Aria-MIDI Motif Dataset: Segmented Piano MIDI Files for Symbolic Music Generation

Aria-MIDI Motif Dataset The Aria-MIDI Motif Dataset is a segmented collection of 3,283,502 piano MIDI files, derived from the original full version of the Aria-MIDI Dataset (1,186,253 files). The dataset consists of short musical motifs, extracted using a filtering script to select relevant segments with a similar number of notes and rhythmic structure. MIDI timings are rendered in absolute time at 120 BPM, preserving the original temporal structure of the performances. File Counts by Genre classical: 2,034,181 soundtrack: 574,612 jazz: 212,135 pop: 173,473 ragtime: 121,373 rock: 71,970 blues: 33,976 folk: 30,616 ambient: 29,500 atonal: 1,666 License The Aria-MIDI Motif Dataset is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). Citation / Related Works @misc{bradshaw2025ariamididatasetpianomidi, title={Aria-MIDI: A Dataset of Piano MIDI Files for Symbolic Music Modeling}, author={Louis Bradshaw and Simon Colton}, year={2025}, eprint={2504.15071}, archivePrefix={arXiv}, primaryClass={cs.SD}, url={https://arxiv.org/abs/2504.15071}, }