---
license: mit
---

# Intro
The Chinese Traditional Instrument Sound Model is a cutting-edge research outcome based on the Chinese Traditional Instrument Sound Dataset. This model covers recordings of over 200 types of Chinese traditional musical instruments, modified instruments, and ethnic minority instruments. Notably, it includes some instruments that are rarely seen among the general public in China. By conducting in-depth analysis and learning of the sound characteristics of these instruments, the model can accurately identify and classify the sounds of various instruments, providing strong technical support for music information retrieval, musical instrument research, and the preservation and inheritance of ethnic music.

## Demo (inference code)
<https://huggingface.co/spaces/ccmusic-database/CTIS>

## Usage
```python
from huggingface_hub import snapshot_download
model_dir = snapshot_download("ccmusic-database/CTIS")
```

## Maintenance
```bash
GIT_LFS_SKIP_SMUDGE=1 git clone git@hf.co:ccmusic-database/CTIS
cd CTIS
```

## Results
|   Backbone    | Size(M) |     Mel     |             CQT             |   Chroma    |
| :-----------: | :-----: | :---------: | :-------------------------: | :---------: |
|   vit_l_32    |  306.5  |    0.936    |            0.921            | **_0.845_** |
|    swin_t     |  28.3   | **_0.956_** |         **_0.940_**         |    0.759    |
|               |         |             |                             |             |
| regnet_y_32gf |   145   | **_0.973_** | [**_0.980_**](#best-result) |    0.848    |
|   vgg19_bn    |  143.7  |    0.966    |            0.965            | **_0.852_** |
|    alexnet    |  61.1   |    0.936    |            0.921            |    0.661    |
|   resnet101   |  44.5   |    0.953    |            0.949            |    0.782    |
| inception_v3  |  27.2   |    0.860    |            0.855            |    0.664    |

### Best result
<table>
    <tr>
        <th>Loss curve</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/CTIS/resolve/master/regnet_y_32gf_cqt_2024-12-02_15-05-57/loss.jpg"></td>
    </tr>
    <tr>
        <th>Training and validation accuracy</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/CTIS/resolve/master/regnet_y_32gf_cqt_2024-12-02_15-05-57/acc.jpg"></td>
    </tr>
    <tr>
        <th>Confusion matrix</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/CTIS/resolve/master/regnet_y_32gf_cqt_2024-12-02_15-05-57/mat.jpg"></td>
    </tr>
</table>

## Dataset
<https://huggingface.co/datasets/ccmusic-database/CTIS>

## Mirror
<https://www.modelscope.cn/models/ccmusic-database/CTIS>

## Evaluation
<https://github.com/monetjoe/ccmusic_eval>

## Cite
```bibtex
@article{Zhou-2025,
  author  = {Monan Zhou and Shenyang Xu and Zhaorui Liu and Zhaowen Wang and Feng Yu and Wei Li and Baoqiang Han},
  title   = {CCMusic: An Open and Diverse Database for Chinese Music Information Retrieval Research},
  journal = {Transactions of the International Society for Music Information Retrieval},
  volume  = {8},
  number  = {1},
  pages   = {22--38},
  month   = {Mar},
  year    = {2025},
  url     = {https://doi.org/10.5334/tismir.194},
  doi     = {10.5334/tismir.194}
}
```