Purvaxxx/ASR

该数据集包含了多种语言的语音和文本数据，包括印地语、马拉地语、泰卢固语和孟加拉语。数据集分为默认配置和针对印地语处理过的配置两种。默认配置中包含音频文件路径、文本、预测文本、音频时长、对齐分数等信息。印地语处理过的配置包含特征向量、标签、标签长度、长度、来源和英文文本等信息。数据集被划分为训练集、验证集和测试集。

The dataset contains speech and text data in multiple languages, including Hindi, Marathi, Telugu, and Bengali. There are two configurations: default and Hindi processed. The default configuration includes audio file paths, text, predicted text, audio duration, alignment scores, etc. The Hindi processed configuration includes feature vectors, labels, label lengths, lengths, source, and English text. The dataset is split into training, validation, and test sets.