mythicinfinity/libritts_r_opus

---
license: cc-by-4.0
task_categories:
- text-to-speech
language:
- en
size_categories:
- 10K<n<100K
configs:
- config_name: dev
data_files:
- split: dev.clean
path: &id001
- data/dev/dev.clean/data-00000.parquet
- data/dev/dev.clean/data-00001.parquet
- data/dev/dev.clean/data-00002.parquet
- data/dev/dev.clean/data-00003.parquet
- data/dev/dev.clean/data-00004.parquet
- data/dev/dev.clean/data-00005.parquet
- data/dev/dev.clean/data-00006.parquet
- data/dev/dev.clean/data-00007.parquet
default: true
- config_name: clean
data_files:
- split: dev.clean
path: *id001
- split: test.clean
path: &id003
- data/clean/test.clean/data-00000.parquet
- data/clean/test.clean/data-00001.parquet
- data/clean/test.clean/data-00002.parquet
- data/clean/test.clean/data-00003.parquet
- data/clean/test.clean/data-00004.parquet
- data/clean/test.clean/data-00005.parquet
- data/clean/test.clean/data-00006.parquet
- data/clean/test.clean/data-00007.parquet
- split: train.clean.100
path: &id005
- data/clean/train.clean.100/data-00000.parquet
- data/clean/train.clean.100/data-00001.parquet
- data/clean/train.clean.100/data-00002.parquet
- data/clean/train.clean.100/data-00003.parquet
- data/clean/train.clean.100/data-00004.parquet
- data/clean/train.clean.100/data-00005.parquet
- data/clean/train.clean.100/data-00006.parquet
- data/clean/train.clean.100/data-00007.parquet
- split: train.clean.360
path: &id006
- data/clean/train.clean.360/data-00000.parquet
- data/clean/train.clean.360/data-00001.parquet
- data/clean/train.clean.360/data-00002.parquet
- data/clean/train.clean.360/data-00003.parquet
- data/clean/train.clean.360/data-00004.parquet
- data/clean/train.clean.360/data-00005.parquet
- data/clean/train.clean.360/data-00006.parquet
- data/clean/train.clean.360/data-00007.parquet
- config_name: other
data_files:
- split: dev.other
path: &id002
- data/other/dev.other/data-00000.parquet
- data/other/dev.other/data-00001.parquet
- data/other/dev.other/data-00002.parquet
- data/other/dev.other/data-00003.parquet
- data/other/dev.other/data-00004.parquet
- data/other/dev.other/data-00005.parquet
- data/other/dev.other/data-00006.parquet
- data/other/dev.other/data-00007.parquet
- split: test.other
path: &id004
- data/other/test.other/data-00000.parquet
- data/other/test.other/data-00001.parquet
- data/other/test.other/data-00002.parquet
- data/other/test.other/data-00003.parquet
- data/other/test.other/data-00004.parquet
- data/other/test.other/data-00005.parquet
- data/other/test.other/data-00006.parquet
- data/other/test.other/data-00007.parquet
- split: train.other.500
path: &id007
- data/other/train.other.500/data-00000.parquet
- data/other/train.other.500/data-00001.parquet
- data/other/train.other.500/data-00002.parquet
- data/other/train.other.500/data-00003.parquet
- data/other/train.other.500/data-00004.parquet
- data/other/train.other.500/data-00005.parquet
- data/other/train.other.500/data-00006.parquet
- data/other/train.other.500/data-00007.parquet
- config_name: all
data_files:
- split: dev.clean
path: *id001
- split: dev.other
path: *id002
- split: test.clean
path: *id003
- split: test.other
path: *id004
- split: train.clean.100
path: *id005
- split: train.clean.360
path: *id006
- split: train.other.500
path: *id007
---
# Dataset Card for LibriTTS-R

<!-- Provide a quick summary of the dataset. -->

LibriTTS-R [1] is a sound quality improved version of the LibriTTS corpus
(http://www.openslr.org/60/) which is a multi-speaker English corpus of approximately
585 hours of read English speech at 24kHz sampling rate, published in 2019.

