datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PianoVAM_v1
PianoVAM v1.1: A Multimodal Piano Performance Dataset
Version History
v1.1 (current). metadata.json is the canonical split file. Video files for the 'Sep 04-05' recordings are the sync-corrected versions, and files previously found to have video-MIDI synchronization issues have been relocated across splits; the current splits reflect these corrections.
v1.0. Initial release (ISMIR 2025).
Summary
This repository contains the PianoVAM (Video, Audio, Midi and… See the full description on the dataset page: https://huggingface.co/datasets/PianoVAM/PianoVAM_v1.PianoCoRe
PianoCoRe: Combined and Refined Piano MIDI Dataset
PianoCoRe is a large-scale piano MIDI dataset that unifies and refines major open-source piano corpora.
It contains 250,046 performances of 5,625 pieces written by 483 composers, totaling 21,763 hours of performed music.
PianoCoRe provides the most diverse composer- and composition-annotated piano MIDI data.
The metadata includes deduplication flags, MIDI quality labels and precise note-level score-performance alignments.
The… See the full description on the dataset page: https://huggingface.co/datasets/SyMuPe/PianoCoRe.pianos
Dataset Card for Piano Sound Quality Dataset
The original dataset is sourced from the Piano Sound Quality Dataset, which includes 12 full-range audio files in .wav/.mp3/.m4a format representing seven models of pianos: Kawai upright piano, Kawai grand piano, Young Change upright piano, Hsinghai upright piano, Grand Theatre Steinway piano, Steinway grand piano, and Pearl River upright piano. Additionally, there are 1,320 split monophonic audio files in .wav/.mp3/.m4a format, bringing… See the full description on the dataset page: https://huggingface.co/datasets/ccmusic-database/pianos.Airbot_MMK2_play_toy_piano
Airbot_MMK2_play_toy_piano
Dataset Description
This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot.
Task Preview
View Video Directly
Overview
Total Episodes: 50
Total Frames: 7683
FPS: 30
Dataset Size: 230.15 MB
Robot Name: Airbot_MMK2
End-Effector Type: five_finger_gripper
Teleoperation Type: Due to some reasons, this dataset temporarily cannot provide the teleoperation type information.… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/Airbot_MMK2_play_toy_piano.AIRBOT_MMK2_place_the_piano_and_the_needle-nose_pliers
AIRBOT_MMK2_place_the_piano_and_the_needle-nose_pliers
📋 Overview
This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot.
Robot Type: discover_robotics_aitbot_mmk2
| Codebase Version: v2.1
End-Effector Type: five_finger_hand
🏠 Scene Types
This dataset covers the following scene types:
home
🤖 Atomic Actions
This dataset includes the following atomic actions:
grasp
place
pick
📊… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/AIRBOT_MMK2_place_the_piano_and_the_needle-nose_pliers.Monster-Piano
Monster Piano
580204 solo Piano MIDI scores representations from Monster MIDI dataset
Installation and use
Load dataset
#===================================================================
from datasets import load_dataset
#===================================================================
monster_piano = load_dataset('asigalov61/Monster-Piano')
dataset_split = 'train'
dataset_entry_index = 0
dataset_entry =… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/Monster-Piano.details_chargoddard__loyal-piano-m7-cdpo
Dataset Card for Evaluation run of chargoddard/loyal-piano-m7-cdpo
Dataset Summary
Dataset automatically created during the evaluation run of model chargoddard/loyal-piano-m7-cdpo on the Open LLM Leaderboard.
The dataset is composed of 63 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard-old/details_chargoddard__loyal-piano-m7-cdpo.piano-protege-curriculum
Piano Protege Curriculum (Stages 1-3)
A human-style piano-practice curriculum of MIDI exercises and beginner
pieces with per-note fingering labels, built for training and evaluating
reinforcement-learning policies for simulated robotic piano playing
(e.g. RoboPianist).
It progresses the way a method book does: scales & arpeggios -> Hanon
finger-independence exercises -> simple real pieces.
Contents
Stage
Folder
Pieces
What
1
stage1_scales
864
Scales: 12… See the full description on the dataset page: https://huggingface.co/datasets/riyaranjan/piano-protege-curriculum.Accurate-Salamander-Grand-Piano-Yamaha-C5
Accurate-Salamander Grand Piano
Mirror of the Accurate-Salamander Grand Piano soundbank.
Accurate-Salamander author: Chisato Yamauchi
Original Salamander Grand Piano author: Alexander Holm
Instrument: Yamaha C5
Format: SFZ + WAV, 48 kHz / 24-bit
License: Creative Commons Attribution 3.0
Modifications in this repository: None. Files are redistributed unchanged.
Original project:
https://www.ir.isas.jaxa.jp/~cyamauch/AccurateSalamander/
Godzilla-Piano
Godzilla Piano
1.1M+ select normalized solo piano scores representations from Godzilla MIDI dataset
Godzilla's Musical Transformation by Microsoft Copilot
In the neon glow of a midnight throng,
A beast with headphones hums along.
Where thunderous roars once led the fray,
Now delicate keystrokes steal the day.
Godzilla, master of storm and lore,
Swaps terror for tunes, a musical score.
