WSC-Train
收藏魔搭社区2026-01-06 更新2025-09-27 收录
下载链接：
https://modelscope.cn/datasets/ASLP-lab/WSC-Train
下载链接
链接失效反馈官方服务：
资源简介：
<h1 align="center">
WenetSpeech-Chuan: A Large-Scale Sichuanese Corpus With Rich Annotation For Dialectal Speech Processing
</h1>
<p align="center">
Yuhang Dai<sup>1</sup><sup>,*</sup>, Ziyu Zhang<sup>1</sup><sup>,*</sup>, Shuai Wang<sup>4</sup><sup>,5</sup>,
Longhao Li<sup>1</sup>, Zhao Guo<sup>1</sup>, Tianlun Zuo<sup>1</sup>,
Shuiyuan Wang<sup>1</sup>, Hongfei Xue<sup>1</sup>, Chengyou Wang<sup>1</sup>,
Qing Wang<sup>3</sup>, Xin Xu<sup>2</sup>, Hui Bu<sup>2</sup>, Jie Li<sup>3</sup>,
Jian Kang<sup>3</sup>, Binbin Zhang<sup>5</sup>, Lei Xie<sup>1</sup><sup>,╀</sup>
</p>
<p align="center">
<sup>1</sup> Audio, Speech and Language Processing Group (ASLP@NPU), Northwestern Polytechnical University <br>
<sup>2</sup> Beijing AISHELL Technology Co., Ltd. <br>
<sup>3</sup> Institute of Artificial Intelligence (TeleAI), China Telecom <br>
<sup>4</sup> School of Intelligence Science and Technology, Nanjing University <br>
<sup>5</sup> WeNet Open Source Community <br>
</p>
<p align="center">
📑 <a href="https://arxiv.org/abs/2509.18004">Paper</a>    |   
🐙 <a href="https://github.com/ASLP-lab/WenetSpeech-Chuan">GitHub</a>    |   
🤗 <a href="https://huggingface.co/collections/ASLP-lab/wenetspeech-chuan-68bade9d02bcb1faece65bda">HuggingFace</a>
<br>
🎤 <a href="https://aslp-lab.github.io/WenetSpeech-Chuan/">Demo Page</a>    |   
💬 <a href="https://github.com/ASLP-lab/WenetSpeech-Chuan?tab=readme-ov-file#contact">Contact Us</a>
</p>
<div align="center">
<img width="800px" src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/logo/WenetSpeech-Chuan-Logo.png?raw=true" />
</div>
## Dataset
### WenetSpeech-Chuan Overview
* Contains 10,000 hours of large-scale Chuan-Yu dialect speech corpus with rich annotations, the largest open-source resource for Chuan-Yu dialect speech research.</li>
* Stores metadata in a single JSON file, including audio path, duration, text confidence, speaker identity, SNR, DNSMOS, age, gender, and character-level timestamps. Additional metadata tags may be added in the future.</li>
* Covers ten domains: Short videos, Entertainment, Live streams, Documentary, Audiobook, Drama, Interview, News and others.</li>
<div align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/figs/domain.png?raw=true" width="300" style="display:inline-block; margin-right:10px;" />
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/figs/quality_distribution.jpg?raw=true" width="300" style="display:inline-block;" />
</div>
### Metadata Format
We store all audio metadata in a standardized JSON format, where the core fields include `utt_id` (unique identifier for each audio segment), `rover_result` (ROVER result of three ASR transcriptions), `confidence` (confidence score of text transcription), `jyutping_confidence` (confidence score of Cantonese pinyin transcriptions), and `duration` (audio duration); speaker attributes include `speaker_id`, `gender`, and `age`; audio quality assessment metrics include `sample_rate`, `DNSMOS`, and `SNR`; timestamp information includes `timestamp` (precisely recording segment boundaries with `start` and `end`); and extended metadata under the `meta_info` field includes `program` (program name), `region` (geographical information), `link` (original content link), and `domain` (domain classification).
#### 📂 Content Tree
```
WenetSpeech-Chuan
├── metadata.jsonl
├── .gitattributes
└── README.md
```
<!-- WenetSpeech-Chuan
├── metadata.jsonl
│
├── audio_labels/
│ ├── wav_utt_id.jsonl
│ ├── wav_utt_id.jsonl
│ ├── ...
