datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
midi-classical-music
MIDI Classical Music
This dataset contains a comprehensive collection of MIDI files representing classical music compositions from various renowned composers.
The collection includes works from composers such as Bach, Beethoven, Chopin, Mozart, and many others.
The dataset is organized into directories by composer, with each directory containing MIDI files of their compositions.
The dataset is ideal for music analysis, machine learning models for music generation, and other… See the full description on the dataset page: https://huggingface.co/datasets/drengskapur/midi-classical-music.simplified_grooveThis is a copy of the Magenta Groove dataset
The script ´simplify_midi_pretty.py` reads the midi data and simplifies it, by removing any midi values that aren't kicks or snares, and quantizing the notes.
Monster-MIDI-Dataset
Monster MIDI Dataset
Giant searchable raw MIDI dataset for MIR and Music AI purposes
Monster MIDI Dataset GPU Search and Filter
[NEW] Monster GPU/CPU Search and Filter stand-alone Python module with improved matching
Installation
Install requirements
!git clone --depth 1 https://github.com/asigalov61/Monster-MIDI-Dataset
!pip install cupy-cuda12x
!pip install numpy==1.24.4
!pip install huggingface_hub
!pip install hf-transfer… See the full description on the dataset page: https://huggingface.co/datasets/projectlosangeles/Monster-MIDI-Dataset.ycuppe-midi
YCU-PPE-III: Piano Performance MIDI Dataset
MIDI transcriptions of the YCU-PPE-III piano performance dataset (Wang et al.), used for unreferenced Performance MOS (PMOS) prediction in EVPMR.
Overview
2,627 MIDI files transcribed from WAV recordings via transkun
13 songs performed by student pianists
2,511 performances with ratings from 3 expert judges (0-100 scale each)
Labels: normalized mean score to [0, 1]
Splits: 1,757 train / 377 val / 377 test (stratified by song)… See the full description on the dataset page: https://huggingface.co/datasets/anusfoil/ycuppe-midi.midisim-embeddings
midisim-embeddings
Pre-computed embeddings for midisim Python package
Main features
Ultra-fast and flexible GPU/CPU MIDI-to-MIDI similarity calculation, search and analysis
Quality pre-trained models and comprehensive pre-computed embeddings sets
Stand-alone, versatile, and extensive codebase for general or custom MIDI-to-MIDI similarity tasks
Full cross-platform compatibility and support
Pre-trained models… See the full description on the dataset page: https://huggingface.co/datasets/projectlosangeles/midisim-embeddings.aria-midi
The Aria-MIDI Dataset
Paper
The Aria-MIDI dataset is a collection of 1,186,253 MIDI files, comprising approximately 100,629 hours of transcribed solo-piano recordings, with metadata in categories including genre, composer, performer, as well as compositional identifiers. We developed Aria-MIDI to serve as a dataset for pre-training generative music models, and are releasing it with the goal of facilitating open research in music information retrieval and generative modeling for… See the full description on the dataset page: https://huggingface.co/datasets/loubb/aria-midi.Los-Angeles-MIDI-Dataset
Los Angeles MIDI Dataset
SOTA kilo-scale MIDI dataset for MIR and Music AI purposes
Search and Explore Los Angeles MIDI dataset
[NEW] Master MIDI Dataset GPU Search and Filter
Master MIDI Dataset Search and Filter
Make your own Los Angeles MIDI Dataset from any MIDI scrape
Make your own Los Angeles MIDI Dataset Metadata
Los Angeles MIDI Dataset is now avaialable for download!!!