## Transcoding notes

- Source dataset: `mythicinfinity/libritts_r`
- Target codec: `opus` `48kbps`

## Overview

This is the LibriTTS-R dataset, adapted for the `datasets` library.

## Usage

### Splits

There are 7 splits (dots replace dashes from the original dataset, to comply with hf naming requirements):

- dev.clean
- dev.other
- test.clean
- test.other
- train.clean.100
- train.clean.360
- train.other.500

### Configurations

There are 3 configurations, each which limits the splits the `load_dataset()` function will download.

The default configuration is "all".

- "dev": only the "dev.clean" split (good for testing the dataset quickly)
- "clean": contains only "clean" splits
- "other": contains only "other" splits
- "all": contains only "all" splits

### Example

Loading the `clean` config with only the `train.clean.360` split.
```
load_dataset("blabble-io/libritts_r", "clean", split="train.clean.100")
```

Streaming is also supported.
```
load_dataset("blabble-io/libritts_r", streaming=True)
```

### Columns

```
{
"audio": datasets.Audio(sampling_rate=24_000),
"text_normalized": datasets.Value("string"),
"text_original": datasets.Value("string"),
"speaker_id": datasets.Value("string"),
"path": datasets.Value("string"),
"chapter_id": datasets.Value("string"),
"id": datasets.Value("string"),
}
```

### Example Row

```
{
'audio': {
'path': '/home/user/.cache/huggingface/datasets/downloads/extracted/5551a515e85b9e463062524539c2e1cb52ba32affe128dffd866db0205248bdd/LibriTTS_R/dev-clean/3081/166546/3081_166546_000028_000002.wav',
'array': ...,
'sampling_rate': 24000
},
'text_normalized': 'How quickly he disappeared!"',
'text_original': 'How quickly he disappeared!"',
'speaker_id': '3081',
'path': '/home/user/.cache/huggingface/datasets/downloads/extracted/5551a515e85b9e463062524539c2e1cb52ba32affe128dffd866db0205248bdd/LibriTTS_R/dev-clean/3081/166546/3081_166546_000028_000002.wav',
'chapter_id': '166546',
'id': '3081_166546_000028_000002'
}
```

## Dataset Details

### Dataset Description

- **License:** CC BY 4.0

### Dataset Sources [optional]

<!-- Provide the basic links for the dataset. -->

- **Homepage:** https://www.openslr.org/141/
- **Paper:** https://arxiv.org/abs/2305.18802

## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

```
@ARTICLE{Koizumi2023-hs,
title = "{LibriTTS-R}: A restored multi-speaker text-to-speech corpus",
author = "Koizumi, Yuma and Zen, Heiga and Karita, Shigeki and Ding,
Yifan and Yatabe, Kohei and Morioka, Nobuyuki and Bacchiani,
Michiel and Zhang, Yu and Han, Wei and Bapna, Ankur",
abstract = "This paper introduces a new speech dataset called
``LibriTTS-R'' designed for text-to-speech (TTS) use. It is
derived by applying speech restoration to the LibriTTS
corpus, which consists of 585 hours of speech data at 24 kHz
sampling rate from 2,456 speakers and the corresponding
texts. The constituent samples of LibriTTS-R are identical
to those of LibriTTS, with only the sound quality improved.
Experimental results show that the LibriTTS-R ground-truth
samples showed significantly improved sound quality compared
to those in LibriTTS. In addition, neural end-to-end TTS
trained with LibriTTS-R achieved speech naturalness on par
with that of the ground-truth samples. The corpus is freely
available for download from
\textbackslashurl\{http://www.openslr.org/141/\}.",
month = may,
year = 2023,
copyright = "http://creativecommons.org/licenses/by-nc-nd/4.0/",
archivePrefix = "arXiv",
primaryClass = "eess.AS",
eprint = "2305.18802"
}
```

