Expanded Groove MIDI Dataset

dataset·aligned drum audio midi·active

Expanded human drum-performance corpus with MIDI-aligned audio rendered across 43 drum kits.

Verified facts

Official site	https://magenta.tensorflow.org/datasets/e-gmd ↗
Geography	United States
size	drum kits: 43 · duration hours: 444.5 · unique sequences: 1059 · rendered sequences: 45537
scope	Human drum sequences re-recorded across electronic and acoustic-style kits for drum transcription.
genres	multiple drum styles
license	CC BY 4.0
steward	Google LLC
creators	Lee Callender; Curtis Hawthorne; Jesse Engel
versions	1.0.0
languages	not_applicable
modalities	WAV; MIDI; CSV
geographies	United States
access method	Google-hosted full, MIDI-only and metadata downloads; Zenodo mirror.
intended uses	automatic drum transcription; velocity estimation
related papers	Improving Perceptual Quality of Drum Transcription with the Expanded Groove MIDI Dataset
annotation method	Audio aligned within 2 ms of MIDI with inherited train, test and validation splits.
collection method	Groove MIDI sequences re-recorded on 43 Roland TD-17 kits at 44.1 kHz/24-bit.
provenance claims	Derived from the Groove MIDI Dataset.
current availability	Official downloads active.
commercial use limits	No dataset-level noncommercial restriction stated; attribution required.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Official page reports unusable tracks from the semi-manual recording pipeline.