DrumGAN interface


DrumGAN is a Generative Adversarial Network (GAN) that can synthesize high-fidelity audio for a wide variety of kick, snare, and cymbal sounds. It enables continuous interpolation of drum classes and provides an encoder that can map any preexisting sound into the latent parameter space of the GAN to generate variations of it.

In 2020, DrumGAN was used in our collaboration with French producer Twenty9. As a hip-hop beatmaker, he’s used to digging into countless kits, scrolling into endless drum sample lists, so as to get a distinctive sound. This habit came in handy when he started using Sony CSL drum-generating plugins in order to create drum samples both ahead of their time and to the point.


In 2022, Steinberg announced the integration of the DrumGAN technology into their drum re-synthesizer Backbone. Hence, DrumGAN became the world’s first commercially available technology to employ Neural Audio Synthesis. Today, thanks to the Sony CSL-Steinberg partnership, DrumGAN is more user-friendly as a tool and is available to creators around the world.

After a few months of experimenting with DrumGAN, Twenty9 releases his Drumkit made out of AI-generated sounds
Steinberg announces collaboration with Sony Computer Science Laboratories – Paris, with its AI-driven DrumGAN development incorporated into the latest release of Backbone.

Related Publications

ISMIR 2020
DrumGAN: Synthesis of Drum Sounds With Timbral Feature Conditioning Using GANs
Javier Nistal, Stefan Lattner, Gaël Richard
drum synthesis, progressive growing GANs, perceptual features, control parameters
ICML 2022 – MLAS Workshop
DrumGAN VST: A Plugin for Drum Sound Analysis/Synthesis with Autoencoding GANs
Javier Nistal, Cyran Aouameur, Ithan Velarde, Stefan Lattner
GANs, sound synthesis, drums, encoder