许可协议：知识共享署名4.0（CC BY 4.0）
任务类别：文本转语音（text-to-speech）
语言：英语（en）
数据规模：10000 < 样本量 < 100000
配置项：
- 配置名称：dev
数据文件：
- 划分：dev.clean
路径：&id001
- data/dev/dev.clean/data-00000.parquet
- data/dev/dev.clean/data-00001.parquet
- data/dev/dev.clean/data-00002.parquet
- data/dev/dev.clean/data-00003.parquet
- data/dev/dev.clean/data-00004.parquet
- data/dev/dev.clean/data-00005.parquet
- data/dev/dev.clean/data-00006.parquet
- data/dev/dev.clean/data-00007.parquet
为默认配置
- 配置名称：clean
数据文件：
- 划分：dev.clean
路径：*id001
- 划分：test.clean
路径：&id003
- data/clean/test.clean/data-00000.parquet
- data/clean/test.clean/data-00001.parquet
- data/clean/test.clean/data-00002.parquet
- data/clean/test.clean/data-00003.parquet
- data/clean/test.clean/data-00004.parquet
- data/clean/test.clean/data-00005.parquet
- data/clean/test.clean/data-00006.parquet
- data/clean/test.clean/data-00007.parquet
- 划分：train.clean.100
路径：&id005
- data/clean/train.clean.100/data-00000.parquet
- data/clean/train.clean.100/data-00001.parquet
- data/clean/train.clean.100/data-00002.parquet
- data/clean/train.clean.100/data-00003.parquet
- data/clean/train.clean.100/data-00004.parquet
- data/clean/train.clean.100/data-00005.parquet
- data/clean/train.clean.100/data-00006.parquet
- data/clean/train.clean.100/data-00007.parquet
- 划分：train.clean.360
路径：&id006
- data/clean/train.clean.360/data-00000.parquet
- data/clean/train.clean.360/data-00001.parquet
- data/clean/train.clean.360/data-00002.parquet
- data/clean/train.clean.360/data-00003.parquet
- data/clean/train.clean.360/data-00004.parquet
- data/clean/train.clean.360/data-00005.parquet
- data/clean/train.clean.360/data-00006.parquet
- data/clean/train.clean.360/data-00007.parquet
- 配置名称：other
数据文件：
- 划分：dev.other
路径：&id002
- data/other/dev.other/data-00000.parquet
- data/other/dev.other/data-00001.parquet
- data/other/dev.other/data-00002.parquet
- data/other/dev.other/data-00003.parquet
- data/other/dev.other/data-00004.parquet
- data/other/dev.other/data-00005.parquet
- data/other/dev.other/data-00006.parquet
- data/other/dev.other/data-00007.parquet
- 划分：test.other
路径：&id004
- data/other/test.other/data-00000.parquet
- data/other/test.other/data-00001.parquet
- data/other/test.other/data-00002.parquet
- data/other/test.other/data-00003.parquet
- data/other/test.other/data-00004.parquet
- data/other/test.other/data-00005.parquet
- data/other/test.other/data-00006.parquet
- data/other/test.other/data-00007.parquet
- 划分：train.other.500
路径：&id007
- data/other/train.other.500/data-00000.parquet
- data/other/train.other.500/data-00001.parquet
- data/other/train.other.500/data-00002.parquet
- data/other/train.other.500/data-00003.parquet
- data/other/train.other.500/data-00004.parquet
- data/other/train.other.500/data-00005.parquet
- data/other/train.other.500/data-00006.parquet
- data/other/train.other.500/data-00007.parquet
- 配置名称：all
数据文件：
- 划分：dev.clean
路径：*id001
- 划分：dev.other
路径：*id002
- 划分：test.clean
路径：*id003
- 划分：test.other
路径：*id004
- 划分：train.clean.100
路径：*id005
- 划分：train.clean.360
路径：*id006
- 划分：train.other.500
路径：*id007