Main… See the full description on the dataset page: https://huggingface.co/datasets/projectlosangeles/Los-Angeles-MIDI-Dataset.slakh2100_midi_only_reduxmidi-classical-music-toio-json
MIDI Classical Music
drengskapur/midi-classical-musicのデータセットをtoioの soundコマンドで再生しやすいように以下のフォーマットのjsonに変換したデータを含めたデータセット
data format
[
{
"track_name": "ALBENIZ: Aragon Op 47/6",
"priority": 1,
"notes": [
{
"note_number": 77,
"start_time_ms": 0,
"duration_units": 26
},
{
},
},
{
"track_name": "apurdam@pcug.org.au",
"priority": 2,
"notes": [
{
"note_number": 53,
"start_time_ms": 0… See the full description on the dataset page: https://huggingface.co/datasets/ayousanz/midi-classical-music-toio-json.MidiCaps
MidiCaps Dataset
The MidiCaps dataset [1] is a large-scale dataset of 168,385 midi music files with descriptive text captions, and a set of extracted musical features.
The captions have been produced through a captioning pipeline incorporating MIR feature extraction and LLM Claude 3 to caption the data from extracted features with an in-context learning task. The framework used to extract the captions is available open source on github.
The original MIDI files originate from the… See the full description on the dataset page: https://huggingface.co/datasets/amaai-lab/MidiCaps.midi-audio-abc_300smidi, synthesized audio, ABC code triples
(this dataset contains those with audio duration in 5-300s, several subsets with smaller duration
60s
30s
10s)
(token_length_abc field represents the token count of the abc text w.r.t. Qwen3's tokenizer)
midi files are from bread-midi-dataset
synthesized audio: use Don Allen's Timbres of Heaven as soundfont and FluidSynth as synthesizer
abc notation: mid2abc by EasyABC (midi2abc.py)
Citation
@misc{jiang2025advancingfoundationmodelmusic… See the full description on the dataset page: https://huggingface.co/datasets/Yi3852/midi-audio-abc_300s.midi-classical-music
MIDI Classical Music
This dataset contains a comprehensive collection of MIDI files representing classical music compositions from various renowned composers.
The collection includes works from composers such as Bach, Beethoven, Chopin, Mozart, and many others.
The dataset is organized into directories by composer, with each directory containing MIDI files of their compositions.
The dataset is ideal for music analysis, machine learning models for music generation, and other… See the full description on the dataset page: https://huggingface.co/datasets/ygonet/midi-classical-music.midi-audio-abc_60smidi, synthesized audio, ABC code triples
(this dataset contains those with audio duration in 5-60s, sampled from the full set with max 300s duration)
(token_length_abc field represents the token count of the abc text w.r.t. Qwen3's tokenizer)
midi files are from bread-midi-dataset
synthesized audio: use Don Allen's Timbres of Heaven as soundfont and FluidSynth as synthesizer
abc notation: mid2abc by EasyABC (midi2abc.py)
Citation
@misc{jiang2025advancingfoundationmodelmusic… See the full description on the dataset page: https://huggingface.co/datasets/Yi3852/midi-audio-abc_60s.pianist8-midiimslp-midi-cc0-1.0
IMSLP MIDI Dataset (CC0-1.0)
This dataset contains MIDI files and metadata crawled from IMSLP (International Music Score Library Project) on July 21-22, 2024.
Data Fields
midi_source: URL to the original MIDI file on IMSLP (incl. original uploader).
metadata_source: URL to the original metadata on IMSLP.
file_name, title, composer, year, era, style, key, license: Metadata fields.
midi: Raw MIDI bytes.
midi_mido: JSON-serialized mido object.
How to Retrieve… See the full description on the dataset page: https://huggingface.co/datasets/TiMauzi/imslp-midi-cc0-1.0.midi-audio-abc_longmidi, synthesized audio, ABC code triples
(this dataset contains those with audio duration in 5 min - 2 hours, less than 5 min data are in 300s
and there are several subsets with smaller duration
60s
30s
10s)
(token_length_abc field represents the token count of the abc text w.r.t. Qwen3's tokenizer)
midi files are from bread-midi-dataset
synthesized audio: use Don Allen's Timbres of Heaven as soundfont and FluidSynth as synthesizer
abc notation: mid2abc by EasyABC (midi2abc.py)… See the full description on the dataset page: https://huggingface.co/datasets/Yi3852/midi-audio-abc_long.midi-classical-music
MIDI Classical Music
This dataset contains a comprehensive collection of MIDI files representing classical music compositions from various renowned composers.