Each note a spark on the ebony keys,
Transforming chaos into rhythmic ease.… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/Godzilla-Piano.SportFashion_512x512youtube-piano-score
YouTube Piano Score
This repository contains piano score samples extracted from YouTube videos. Each sample lives under scores/<video_id>/ and may include:
meta.yaml: metadata for video segmentation, score layout, and staff grouping.
score.webp: the generated score image grid.
audio.*: the original audio extracted from video.
transkun.mid: the MIDI transcription result.
transkun-segmentation.yaml: measure and segment alignment metadata generated for the Transkun transcription… See the full description on the dataset page: https://huggingface.co/datasets/k-l-lambda/youtube-piano-score.team11_pianobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "so100",
"total_episodes": 40,
"total_frames": 11526,
"total_tasks": 1,
"total_videos": 40,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:40"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/roboticshack/team11_pianobot.chunked_text_piano_llm
Dataset Card for "chunked_text_piano_llm"
More Information needed
mmd_piano_tokenizedclassical-piano-midiIMSLP-Piano-Scores-leaderboard-resultsimnet1k_accordion_piano_accordion_squeeze_boxDiscover-Piano
Discover Piano
Ultimate pre-tokenized solo Piano MIDI dataset for symbolic music AI and MIR purposes
Installation and use
Load dataset
#===================================================================
from datasets import load_dataset
#===================================================================
discover_piano = load_dataset('asigalov61/Discover-Piano')
dataset_split = 'train'
dataset_entry_index = 0
dataset_entry =… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/Discover-Piano.piano
Music Demixing Benchmarks Piano
Unofficial, unpacked mirror of MVSep Quality Checker Piano dataset.
Piano Leaderboard (Piano)
Official zip file: 0.9 GB
Download
hf CLI and the huggingface_hub Python package are orders of magnitude faster than git.
Use
Hugging Face Command Line Interface (CLI)
git
huggingface_hub Python package
Choose
individual files
all files in the repository
subsets of files with filename pattern matching (fnmatch)
Read more
Hugging Face… See the full description on the dataset page: https://huggingface.co/datasets/MusicDemixingBenchmarks/piano.lmd-matched-piano
LMD Matched Piano
All LMD Matched titles with 2+ variations as normalized solo piano score representations with captions
Installation and use
Load dataset
#===================================================================
from datasets import load_dataset
#===================================================================
lmd_matched_piano = load_dataset('asigalov61/lmd-matched-piano')
dataset_split = 'train'
dataset_entry_index = 0… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/lmd-matched-piano.PianoScoreGenerationSample
Piano Score Generation Sample
A continuation produced by a compact Transformer (16.6M parameters) that generates piano scores at the notation level: the representation covers staves, voices, ties, slurs, dynamics, pedal and ornaments, rather than MIDI note events. The sample is provided as playable audio and as the underlying MusicXML score, next to the prompt it was given.
This is a demo of research output, not a model release. It accompanies a research internship at UC… See the full description on the dataset page: https://huggingface.co/datasets/Ant-Dlc/PianoScoreGenerationSample.BEAT-ICML2026-piano-dataset
BEAT ICML 2026 Piano Dataset
This repository hosts the piano-score dataset release for BEAT: Tokenizing
and Generating Symbolic Music by Uniform Temporal Steps (ICML 2026).
Released archive
BEAT-ICML2026-piano-v3-dedup-musicxml.tar.gz contains 169,283
deduplicated piano scores in score-partwise MusicXML format. This is the v3
deduplicated release of the MuseScore-derived XML portion of the BEAT piano
corpus. The paper reports a broader piano training collection of… See the full description on the dataset page: https://huggingface.co/datasets/marisa0v0/BEAT-ICML2026-piano-dataset.376-RobotLancers-PianoThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "so100_follower",
"total_episodes": 2,
"total_frames": 3028,
"total_tasks": 1,
"total_videos": 2,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:2"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/LeRobot-worldwide-hackathon/376-RobotLancers-Piano.PianoDatasetasap-piano-excerpts
ASAP Piano Excerpts
Multimodal piano dataset of 6,019 short excerpts (3–8 measures each) cut from the
ASAP piano dataset. Each row aligns four
modalities for the same musical window:
column
type
description
audio
Audio
excerpt audio, 22 050 Hz mono FLAC
kern
string
full **kern score text of the excerpt (**kern header included)
st_plus
Sequence[str]
ScoreTransformer+ token list (bar/key/time/note/chord/rest/clef)
midi2score
{stream: Sequence[int64]}
14 aligned… See the full description on the dataset page: https://huggingface.co/datasets/BarbosaTest/asap-piano-excerpts.imnet1k_grand_piano_grandpianos-parquet
Pianos Sound Quality Dataset
This version of dataset comprises seven models of pianos:
Kawai upright piano
Kawai grand piano
Young Change upright piano
Hsinghai upright piano
Grand Theatre Steinway piano
Steinway grand piano
Pearl River upright piano.
Note: the paper (Zhou et al., 2023) only uses the first 7 piano classes in the dataset, its future work has finished the 8-class evaluation.
License
MIT License
Copyright (c) CCMUSIC
Permission is hereby granted… See the full description on the dataset page: https://huggingface.co/datasets/confit/pianos-parquet.imnet1k_upright_upright_pianogrand-piano