---
title: pianos
canonical_url: "https://www.modelscope.cn/datasets/ccmusic-database/pianos"
md_url: "https://www.modelscope.cn/datasets/ccmusic-database/pianos.md"
repository: ccmusic-database/pianos
chinese_name: "钢琴音质数据集"
last_updated: 2026-08-05
license: CC-BY-NC-ND
storage_size: "616 MB"
domain:
  - audio
tasks:
  - audio-classification
size_scale:
  - 10k-1m
downloads: 3834
stars: 18
---

# pianos

> pianos - ccmusic-database 在 ModelScope 开源的数据集。原始数据集包含来自中国音乐学院钢琴房的12个音域音频文件（.wav / .mp3 / .m4a 格式）以及1320个拆分的单音音频文件（.wav / .mp3 / .m4a 格式），涵盖了7种不同类型的钢琴（Kawai立式钢琴、Kawai大钢琴、YOUNG CHANG立式钢琴、HSINGHAI立式钢琴、Steinway大剧院里的大钢琴、Steinway大钢琴和Pearl…

ccmusic-database/pianos 是 ModelScope 魔搭社区上的audio-classification数据集，涉及 audio 领域，存储大小 616 MB，采用 CC-BY-NC-ND 许可。

- **Repository**: ccmusic-database/pianos
- **License**: CC-BY-NC-ND
- **Tasks**: audio-classification
- **Domain**: audio
- **Storage size**: 616 MB
- **Downloads**: 3834
- **Stars**: 18
- **Last updated**: 2026-08-05

Source: https://www.modelscope.cn/datasets/ccmusic-database/pianos

---

# 简介 Dataset Card for Piano Sound Quality
本数据集的原始数据源来自 [钢琴音质数据集](https://ccmusic-database.github.io/database/ccm.html#shou1), 包括 12 个全音域音频文件, 格式为.wav/.mp3/.m4a, 涵盖七种型号的钢琴: Kawai 立式钢琴、Kawai 三角钢琴、英昌立式钢琴、星海立式钢琴、大剧院施坦威钢琴、施坦威三角钢琴和珠江立式钢琴。此外, 还有 1,320 个拆分后的单音音频文件, 格式同样为.wav/.mp3/.m4a, 使得文件总数达到 1,332 个。此外, 数据集还包括一个 .xls 格式的评分表, 包含 29 位具有音乐背景的参与者对钢琴音质的主观评价。

基于上述原始数据, 经过数据处理后我们构建了当前集成版数据集的 [默认子集](#快速使用-usage), 其数据结构见 [数据预览](https://www.modelscope.cn/datasets/ccmusic-database/pianos/dataPeview)。由于默认子集缺少一个流行的钢琴品牌 Yamaha, 因此通过录制一架立式 Yamaha 钢琴, 将默认子集扩展为 [8_class子集](#快速使用-usage)。鉴于当前数据集已经被发表的文章验证过有效性, 基于 8_class 子集, 我们采用了该文章中对数据集评估的数据处理方法, 又构建出了 [评估子集](#快速使用-usage), 评估结果在 [钢琴分类模型](https://www.modelscope.cn/models/ccmusic-database/pianos) 中体现。除默认子集, 其余子集未在文章中体现。以下是各子集的数据结构简介。

The original data source of this dataset is [Piano Sound Quality Dataset](https://ccmusic-database.github.io/en/database/ccm.html#shou1), 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 the total number of files to 1,332. The dataset also includes a score sheet in .xls format containing subjective evaluations of piano sound quality provided by 29 participants with musical backgrounds.

Based on the aforementioned original dataset, after data processing, we constructed the [default subset](#快速使用-usage) of the current integrated version of the dataset, and its data structure can be viewed in the [viewer](https://www.modelscope.cn/datasets/ccmusic-database/pianos/dataPeview). Due to the need to increase the dataset size and the absence of a popular piano brand, Yamaha, the default subset is expanded by recording an upright Yamaha piano into the [8_class subset](#快速使用-usage). Since the current dataset has been validated by published articles, based on the 8_class subset, we adopted the data processing method for dataset evaluation from the article and constructed the [eval subset](#快速使用-usage), whose result has been shown in [pianos](https://www.modelscope.cn/models/ccmusic-database/pianos). Except for the default subset, the rest of the subsets are not represented in our paper. Below is a brief introduction to the data structure of each subset.

## 数据结构 Dataset Structure
### 默认子集 Default Subset Structure
|     audio     | mel (spectrogram) |                               label (7-class)                                |    pitch (88-class)    | bass_score | mid_score | treble_score | avg_score |
| :-----------: | :---------------: | :--------------------------------------------------------------------------: | :--------------------: | :--------: | :-------: | :----------: | :-------: |
| .wav, 44100Hz |   .jpg, 44100Hz   | PearlRiver / YoungChang / Steinway-T / Hsinghai / Kawai / Steinway / Kawai-G | 88/52 pitches on piano |  float32   |  float32  |   float32    |  float32  |

