---
title: pianos
canonical_url: "https://www.modelscope.cn/models/ccmusic-database/pianos"
md_url: "https://www.modelscope.cn/models/ccmusic-database/pianos.md"
repository: ccmusic-database/pianos
chinese_name: "钢琴分类模型"
last_updated: 2026-08-05
license: "MIT License"
downloads: 1810
stars: 20
---

# pianos

> pianos - ccmusic-database 在 ModelScope 开源的模型。本研究基于深度学习技术，借鉴了计算机视觉领域的经典骨干网络结构，通过音频数据处理构建了一种创新的8分类钢琴音质判别器模型。该模型以Kawai、Kawai Grand、YOUNG CHANG、HSINGHAI、Steinway Theatre、Steinway Grand、Pearl…

ccmusic-database/pianos 是 ModelScope 魔搭社区上的机器学习模型，采用 MIT License 许可。

- **Repository**: ccmusic-database/pianos
- **License**: MIT License
- **Downloads**: 1810
- **Stars**: 20
- **Last updated**: 2026-08-05

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

---

# 简介 Intro
本研究基于深度学习技术, 借鉴了计算机视觉领域的经典骨干网络结构, 通过音频数据处理构建了一种创新的8分类钢琴音质判别器模型。该模型以Kawai、Kawai Grand、YOUNG CHANG、HSINGHAI、Steinway Theatre、Steinway Grand、Pearl River和Yamaha等8个品牌和类型的钢琴为研究对象。通过将音频数据转换为Mel频谱并在微调阶段进行监督学习, 模型能够准确地区分不同钢琴音质, 并在实际测试中表现出色。在训练过程中, 我们使用了大规模标注的音频数据集, 通过逐步学习从音频中提取关键特征, 深度学习技术的引入为模型的性能提升提供了关键支持。该音质判别器模型在音乐评估、音响工程等领域中具有广泛应用潜力, 为钢琴音质判别提供了一种先进而可靠的解决方案。这一研究为深度学习在音频领域的应用拓展了新的可能性, 为未来相关领域的研究和应用提供了有益的参考。

In this study, an innovative 8-categorized piano sound quality discriminator model is constructed by audio data processing based on deep learning techniques and drawing on the classical backbone network structure in computer vision. The model takes 8 brands and types of pianos such as Kawai, Kawai Grand, Young Chang, HSINGHAI, Steinway Theatre, Steinway Grand, Pearl River, and Yamaha as research objects. By converting the audio data into Mel spectra and performing supervised learning in the fine-tuning stage, the model is able to accurately distinguish different piano sound qualities and performs well in real-world tests. During the training process, we used a large-scale labeled audio dataset and extracted key features from the audio through stepwise learning, and the introduction of deep learning techniques provided key support for the model's performance improvement. The sound quality discriminator model has the potential for a wide range of applications in the fields of music evaluation and sound engineering, providing an advanced and reliable solution for piano sound quality discrimination. This research expands new possibilities for the application of deep learning in the audio field, and provides a useful reference for future research and application in related fields.

## 在线体验 Demo (推理代码)
<https://www.modelscope.cn/studios/ccmusic-database/pianos>

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

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

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

## 训练结果 Results
<table>
    <tr>
        <th>Loss curve</th>
        <td><img src="./loss.jpg"></td>
    </tr>
    <tr>
        <th>Training and validation accuracy</th>
        <td><img src="./acc.jpg"></td>
    </tr>
    <tr>
        <th>Confusion matrix</th>
        <td><img src="./mat.jpg"></td>
    </tr>
</table>

## 数据集 Dataset
<https://www.modelscope.cn/datasets/ccmusic-database/pianos>

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

## 校验 Evaluation
<https://github.com/monetjoe/pianos>

## 引用 Cite
```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}
}
```