---
license: mit
datasets:
- ccmusic-database/chest_falsetto
language:
- en
metrics:
- accuracy
pipeline_tag: audio-classification
tags:
- music
- art
---
# Intro
The design of the chest-falsetto voice discrimination model aims to effectively differentiate between real and synthetic voices in audio samples, with four specific categories including male chest, male falsetto, female chest, and female falsetto voices. The model's training is based on a backbone network from the computer vision (CV) domain, which involves transforming audio data into spectrograms and fine-tuning to enhance the network's accuracy in recognizing different voice categories. During training, a dataset containing both real and synthetic voice samples is utilized to ensure the model adequately learns and captures features relevant to male and female chest and falsetto voices. Through this approach, the model can finely classify different genders and chest/falsetto voices, providing a reliable solution for accurate voice discrimination in audio. This model holds broad potential applications in fields such as speech processing and music production, offering an efficient and precise tool for audio analysis and processing. Its training and fine-tuning strategies based on computer vision principles highlight the model's adaptability and robustness across different domains, providing beneficial examples for further research and application.

## Demo
<https://huggingface.co/spaces/ccmusic-database/chest-falsetto>

## Usage
```python
from modelscope import snapshot_download
model_dir = snapshot_download('ccmusic-database/chest_falsetto')
```

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

## Results
A fine-tuning result of MaxViT-T on CQT:
<style>
  #falsetto td {
    vertical-align: middle !important;
    text-align: center;
  }
  #falsetto th {
    text-align: center;
  }
</style>
<table id="falsetto">
    <tr>
        <th>Loss curve</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/chest_falsetto/resolve/master/maxvit_t_cqt_2024-07-30_17-23-12/loss.jpg"></td>
    </tr>
    <tr>
        <th>Training and validation accuracy</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/chest_falsetto/resolve/master/maxvit_t_cqt_2024-07-30_17-23-12/acc.jpg"></td>
    </tr>
    <tr>
        <th>Confusion matrix</th>
        <td><img src="https://www.modelscope.cn/models/ccmusic-database/chest_falsetto/resolve/master/maxvit_t_cqt_2024-07-30_17-23-12/mat.jpg"></td>
    </tr>
</table>

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

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

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

## Cite
```bibtex
@dataset{zhaorui_liu_2021_5676893,
  author       = {Monan Zhou, Shenyang Xu, Zhaorui Liu, Zhaowen Wang, Feng Yu, Wei Li and Baoqiang Han},
  title        = {CCMusic: an Open and Diverse Database for Chinese and General Music Information Retrieval Research},
  month        = {mar},
  year         = {2024},
  publisher    = {HuggingFace},
  version      = {1.2},
  url          = {https://huggingface.co/ccmusic-database}
}
```