### 8分类子集 8_class Subset Structure
|     audio     | mel (spectrogram) |                                    label (8-class)                                    |    pitch (88-class)    | bass_score | mid_score | treble_score | avg_score |
| :-----------: | :---------------: | :-----------------------------------------------------------------------------------: | :--------------------: | :--------: | :-------: | :----------: | :-------: |
| .wav, 44100Hz |   .jpg, 44100Hz   | PearlRiver / YoungChang / Steinway-T / Hsinghai / Kawai / Steinway / Kawai-G / Yamaha | 88/52 pitches on piano |  float32   |  float32  |   float32    |  float32  |

### 评估子集 Eval Subset Structure
|  mel (spectrogram)   |                                    label (8-class)                                    |    pitch (88-class)    | bass_score | mid_score | treble_score | avg_score |
| :------------------: | :-----------------------------------------------------------------------------------: | :--------------------: | :--------: | :-------: | :----------: | :-------: |
| .jpg, 0.18s, 44100Hz | PearlRiver / YoungChang / Steinway-T / Hsinghai / Kawai / Steinway / Kawai-G / Yamaha | 88/52 pitches on piano |  float32   |  float32  |   float32    |  float32  |

### 文件格式 Data Instances
.zip(.wav, jpg)

### 标签 Data Fields
|           Piano           | Label |
| :-----------------------: | :---: |
|       1_PearlRiver        |   0   |
|       2_YoungChang        |   1   |
|       3_Steinway-T        |   2   |
|        4_Hsinghai         |   3   |
|          5_Kawai          |   4   |
|        6_Steinway         |   5   |
|         7_Kawai-G         |   6   |
| 8_Yamaha(For Non-default) |   7   |

### 数据分割 Data Splits
|       Split       |              Default |             8_class |                Eval |
| :---------------: | -------------------: | ------------------: | ------------------: |
|    train (80%)    |                  461 |                 531 |              14,678 |
| validation (10%)  |                   59 |                  68 |               1,835 |
|    test (10%)     |                   60 |                  69 |               1,839 |
|       total       |                  580 |                 668 |              18,352 |
| Total duration(s) | `2851.6933333333354` | `3247.941395833335` | `3247.941395833335` |

## 快速使用 Usage
:modelscope-code[]{type="sdk"}

## 维护 Maintenance
```bash
GIT_LFS_SKIP_SMUDGE=1 
```

:modelscope-code[]{type="git"}

### 环境(仅适配数据处理脚本) Requirements(Only for data processor)
```bash
cd data
echo y | conda create -n data python=3.x
conda activate data
pip install -r requirements.txt
```

### 数据处理脚本 Data processor
1. Open project with `VSCode`
2. Select file `./data/data.py`
3. Press `F5`

## 数据集描述 Dataset Description
### 数据集总结 Dataset Summary
For the integrated version of the dataset, we provide the expanded version which consists of 2,020 audio clips, along with the spectrogram generated above. The function to automatically shuffle the dataset and split it into training, validation, and test sets in an 8:1:1 ratio is also given. This dataset can be used for piano timbre analysis.

### 支持任务 Supported Tasks and Leaderboards
Piano Sound Classification, pitch detection

### 语言 Languages
English

## 数据集创建 Dataset Creation
### 动机 Curation Rationale
Lack of a dataset for piano sound quality

### 数据源 Source Data
#### 源数据收集与正规化 Initial Data Collection and Normalization
Zhaorui Liu, Shaohua Ji, Monan Zhou
#### 语言支持 Who are the source language producers?
Students from CCMUSIC & CCOM

### 标注 Annotations
#### 标注步骤 Annotation process
Students from CCMUSIC recorded different piano sounds and labeled them, and then a subjective survey of sound quality was conducted to score them.

#### 标注者 Who are the annotators?
Students from CCMUSIC & CCOM

### 敏感信息 Personal and Sensitive Information
Piano brands

## 使用考量 Considerations for Using the Data
### 社会影响 Social Impact of Dataset
Help develop piano sound quality scoring apps

### 偏好 Discussion of Biases
Only for pianos

### 其它限制 Other Known Limitations
Lack of black keys for Steinway, data imbalance

## 附加信息 Additional Information
### 策划人 Dataset Curators
Zijin Li

### 镜像 Mirror
<https://huggingface.co/datasets/ccmusic-database/pianos>

### 评估 Evaluation
[1] [Monan Zhou, Shangda Wu, Shaohua Ji, Zijin Li, and Wei Li. A Holistic Evaluation of Piano Sound Quality[C]//Proceedings of the 10th Conference on Sound and Music Technology (CSMT). Springer, Singapore, 2023.](https://arxiv.org/pdf/2310.04722.pdf)<br>
(Note: this paper only uses the first 7 piano classes in the dataset, its future work has finished the 8-class evaluation <a href="https://www.modelscope.cn/models/ccmusic-database/pianos">here</a>)

### 引用 Citation Information
```bibtex
@inproceedings{zhou2023holistic,
  title        = {A Holistic Evaluation of Piano Sound Quality},
  author       = {Monan Zhou and Shangda Wu and Shaohua Ji and Zijin Li and Wei Li},
  booktitle    = {National Conference on Sound and Music Technology},
  pages        = {3--17},
  year         = {2023},
  organization = {Springer}
}
```

### 贡献 Contributions
Provide a dataset for piano sound quality

<div style="display:none">
