← All modules

@maru shared this module so you can build on it and use it in your DAW.

AvailableMIDI & ModulationExperimental

Groove MIDI

Browse human drum performances and audition selected bar regions.

About

A browser for real drum performances that plays them as MIDI. Pick one of the 1,150 human drum performances in Google's Groove MIDI Dataset, choose a few bars, and send them to a drum sampler or any other MIDI instrument.

The idea

Programmed drums sit exactly on the grid; human drummers do not. These performances keep their original timing and velocity, with no quantisation, so you hear and see the small pushes, drags and accents that make a groove feel played. They are whole performances, beats and fills, not loops cut to length.

Finding a groove

Filter by style or by beat and fill, or search by tempo, meter, drummer and dataset split. Then choose a Start bar and how many Bars to play, from one to four. The lane view shows each hit across nine drum lanes where it was actually played, with its velocity, so you can see the microtiming as well as hear it.

Drum mapping chooses how notes come out. The default reduces the kit to nine standard General MIDI drum classes, which most drum samplers cover; this merges articulations such as rim and side stick, hi-hat edges and pedal, and cymbal variants. The original Roland mapping keeps every pitch as recorded.

Getting started

Groove MIDI makes no sound by itself: connect its MIDI output to a drum sampler. The Groove drum lab rack does this for you with a kit that covers all nine drum classes. Preview auditions the selected region while the rack is stopped, and Loop repeats it; when the rack's transport plays, the player starts and stops with it.

Limits: performances play at their original tempo and do not follow the rack's tempo. Only notes are sent; the hi-hat pedal controller data in the dataset is not transmitted.

Dataset attribution

Groove MIDI Dataset v1.0.0, Google LLC, licensed CC BY 4.0. Source and drum mapping. Jon Gillick, Adam Roberts, Jesse Engel, Douglas Eck and David Bamman, Learning to Groove with Inverse Sequence Transformations, ICML 2019. Changes: MIDI decoded, metadata and rhythm descriptors derived, resources repacked, regions selected and optional nine-class pitch reduction applied.

The navbar's menu opens an About dialog with a summary, the dataset and paper links, and the full attribution.

What would you make with Groove MIDI?

Describe what you want to make. Build it with the Studio agent.