gigant/m-ailabs_speech_dataset_fr

M-AILABS语音数据集是第一个免费提供的大型数据集，可用于语音识别和语音合成的训练数据。大部分数据基于LibriVox和Project Gutenberg。训练数据包括近千小时的音频和格式化的文本文件。每个音频片段都提供了转录，片段长度从1到20秒不等。文本发表于1884年至1964年之间，属于公共领域。音频由LibriVox项目录制，也属于公共领域，除了乌克兰语部分。乌克兰语音频由Nash Format或Gwara Media提供，仅用于机器学习目的。数据集包含音频路径和句子，数据字段包括音频和句子。数据集未进行分割，所有数据都在训练集中。

The M-AILABS Speech Dataset is the first freely available large-scale dataset for use as training data in speech recognition and speech synthesis. Most of its data is sourced from LibriVox and Project Gutenberg. The training data includes nearly one thousand hours of audio and formatted text files. Each audio segment is provided with a transcription, with the length of the segments ranging from 1 to 20 seconds. The texts were published between 1884 and 1964 and are in the public domain. The audio recordings were made by the LibriVox project and are also in the public domain, with the exception of the Ukrainian language subset. The Ukrainian audio was provided by Nash Format or Gwara Media, and is exclusively intended for machine learning purposes. The dataset includes audio paths and sentences, and its data fields cover audio and sentences. The dataset has not been split, and all data is included in the training set.

原始信息汇总

数据集概述

数据集名称

M-AILABS Speech Dataset (French)

数据集描述

M-AILABS Speech Dataset 是一个免费提供的大型语音数据集，用于语音识别和语音合成训练。数据集包含近千小时的音频和相应的文本文件，文本发布于1884至1964年间，属于公共领域。音频主要由LibriVox项目录制，除乌克兰语外，均属于公共领域。

语言

• 法语

数据集结构

数据实例

每个数据点包括音频文件路径和对应的句子。

数据字段

• audio: 包含音频文件路径、解码后的音频数组和采样率的字典。
• sentence: 用户被提示朗读的句子。

数据分割

数据未被分割，所有内容均在"train"分割中，包含82825个音频片段及其相关句子。

贡献者

@gigant