IamXiangyu/osu-beatmaps-duplicated

---
license: other
license_name: osu-terms
license_link: https://osu.ppy.sh/legal/terms
task_categories:
- audio-classification
tags:
- music
- rhythm-game
- beatmaps
- osu
size_categories:
- 10K<n<100K
configs:
- config_name: original
data_files:
- split: train
path: original/*.tar
- config_name: compressed
data_files:
- split: train
path: compressed/*.tar
---

# osu! Beatmaps Dataset (WebDataset)

A collection of ranked/loved osu! beatmaps with audio and chart data, in WebDataset format.

## Dataset Variants

| Variant | Audio Format | Description |
|---------|--------------|-------------|
| `original` | MP3/OGG/WAV | Full quality original audio files |
| `compressed` | 64kbps Mono Opus | Compressed audio for smaller download |

```python
from datasets import load_dataset

# Load original audio variant
ds = load_dataset("project-riz/osu-beatmaps", "original", streaming=True)

# Load compressed audio variant
ds = load_dataset("project-riz/osu-beatmaps", "compressed", streaming=True)
```

## Statistics

| Metric | Value |
|--------|-------|
| Total beatmaps | 213,068 |
| Unique audio tracks (samples) | 46,386 |
| Total audio duration | 2,526 hours |
| Total beatmap playable length | 8,419 hours |
| Date range | 2007-10-06 to 2025-12-31 |

## Sample Structure

Each sample represents one unique audio track with all its associated beatmaps:

```plain
{seq_num:06d}.mp3|opus # Audio file
{seq_num:06d}.json # Metadata + embedded beatmap charts
```

### JSON Schema

```json
{
"audio_hash": "sha256_hash_of_audio",
"audio_length": 167.92,
"beatmaps": [
{
"beatmapset_id": 1,
"beatmap_id": 75,
"approved": 1,
// ...
"content": "osu file format v14\n..."
}
]
}
```

Audio files are deduplicated by SHA256 hash. Multiple beatmaps sharing the same audio are grouped into a single sample. In both variants, the hash refers to the original audio.

In adherence to WebDataset conventions, all original audio files share the extension `.mp3` **regardless of their actual format**.

## Metadata Fields

Most fields are taken from the [osu! API v1 response](https://github.com/ppy/osu-api/wiki#response). The `content` field contains the full `.osu` beatmap chart file (UTF-8 encoded).

## License

This dataset contains user-generated content from osu! and is subject to the [osu! Terms of Service](https://osu.ppy.sh/legal/terms). The beatmap charts are created by community mappers and the audio tracks are copyrighted by their respective owners.

This dataset is provided for research purposes only.