数据集介绍

MAESTRO----MIDI and Audio Edited for Synchronous TRacks and Organization

 

MASETRO contains over a week of paired audio and MIDI recordings from nine years of International Piano-e-Competition events.

The MIDI data includes key strike velocities and sustain pedal positions.

Audio and MIDI files are aligned with ≈3 ms accuracy and sliced to individual musical pieces, which are annotated with composer, title, and year of performance. 

Uncompressed audio is of CD quality or higher (44.1–48 kHz 16-bit PCM stereo).

A train/validation/test split configuration is also proposed, so that the same composition, even if performed by multiple contestants, does not appear in multiple subsets.

MusicNet

https://homes.cs.washington.edu/~thickstn/musicnet.html

MusicNet contains recordings of human performances, but separately sourced score.

the alignment between audio and score is not fully accurate.

One advantage of MusicNet is that it contains instruments other than piano and a wider variety of recording environments.

 

MAPS

MAPS数据集

MAPS contains Disklavier recordings and synthesized audio created from MIDI files that were originally entered via sequencer.

As such, the "performances" are not as natural as the MAESTRO performances captured from live performances.

In addition, synthesized audio makes up a large fraction of the MAPS dataset.

MAPS also contains syntheses and recordings of individual notes and chords.

 

Saarland Music Data (SMD)

SMD is similar to MAESTRO in that it contains recordings and aligned MIDI of human performances on a Disklavier, but is 30 times smaller.