Bach Doodle Dataset: 21.6 Million User Melodies with Coconet Four-Voice Bach-Style Harmonizations (Google Magenta)
- Provider
- Google Magenta (Cheng-Zhi Anna Huang et al.)
- License
- CC-BY-4.0 (source)
- Stored at
- Google Cloud Storage, gs://magentadata/datasets/bach-doodle (us)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Speech & Audio
- Source
- https://magenta.tensorflow.org/datasets/bach-doodle
About this dataset
The Bach Doodle Dataset collects compositions made with the Google Doodle of March 2019, where users entered a short melody and a Coconet model harmonized it in four voices in the style of J.S. Bach chorales. It holds over 21.6 million miniature compositions from 8.5 million anonymous sessions (one session is a single page load, which can contain several compositions), about 6 years of music in total. Each data point contains the user's input melody, the four-voice harmonization returned by Coconet, and metadata: country of origin, the user's rating, the key signature, time spent composing and how many times the composition was played back. The folder holds the data sharded as 192 gzipped JSON-lines files (bach-doodle.jsonl-00000-of-00192.gz and onward, about 37 MB each) and as matching TFRecord shards of NoteSequence protocol buffers, readable with the note-seq library. Used for symbolic music generation, melody and harmony modeling, music education research and ethnomusicological comparison of melodies across countries. License: CC BY 4.0, made available by Google LLC. Cite Huang et al., The Bach Doodle: Approachable music composition with machine learning at scale, ISMIR 2019.
How to access this dataset in Python
import gcsfs
fs = gcsfs.GCSFileSystem(token='anon')
fs.ls('magentadata/datasets/bach-doodle/', detail=False)[:10]Lists the files with anonymous access. Runs on any machine, with no cloud account.
Data structure
Files
Example file: datasets/bach-doodle/
Analyze this dataset without downloading it
On Scigantic, Bach Doodle Dataset: 21.6 Million User Melodies with Coconet Four-Voice Bach-Style Harmonizations (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.
Related Speech & Audio datasets
- CVSS: Multilingual Speech-to-Speech Translation Corpus, 21 Languages into English (CVSS-C and CVSS-T, Google Research)
- Multilingual Librispeech
- E-GMD: Expanded Groove MIDI Dataset, 444 Hours of Human Drumming on 43 Drum Kits with Aligned MIDI and Audio (Google Magenta)
- CocoChorales: 240,000 Synthesized Four-Part Chorale Mixtures with Per-Instrument Stems, MIDI and Note Expression (Google Magenta)
- MInDS-14: multilingual spoken intent detection
- Nemotron Content Safety Audio Dataset
- Voxpopuli
- Groove MIDI Dataset (GMD): 13.6 Hours of Expressive Human Drumming as MIDI with Synthesized Audio (Google Magenta)