJ-HeY/view
SCCH ID# 18095

Tools which support artists with the generation of drum tracks for music productions have gained popularity in recent years. Especially when utilizing digital audio workstations, using e.g. predefined drum loops can speed up drafting a musical idea. However, regarding variation and originality of drum patterns, methods based on generative models would be more suited for this task. In this work we present a drum pattern generation prototype based on a step sequencer interface which allows to explore the suitability of generative adversarial networks for this task.