zeio/baneks-speech

---
language:
- ru
- en
license: apache-2.0
tags:
- not-for-all-audiences
- art
- humour
- jokes
annotation_creators:
- crowdsourced
- original
- machine-generated
language_creators:
- crowdsourced
- original
- machine-generated
pretty_name: baneks-speech
size_categories:
- 10K<n<100K
task_categories:
- text-to-speech
- automatic-speech-recognition
---

# Dataset card for baneks-speech

## Table of contents

- [Dataset description](#dataset-description)
- [Dataset summary](#dataset-summary)
- [Dataset structure](#dataset-structure)
- [Dataset instance](#dataset-instance)
- [Dataset fields](#dataset-fields)

## Dataset description

- **Homepage:** [baneks-speech homepage]()
- **Repository:** [baneks-speech repository](https://huggingface.co/datasets/zeio/baneks-speech)
- **Point of contact:** [Zeio Nara](mailto:[e-mail])
- **Dataset version:** `10.10.2023`

### Dataset summary

This dataset contains speech generated for anekdotes parsed from a few vk social network communities using [stc model](https://cloud.speechpro.com/home).
The dataset corresponds to the **default** configuration from the [baneks](https://huggingface.co/datasets/zeio/baneks) dataset.
Since the dataset is regularly updated, there is no fixed number of entries, so stay tuned.

## Dataset structure

### Data instance

An example of an entry from the dataset is given below:

```json
{
'text': 'Сидят русский и казах Русский спрашивает: - Слушай, а у тебя жена есть? - Жок - Молодец! Я бы свою тоже сжег нахуй!',
'audio': {
'path': '/root/.cache/huggingface/datasets/downloads/extracted/e8abea39e83c61e4a60c5a4b0661dc044ae82cfcd84b26f966b87999f73ae92e/00476018.anekdotikategoriib.mp3',
'array': array([ 8.09500818e-07, 1.27653129e-06, 4.17583010e-07, ..., -8.52341486e-07, 9.19626189e-07, 1.67368569e-06]),
'sampling_rate': 22050
},
'artist': 'Vladimir_n',
'id': 476018,
'source': 'anekdotikategoriib'
}
```

### Data fields

Each dataset entry therefore consists of the following fields:

- `text` - text representation of the anecdote;
- `id` - id of the corresponding post;
- `source` - community name in which the corresponding post has been published;
- `artist` - identifier of the voice which was used for speech generation;
- `audio` - audio data read from an `mp3` file.

原始信息汇总

数据集卡片 for baneks-speech

数据集描述

数据集摘要

该数据集包含为从几个VK社交网络社区解析的轶事生成的语音，使用stc模型生成。数据集对应于baneks数据集的默认配置。由于数据集定期更新，没有固定的条目数量，请保持关注。

数据集结构

数据实例

以下是一个数据集条目的示例：

json { text: Сидят русский и казах Русский спрашивает: - Слушай, а у тебя жена есть? - Жок - Молодец! Я бы свою тоже сжег нахуй!, audio: { path: /root/.cache/huggingface/datasets/downloads/extracted/e8abea39e83c61e4a60c5a4b0661dc044ae82cfcd84b26f966b87999f73ae92e/00476018.anekdotikategoriib.mp3, array: array([ 8.09500818e-07, 1.27653129e-06, 4.17583010e-07, ..., -8.52341486e-07, 9.19626189e-07, 1.67368569e-06]), sampling_rate: 22050 }, artist: Vladimir_n, id: 476018, source: anekdotikategoriib }

数据字段

每个数据集条目包含以下字段：

• text - 轶事的文本表示；
• id - 对应帖子的ID；
• source - 发布对应帖子的社区名称；
• artist - 用于语音生成的声音标识符；
• audio - 从mp3文件读取的音频数据。