---
title: speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k
canonical_url: "https://www.modelscope.cn/models/iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k"
md_url: "https://www.modelscope.cn/models/iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k.md"
repository: iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k
chinese_name: "Regularized DINO说话人确认-中文-3D-Speaker-16k"
last_updated: 2024-12-26
license: "Apache License 2.0"
pipeline_tag: speaker-verification
tasks:
  - speaker-verification
model_type:
  - ecapa-tdnn
library_name:
  - pytorch
frameworks:
  - pytorch
language:
  - cn
domain:
  - audio
downloads: 8839
stars: 10
tags:
  - "SSL speaker verification"
  - "RDINO framework"
  - 3D-Speaker
---

# speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k

> speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k - iic 在 ModelScope 开源的模型。Regularized DINO (RDINO) 是基于时延神经网络构建的自监督说话人模型，该模型使用开源数据集3D-Speaker训练，不使用任何说话人标签。适用于16k中文测试数据，可以用于说话人确认、说话人日志，语音合成等任务。

iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k 是 ModelScope 魔搭社区上的speaker-verification模型，采用 Apache License 2.0 许可。

- **Repository**: iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k
- **License**: Apache License 2.0
- **Tasks**: speaker-verification
- **Tags**: SSL speaker verification, RDINO framework, 3D-Speaker
- **Downloads**: 8839
- **Stars**: 10
- **Last updated**: 2024-12-26

Source: https://www.modelscope.cn/models/iic/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k

---

## 3D-Speaker RDINO 说话人识别模型
RDINO模型是基于时延神经网络构建的自监督说话人模型，可用于说话人确认、说话人日志等任务。
## 模型简述
Regularized DINO使用使用teacher-student模型结构，通过最大化同一语句中不同增强片段的特征分布相似性，自监督地获取说话人特征，同时使用多样性正则和冗余度消除正则缓解训练中存在的模型坍塌问题。

<div align=center>
<img src="images/RDINO_architecture.jpg" width="800" />
</div>

更详细的信息见
- ERes2Net论文：[Pushing the limits of self-supervised speaker verification using regularized distillation framework](https://arxiv.org/pdf/2211.04168.pdf)
- 3D-Speaker论文：[3D-Speaker: A Large-Scale Multi-Device, Multi-Distance, and Multi-Dialect Corpus for Speech Representation Disentanglement](https://arxiv.org/pdf/2306.15354.pdf)
- github项目地址：[3D-Speaker](https://github.com/alibaba-damo-academy/3D-Speaker)

## 训练数据
本模型使用开源数据集[3D-Speaker](https://3dspeaker.github.io/)数据集进行无标签训练，包含约10k个说话人，可以对16k采样率的中文音频进行识别。
## 模型效果评估
在3D-Speaker中文测试集:Cross Device, Cross-Distance, Cross-Dialect中EER评测结果如下：
| Model | Params | Cross-Device | Cross-Distance | Cross-Dialect |
|:-----:|:------:| :------:|:------:|:------:|
| RDINO | 45.44M | 20.41% | 21.92% | 25.53% |

## 在线体验 开发中...
在页面右侧，可以在“在线体验”栏内看到我们预先准备好的示例音频，点击播放按钮可以试听，点击“执行测试”按钮，会在下方“测试结果”栏中显示相似度得分(范围为[-1,1])和是否判断为同一个人。如果您想要测试自己的音频，可点“更换音频”按钮，选择上传或录制一段音频，完成后点击执行测试，识别内容将会在测试结果栏中显示。
## 在Notebook中体验
```python
from modelscope.pipelines import pipeline
sv_pipline = pipeline(
    task='speaker-verification',
    model='damo/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k',
    model_revision='v1.0.1'
)
speaker1_a_wav = 'https://modelscope.cn/api/v1/models/damo/speech_campplus_sv_zh-cn_16k-common/repo?Revision=master&FilePath=examples/speaker1_a_cn_16k.wav'
speaker1_b_wav = 'https://modelscope.cn/api/v1/models/damo/speech_campplus_sv_zh-cn_16k-common/repo?Revision=master&FilePath=examples/speaker1_b_cn_16k.wav'
speaker2_a_wav = 'https://modelscope.cn/api/v1/models/damo/speech_campplus_sv_zh-cn_16k-common/repo?Revision=master&FilePath=examples/speaker2_a_cn_16k.wav'
# 相同说话人语音
result = sv_pipline([speaker1_a_wav, speaker1_b_wav])
print(result)
# 不同说话人语音
result = sv_pipline([speaker1_a_wav, speaker2_a_wav])
print(result)
# 可以自定义得分阈值来进行识别
result = sv_pipline([speaker1_a_wav, speaker2_a_wav], thr=0.198)
print(result)
```
## 训练和测试自己的RDINO模型
本项目已在[3D-Speaker](https://github.com/alibaba-damo-academy/3D-Speaker)开源了训练、测试和推理代码，使用者可按下面方式下载安装使用：
``` sh
git clone https://github.com/alibaba-damo-academy/3D-Speaker.git && cd 3D-Speaker
conda create -n 3D-Speaker python=3.8
conda activate 3D-Speaker
pip install -r requirements.txt
```

运行RDINO在3D-Speaker数据集上的训练脚本
``` sh
cd egs/3dspeaker/sv-rdino
# 需要在run.sh中提前配置训练使用的GPU信息，默认是4卡
bash run.sh
```
## 使用本预训练模型快速提取embedding
``` sh
pip install modelscope
cd 3D-Speaker
# 配置模型名称并指定wav路径，wav路径可以是单个wav，也可以包含多条wav路径的list文件
model_id=damo/speech_rdino_ecapa_tdnn_sv_zh-cn_3dspeaker_16k
# 提取embedding
python speakerlab/bin/infer_sv_ssl.py --model_id $model_id --wavs $wav_path --yaml egs/voxceleb/sv-rdino/conf/rdino.yaml
```

## 相关论文以及引用信息
如果你觉得这个该模型有所帮助，请引用下面的相关的论文
```BibTeX
@inproceedings{chen2023pushing,
  title={Pushing the limits of self-supervised speaker verification using regularized distillation framework},
  author={Chen, Yafeng and Zheng, Siqi and Wang, Hui and Cheng, Luyao and Chen, Qian},
  booktitle={ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={1--5},
  year={2023},
  organization={IEEE}
}
@inproceedings{chen2023pushing,
  title={3D-Speaker: A Large-Scale Multi-Device, Multi-Distance, and Multi-Dialect Corpus for Speech Representation Disentanglement},
  author={Siqi Zheng, Luyao Cheng, Yafeng Chen, Hui Wang and Qian Chen},
  url={https://arxiv.org/pdf/2306.15354.pdf},
  year={2023}
}
```

## 3D-Speaker 开发者社区钉钉群
<div align=left>
<img src="images/ding.jpg" width="280" />
</div>