CocoChorales

dataset·synthetic multitrack music·active

Synthetic four-part chamber-ensemble music with mixtures, stems, MIDI and fine-grained performance annotations.

Recorded facts

Official site	https://magenta.tensorflow.org/ceg-and-cocochorales ↗
Geography	United States
size	pieces: 240000 · instruments: 13 · mixture hours: 1411 · ensemble types: 4
scope	Generated Bach-style four-part chamber ensembles across string, brass, woodwind and random instrumentations.
genres	Bach chorales; chamber ensemble
steward	Google Magenta
creators	Yusong Wu; Josh Gardner; Ethan Manilow; Ian Simon; Curtis Hawthorne; Jesse Engel
languages	not_applicable
modalities	audio mixtures; audio stems; MIDI; multi-f0; performance annotations; synthesis parameters
geographies	Global
access method	Official project page and GitHub download instructions.
intended uses	transcription; source separation; multi-f0 estimation; performance analysis
related papers	The Chamber Ensemble Generator: Limitless High-Quality MIR Data via Generative Modeling
annotation method	Structured generators emit source, note, expression and synthesis ground truth.
collection method	Generated by the Chamber Ensemble Generator using Coconet and MIDI-DDSP.
provenance claims	Coconet used 382 Bach chorales; MIDI-DDSP used 44 URMP examples before amplification.
current availability	Download instructions linked from official project page.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Entire corpus is synthetic and stylistically concentrated on Bach chorales.
related models tools benchmarks	Coconet; MIDI-DDSP; URMP