│ └── wav_utt_id.jsonl
│
├── .gitattributes
└── README.md -->
#### Data sample：
###### metadata.jsonl
{<br>
"utt": 音频id, <br>
"filename"：音频文件名(type: str), <br>
"text": 转录抄本(type: str), <br>
"domain": 参考领域信息(type: list[str]), <br>
"gender": 说话人性别(type: str), <br>
"age": 说话人年龄标签 (type: int范围, eg: 中年（36~59）), <br>
"wvmos": 音频质量分数(type: float), <br>
"confidence": 转录文本置信度（0-1）(type: str), <br>
"emotion": 说话人情感标签 (type: str，eg: 愤怒), <br>
} <br>
**example:**
{ <br>
"utt": "013165495633_09mNC_9_5820", <br>
"filename": "013165495633_09mNC_9_5820.wav", <br>
"text": "还是选二手装好了的别墅诚心入如意的直接入住的好好", <br>
"domain": [ <br>
"短视频" <br>
], <br>
"gender": "Male", <br>
"age": "YOUTH", <br>
"wvmos": 2.124380588531494, <br>
"confidence": 0.8333, <br>
"emotion": angry, <br>
} <br>
<!-- ###### audio_labels/wav_utt_id.jsonl：
{ <br>
"wav_utt_id_timestamp": 以 转化为wav后的长音频id_时间戳信息 作为切分后的短音频id (type: str), <br>
"wav_utt_id_timestamp_path": 短音频数据路径 (type: str), <br>
"audio_clip_id": 该段短音频在长音频中的切分顺序编号, <br>
"timestamp": 时间戳信息， <br>
"wvmos_score": wvmos分数，衡量音频片段质量 (type: float), <br>
"text": 对应时间戳的音频片段的抄本 (type: str), <br>
"text_punc": 带标点的抄本 (type: str), <br>
"spk_num": 音频片段说话人个数，single/multi (type: str) <br>
"confidence": 抄本置信度 (type: float), <br>
"emotion": 说话人情感标签 (type: str，eg: 愤怒), <br>
"age": 说话人年龄标签 (type: int范围, eg: 中年（36~59）), <br>
"gender": 说话人性别标签 (type: str，eg: 男/女), <br>
} <br>
-->
<!-- #### Data sample（EN）：
###### metadata.jsonl
{ <br>
"utt_id": Original long audio ID, <br>
"wav_utt_id": Converted long audio ID after transforming to WAV format, <br>
"source_audio_path": Path to the original long audio file, <br>
"audio_labels": Path to the label file of short audio segments cut from the converted long audio, <br>
"url": Download link for the original long audio <br>
} <br>
###### audio_labels/wav_utt_id.jsonl：
{ <br>
"wav_utt_id_timestamp": Short audio segment ID, composed of the converted long audio ID + timestamp information (type: str), <br>
"wav_utt_id_timestamp_path": Path to the short audio data (type: str), <br>
"audio_clip_id": Sequence number of this short segment within the long audio, <br>
"timestamp": Timestamp information, <br>
"wvmos_score": WVMOS score, measuring the quality of the audio segment (type: float), <br>
"text": Transcript of the audio segment corresponding to the timestamp (type: str), <br>
"text_punc": Transcript with punctuation (type: str), <br>
"spk_num": Number of speakers in the audio segment, single/multi (type: str), <br>
"confidence": Confidence score of the transcript (type: float), <br>
"emotion": Speaker’s emotion label (type: str, e.g., anger), <br>
"age": Speaker’s age label (type: int range, e.g., middle-aged (36–59)), <br>
"gender": Speaker’s gender label (type: str, e.g., male/female) <br>
} <br>
-->
### WenetSpeech Usage
You can obtain the original video source through the `link` field in the metadata file (`metadata.json`). Segment the audio according to the `timestamps` field to extract the corresponding record. For pre-processed audio data, please contact us using the information provided below.
## Contact
If you have any questions or would like to collaborate, feel free to reach out to our research team via email: yhdai@mail.nwpu.edu.cn or ziyu_zhang@mail.nwpu.edu.cn.
You’re also welcome to join our WeChat group for technical discussions, updates, and — as mentioned above — access to pre-processed audio data.