The collection includes works from composers such as Bach, Beethoven, Chopin, Mozart, and many others.
The dataset is organized into directories by composer, with each directory containing MIDI files of their compositions.
The dataset is ideal for music analysis, machine learning models for music generation, and other… See the full description on the dataset page: https://huggingface.co/datasets/elizawhitfield/midi-classical-music.midi-page-assetsLyrics-MIDI-Dataset
Lyrics MIDI Dataset
~179k original MIDI files with matched lyrics, summaries, embeddings, keywords, and chords
Overview
The Lyrics MIDI Dataset is a large-scale multimodal collection of symbolic music files paired with corresponding lyrics in plain text, enriched with AI-generated summaries, pre-computed embeddings, extracted keywords, and chord annotations. It enables research on lyric-conditioned music generation, alignment between textual and musical… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/Lyrics-MIDI-Dataset.MIDI-Images
A dataset of MIDI images designed for use with diffusion models for music generation, music classification, text-to-music and other purposes
🤗 Check out Imagen MIDI Images LIVE demo on Hugging Face Spaces 🤗
Installation
from huggingface_hub import snapshot_download
repo_id = "asigalov61/MIDI-Images"
repo_type = 'dataset'
local_dir = "./MIDI-Images"
snapshot_download(repo_id, repo_type=repo_type, local_dir=local_dir)
MIDI Images… See the full description on the dataset page: https://huggingface.co/datasets/asigalov61/MIDI-Images.e-gmd-v1.0.0-midi
Expanded Groove MIDI Dataset (E-GMD v1.0.0) — MIDI-only mirror
Unofficial mirror. This is a convenience mirror of the MIDI-only portion
of the Expanded Groove MIDI Dataset (E-GMD), redistributed under its CC BY 4.0
license (see below). It is not affiliated with or endorsed by Google / Magenta.
The canonical source is https://g.co/magenta/e-gmd.
E-GMD is a large dataset of human drum performances (~444 hours) recorded on a
Roland TD-17 electronic kit, captured as MIDI with… See the full description on the dataset page: https://huggingface.co/datasets/yalishanda/e-gmd-v1.0.0-midi.midicaps-rqvaemidi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-aug-lessgroove_midineuropiano-midiclassical-piano-midisampleflip-midi
SampleFlip MIDI Chord Progressions
3,764 MIDI chord progression files for music production. Used by SampleFlip for melody derivation and harmonic reference.
Collections
Collection
Files
Description
chordprogressions/Altered Dominant Chords
Altered dominant voicings
Jazz/neo-soul
chordprogressions/EDM Progressions
EDM chord patterns
House, trance, progressive
chordprogressions/Freemidis2025
Free MIDI pack 2025
Mixed genres
chordprogressions/More Genres… See the full description on the dataset page: https://huggingface.co/datasets/ronantakizawa/sampleflip-midi.MIDI_Art
MIDI Art images
MIDI art images with human-generated description txt files
Source
./data_source.md
Type
MIDIArt: MIDI art images
PianoRoll: piano roll images
Original size
./Original_size
Specific resized width and height
Folder naming: {width}_{height}
midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs
midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs
Pre-tokenized MIDI pieces for IsoFLOP scaling-law runs. Each row is one full
piece (no time-windowing); training crops sequences from packed token bins.
The source column is the original piece metadata as JSON so a row can be
traced back to its EPR Labs source dataset.
Based on MIDI datasets gathered by EPR Labs.
Codec
name: dyadic
tokenizer vocab size: 512
max_time_step: 1.0
n_velocity_bins: 32… See the full description on the dataset page: https://huggingface.co/datasets/wmatejuk/midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs.midi-dataset
[aria-midi-v1-pruned-ext] with midi column for midi file