Groove MIDI Dataset (GMD): 13.6 Hours of Expressive Human Drumming as MIDI with Synthesized Audio (Google Magenta)
- Provider
- Google Magenta (Jon Gillick, Adam Roberts, Jesse Engel, Douglas Eck, David Bamman)
- License
- CC-BY-4.0 (source)
- Stored at
- Google Cloud Storage, gs://magentadata/datasets/groove (us)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Speech & Audio
- Source
- https://magenta.tensorflow.org/datasets/groove
About this dataset
The Groove MIDI Dataset (GMD) contains about 13.6 hours of tempo-aligned, expressive drumming recorded in MIDI on a Roland TD-11 electronic drum kit, with the kit's synthesized audio captured and aligned to the MIDI within 2 ms. It has 1,150 MIDI files and over 22,000 measures of drumming by 10 drummers, more than 80 percent of the duration from hired professionals, who played to a metronome in a wide range of styles, mixing long continuous performances with short beats and fills. Each performance is annotated with genre (given by the drummer), tempo and an anonymized drummer ID; most are in 4/4. Four drummers recorded the same 10 beats in their own style (eval-session/groove1-10 in the test split). A train/validation/test split is provided. The folder holds groove-v1.0.0.zip (MIDI, WAV and a metadata CSV, about 5.1 GB) and groove-v1.0.0-midionly.zip (3.3 MB). Read the MIDI with pretty_midi or note-seq; the dataset is also in TensorFlow Datasets as groove. Used for drum pattern generation, humanization of quantized beats (groove transfer), tap-to-drum and drum transcription. License: CC BY 4.0, made available by Google LLC. Cite Gillick, Roberts, Engel, Eck and Bamman, Learning to Groove with Inverse Sequence Transformations, ICML 2019.
How to access this dataset in Python
import gcsfs
fs = gcsfs.GCSFileSystem(token='anon')
fs.ls('magentadata/datasets/groove/', detail=False)[:10]Lists the files with anonymous access. Runs on any machine, with no cloud account.
Data structure
Files
Example file: datasets/groove/
Analyze this dataset without downloading it
On Scigantic, Groove MIDI Dataset (GMD): 13.6 Hours of Expressive Human Drumming as MIDI with Synthesized Audio (Google Magenta) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.