<p align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/raw/main/src/figs/wechat.jpg" width="300" alt="WeChat Group QR Code"/>
<em>Scan to join our WeChat discussion group</em>
</p>
<p align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Yue/raw/main/figs/npu@aslp.jpeg" width="300" alt="Official Account QR Code"/>
</p>
<h1 align="center">WenetSpeech-Chuan：面向方言语音处理的大规模川渝方言标注语料库</h1>
<p align="center">
戴宇航<sup>1</sup><sup>,*</sup>, 张子瑜<sup>1</sup><sup>,*</sup>, 王帅<sup>4</sup><sup>,5</sup>,
李豪龙<sup>1</sup>, 郭昭<sup>1</sup>, 左天伦<sup>1</sup>,
王水元<sup>1</sup>, 薛鸿飞<sup>1</sup>, 王承友<sup>1</sup>,
王清<sup>3</sup>, 徐鑫<sup>2</sup>, 卜辉<sup>2</sup>, 李杰<sup>3</sup>,
康健<sup>3</sup>, 张彬彬<sup>5</sup>, 谢磊<sup>1</sup><sup>,╀</sup>
</p>
<p align="center">
<sup>1</sup> 西北工业大学音频、语音与语言处理课题组（ASLP@NPU）<br>
<sup>2</sup> 北京爱数智慧科技有限公司（AISHELL Technology Co., Ltd.）<br>
<sup>3</sup> 中国电信人工智能研究院（TeleAI）<br>
<sup>4</sup> 南京大学智能科学与技术学院<br>
<sup>5</sup> WeNet开源社区<br>
</p>
<p align="center">
📑 <a href="https://arxiv.org/abs/2509.18004">论文</a>    |   
🐙 <a href="https://github.com/ASLP-lab/WenetSpeech-Chuan">GitHub仓库</a>    |   
🤗 <a href="https://huggingface.co/collections/ASLP-lab/wenetspeech-chuan-68bade9d02bcb1faece65bda">HuggingFace数据集集合</a>
<br>
🎤 <a href="https://aslp-lab.github.io/WenetSpeech-Chuan/">演示页面</a>    |   
💬 <a href="https://github.com/ASLP-lab/WenetSpeech-Chuan?tab=readme-ov-file#contact">联系我们</a>
</p>
<div align="center">
<img width="800px" src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/logo/WenetSpeech-Chuan-Logo.png?raw=true" />
</div>
## 数据集
### WenetSpeech-Chuan 数据集概览
* 包含总计10000小时的大规模川渝方言语音语料库及丰富标注信息，是目前面向川渝方言语音研究的最大规模开源资源。
* 元数据以单一JSON文件存储，涵盖音频路径、时长、文本置信度、说话人身份、信噪比（SNR）、DNSMOS评分、年龄、性别以及字符级时间戳等信息，未来将新增更多元数据标签。
* 涵盖短视频、娱乐、直播、纪录片、有声书、戏曲、访谈、新闻等十大领域。
<div align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/figs/domain.png?raw=true" width="300" style="display:inline-block; margin-right:10px;" />
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/blob/main/src/figs/quality_distribution.jpg?raw=true" width="300" style="display:inline-block;" />
</div>
### 元数据格式
我们将所有音频元数据以标准化JSON格式存储，核心字段包括`utt_id`（单条音频片段的唯一标识符）、`rover_result`（三种自动语音识别（Automatic Speech Recognition, ASR）转录结果的ROVER融合结果）、`confidence`（文本转录置信度评分）、`jyutping_confidence`（粤式拼音转录置信度评分）以及`duration`（音频时长）；说话人属性字段包含`speaker_id`、`gender`（性别）和`age`（年龄）；音频质量评估指标包含`sample_rate`（采样率）、`DNSMOS`以及`SNR`（信噪比）；时间戳信息字段`timestamp`通过`start`和`end`精准记录片段边界；`meta_info`字段下的扩展元数据包含`program`（节目名称）、`region`（地域信息）、`link`（原始内容链接）以及`domain`（领域分类）。
#### 📂 目录结构
WenetSpeech-Chuan
├── metadata.jsonl
├── .gitattributes
└── README.md
#### 数据示例
###### metadata.jsonl 文件示例
{<br>
"utt": "音频片段唯一标识符", <br>
"filename": "音频文件名（字符串类型）", <br>
"text": "语音转录文本（字符串类型）", <br>
"domain": "领域分类列表（字符串数组类型）", <br>
"gender": "说话人性别（字符串类型）", <br>
"age": "说话人年龄标签（整数范围，例如：中年（36~59岁））", <br>
"wvmos": "音频质量评分（浮点类型）", <br>
"confidence": "转录文本置信度（取值范围0-1，字符串类型）", <br>
"emotion": "说话人情感标签（字符串类型，例如：愤怒）", <br>
} <br>
**示例：**
{ <br>
"utt": "013165495633_09mNC_9_5820", <br>
"filename": "013165495633_09mNC_9_5820.wav", <br>
"text": "还是选二手装好了的别墅诚心入如意的直接入住的好好", <br>
"domain": [ <br>
"短视频" <br>
], <br>
"gender": "男性", <br>
"age": "青年", <br>
"wvmos": 2.124380588531494, <br>
"confidence": 0.8333, <br>
"emotion": "愤怒", <br>
} <br>
### 数据集使用方式
您可通过元数据文件（`metadata.json`）中的`link`字段获取原始视频源，亦可依据`timestamps`字段对音频进行切分以提取对应语音片段。如需获取预处理后的音频数据，请通过下方联系方式与我们取得联系。
## 联系方式
若您有任何疑问或合作意向，可通过以下邮箱联系我们的研究团队：yhdai@mail.nwpu.edu.cn 或 ziyu_zhang@mail.nwpu.edu.cn。
您也可加入我们的微信技术交流群，获取最新动态、开展技术讨论，并且如前文所述，可通过该群获取预处理后的音频数据。
<p align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Chuan/raw/main/src/figs/wechat.jpg" width="300" alt="微信交流群二维码"/>
<em>扫码加入我们的微信讨论群</em>
</p>
<p align="center">
<img src="https://github.com/ASLP-lab/WenetSpeech-Yue/raw/main/figs/npu@aslp.jpeg" width="300" alt="公众号二维码"/>
</p>
提供机构：
maas
创建时间：
2025-09-24
