资源简介:
---
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: speaker_id
dtype: int64
- name: age
dtype: int64
- name: sex
dtype: string
- name: dialect
dtype: string
- name: recording_datetime
dtype: string
splits:
- name: train
num_bytes: 55199435558.0
num_examples: 182605
- name: test
num_bytes: 8894080220.0
num_examples: 54747
download_size: 5358057252
dataset_size: 64093515778.0
size_categories:
- 100K<n<1M
license: cc0-1.0
task_categories:
- automatic-speech-recognition
- text-to-speech
language:
- da
pretty_name: NST-da
---
# Dataset Card for NST-da
## Dataset Description
- **Repository:** <https://www.nb.no/sprakbanken/en/resource-catalogue/oai-nb-no-sbr-55/>
- **Point of Contact:** [Dan Saattrup Nielsen](mailto:dan.nielsen@alexandra.dk)
- **Size of downloaded dataset files:** 5.36 GB
- **Size of the generated dataset:** 64.09 GB
- **Total amount of disk used:** 69.45 GB
### Dataset Summary
This dataset is an upload of the [NST Danish ASR Database (16 kHz) – reorganized](https://www.nb.no/sprakbanken/en/resource-catalogue/oai-nb-no-sbr-55/).
The training and test splits are the original ones.
### Supported Tasks and Leaderboards
Training automatic speech recognition is the intended task for this dataset. No leaderboard is active at this point.
### Languages
The dataset is available in Danish (`da`).
## Dataset Structure
### Data Instances
- **Size of downloaded dataset files:** 5.36 GB
- **Size of the generated dataset:** 64.09 GB
- **Total amount of disk used:** 69.45 GB
An example from the dataset looks as follows.
```
{
'audio': {
'path': 'dk14x404-05072000-1531_u0008121.wav',
'array': array([ 0.00265503, 0.00248718, 0.00253296, ..., -0.00030518,
-0.00035095, -0.00064087]),
'sampling_rate': 16000
},
'text': 'Desuden er der en svømmeprøve, en fremmedsprogstest samt en afsluttende samtale.',
'speaker_id': 404,
'age': 24,
'sex': 'Female',
'dialect': 'Storkøbenhavn',
'recording_datetime': '2000-07-05T15:31:14'
}
```
### Data Fields
The data fields are the same among all splits.
- `audio`: an `Audio` feature.
- `text`: a `string` feature.
- `speaker_id`: an `int64` feature.
- `age`: an `int64` feature.
- `sex`: a `string` feature.
- `dialect`: a `string` feature.
- `recording_datetime`: a `string` feature.
### Dataset Statistics
There are 183,205 samples in the training split, and 54,747 samples in the test split.
#### Speakers
There are 539 unique speakers in the training dataset and 56 unique speakers in the test dataset, where 54 of them are also present in the training set.
#### Age Distribution

#### Dialect Distribution

#### Sex Distribution

#### Transcription Length Distribution

## Dataset Creation
### Curation Rationale
There are not many large-scale ASR datasets in Danish.
### Source Data
The data originates from the now bankrupt company Nordisk språkteknologi (NST), whose data was transferred to the National Library of Norway, who subsequently released it into the public domain.
## Additional Information
### Dataset Curators
[Dan Saattrup Nielsen](https://saattrupdan.github.io/) from the [The Alexandra
Institute](https://alexandra.dk/) reorganised the dataset and uploaded it to the Hugging Face Hub.
### Licensing Information
The dataset is licensed under the [CC0
license](https://creativecommons.org/share-your-work/public-domain/cc0/).
配置项:
- 配置名称:default
数据文件:
- 拆分集:训练集(train),路径:data/train-*
- 拆分集:测试集(test),路径:data/test-*
数据集信息(dataset_info):
特征字段:
- 字段名:音频(audio),数据类型:
音频的采样率(sampling_rate)为16000
- 字段名:文本(text),数据类型:字符串(string)
- 字段名:说话人ID(speaker_id),数据类型:int64
- 字段名:年龄(age),数据类型:int64
- 字段名:性别(sex),数据类型:字符串
- 字段名:方言(dialect),数据类型:字符串
- 字段名:录音时间(recording_datetime),数据类型:字符串
拆分集:
- 拆分名称:train,占用字节数:55199435558.0,样本数:182605
- 拆分名称:test,占用字节数:8894080220.0,样本数:54747
下载总大小:5358057252,数据集生成后总大小:64093515778.0
样本规模分类:100K<n<1M
许可证:cc0-1.0(CC0 1.0公共领域授权)
任务类别:自动语音识别(automatic-speech-recognition)、文本转语音(text-to-speech)
语言:丹麦语(da)
美观名称:NST-da
# NST-da 数据集卡片
## 数据集描述
- **仓库地址**:<https://www.nb.no/sprakbanken/en/resource-catalogue/oai-nb-no-sbr-55/>
- **联系人**:[Dan Saattrup Nielsen](mailto:dan.nielsen@alexandra.dk)
- **下载数据集文件大小**:5.36 GB
- **生成后数据集总大小**:64.09 GB
- **总磁盘占用量**:69.45 GB
### 数据集摘要
本数据集为[NST丹麦语自动语音识别数据库(16 kHz)—— 重构版](https://www.nb.no/sprakbanken/en/resource-catalogue/oai-nb-no-sbr-55/)的上传版本。训练集与测试集的拆分方式沿用原始划分。
### 支持任务与基准排行榜
本数据集的预设任务为训练自动语音识别模型,当前暂无活跃的基准排行榜。
### 语言
本数据集为丹麦语(`da`)版本。
## 数据集结构
### 数据实例
- **下载数据集文件大小**:5.36 GB
- **生成后数据集总大小**:64.09 GB
- **总磁盘占用量**:69.45 GB
本数据集的一个样本示例如下:
{
'audio': {
'path': 'dk14x404-05072000-1531_u0008121.wav',
'array': array([ 0.00265503, 0.00248718, 0.00253296, ..., -0.00030518,
-0.00035095, -0.00064087]),
'sampling_rate': 16000
},
'text': 'Desuden er der en svømmeprøve, en fremmedsprogstest samt en afsluttende samtale.',
'speaker_id': 404,
'age': 24,
'sex': 'Female',
'dialect': 'Storkøbenhavn',
'recording_datetime': '2000-07-05T15:31:14'
}
### 数据字段
所有拆分集的字段均保持统一:
- `audio`:音频(Audio)特征
- `text`:字符串(string)特征
- `speaker_id`:int64类型说话人ID特征
- `age`:int64类型年龄特征
- `sex`:字符串类型性别特征
- `dialect`:字符串类型方言特征
- `recording_datetime`:字符串类型录音时间特征
### 数据集统计
训练集包含183205条样本,测试集包含54747条样本。
#### 说话人统计
训练集共有539名唯一说话人,测试集共有56名唯一说话人,其中54名说话人同时出现在训练集中。
#### 年龄分布

#### 方言分布

#### 性别分布

#### 转录文本长度分布

## 数据集构建
### 构建依据
当前丹麦语大规模自动语音识别数据集较为稀缺。
### 源数据
本数据集源自已破产的北欧语言技术公司Nordisk språkteknologi(NST),其数据已移交至挪威国家图书馆,并随后被释放至公共领域。
## 附加信息
### 数据集整理者
来自亚历山大研究所(The Alexandra Institute)的Dan Saattrup Nielsen重构了本数据集并上传至Hugging Face Hub。
### 许可信息
本数据集采用[CC0许可证(CC0 license)](https://creativecommons.org/share-your-work/public-domain/cc0/)进行授权。