We all know that a cowbell sounds different than a crash cymbal, but it’s hard to pinpoint in words exactly how those sounds differ.

The Infinite Drum Machine aims to organize sounds in a way that’s more intuitive and direct. Part of Google’s A.I. Experiments series, the Infinite Drum Machine was built by Kyle McDonald, Manny Tan, and Yotam Mann, and uses machine learning to catalog thousands of everyday sounds, grouping the ones that sound similar on a visual plane with a a technique called t-SNE.

You can visually and aurally explore these “neighborhoods of sound” on their handy, colorful map, and even make your own beats with the drum sequencer. (But you can’t share the sick beats you create. Not yet, at least.)

Drag your cursor around the sound map to see the wide array of options, from the specific “dishes dog bowl plastic light set down on concrete” to the more generic “human cough,” as well as more expected percussive sounds, such as cowbell or Chinese hand cymbal.


Navigating huge sound libraries can be really difficult, especially because composers often have to find sounds  by name alone. “Disturbing industrial ambient noise,” anyone? When it launches, the Infinite Drum Machine could be a huge boon for drummers, creators, and producers looking to capture specific sounds.

Plus, it’s really fun. Do yourself a favor and waste the next 15 minutes (or more!) playing around with some of these sounds.

And watch the short video above to learn more about the experiment.


Anna Pulley is associate editor at Drum. Follow her on Twitter.