# LibriTTS-R 数据集卡片
LibriTTS-R[1]是LibriTTS语料库（http://www.openslr.org/60/）的音质增强版本，该语料库是2019年发布的多说话人英语语料库，包含约585小时的24kHz采样率朗读英语语音数据。

## 转码说明
- 源数据集：`mythicinfinity/libritts_r`
- 目标编码格式：Opus，48kbps

## 概述
本数据集为适配`datasets`库的LibriTTS-R数据集。

## 使用方法
### 数据集划分
本数据集包含7种划分（为符合Hugging Face命名规范，将原数据集的横杠替换为点号）：
- dev.clean
- dev.other
- test.clean
- test.other
- train.clean.100
- train.clean.360
- train.other.500

### 配置项
本数据集共包含4种配置项，每种配置项均可限制`load_dataset()`函数需下载的数据集划分。默认配置为`all`。
- `dev`：仅包含`dev.clean`划分（适合快速测试数据集）
- `clean`：仅包含所有`clean`类划分
- `other`：仅包含所有`other`类划分
- `all`：包含全部数据集划分

### 使用示例
加载`clean`配置下的`train.clean.100`划分：
python
load_dataset("blabble-io/libritts_r", "clean", split="train.clean.100")

支持流式加载：
python
load_dataset("blabble-io/libritts_r", streaming=True)

### 数据字段结构
python
{
"audio": datasets.Audio(sampling_rate=24_000),
"text_normalized": datasets.Value("string"),
"text_original": datasets.Value("string"),
"speaker_id": datasets.Value("string"),
"path": datasets.Value("string"),
"chapter_id": datasets.Value("string"),
"id": datasets.Value("string"),
}

### 样本示例
python
{
'audio': {
'path': '/home/user/.cache/huggingface/datasets/downloads/extracted/5551a515e85b9e463062524539c2e1cb52ba32affe128dffd866db0205248bdd/LibriTTS_R/dev-clean/3081/166546/3081_166546_000028_000002.wav',
'array': ...,
'sampling_rate': 24000
},
'text_normalized': 'How quickly he disappeared!"',
'text_original': 'How quickly he disappeared!"',
'speaker_id': '3081',
'path': '/home/user/.cache/huggingface/datasets/downloads/extracted/5551a515e85b9e463062524539c2e1cb52ba32affe128dffd866db0205248bdd/LibriTTS_R/dev-clean/3081/166546/3081_166546_000028_000002.wav',
'chapter_id': '166546',
'id': '3081_166546_000028_000002'
}

## 数据集详情
### 数据集描述
- **许可协议**：知识共享署名4.0（CC BY 4.0）

### 数据集来源（可选）
- **主页**：https://www.openslr.org/141/
- **论文**：https://arxiv.org/abs/2305.18802

## 引用信息
bibtex
@ARTICLE{Koizumi2023-hs,
title = "{LibriTTS-R}: A restored multi-speaker text-to-speech corpus",
author = "Koizumi, Yuma and Zen, Heiga and Karita, Shigeki and Ding,
Yifan and Yatabe, Kohei and Morioka, Nobuyuki and Bacchiani,
Michiel and Zhang, Yu and Han, Wei and Bapna, Ankur",
abstract = "本文介绍了一款专为文本转语音（text-to-speech, TTS）任务设计的新型语音数据集"LibriTTS-R"。该数据集通过对LibriTTS语料库进行语音修复处理得到，LibriTTS语料库包含来自2456名说话人的585小时24kHz采样率语音数据及其对应文本。LibriTTS-R的样本与LibriTTS完全一致，仅音质得到了提升。实验结果表明，与LibriTTS相比，LibriTTS-R的基准样本音质有显著提升。此外，基于LibriTTS-R训练的神经端到端TTS系统的语音自然度可媲美基准样本。该语料库可从http://www.openslr.org/141/免费下载。",
month = may,
year = 2023,
copyright = "http://creativecommons.org/licenses/by-nc-nd/4.0/",
archivePrefix = "arXiv",
primaryClass = "eess.AS",
eprint = "2305.18802"
}