cobrayyxx/FLEURS_INDO-ENG_Speech_Translation

这是一个用于语音翻译任务的印度尼西亚语到英语的数据集，来源于FLEURS项目。FLEURS是FLoRes机器翻译基准的语音版本，包含多种语言，其中印度尼西亚语部分包含约3561个话语和约12小时24分钟的音频数据。数据集经过预处理步骤，包括移除未使用的列、删除英语数据集中的重复行、基于id列合并英语转录与印度尼西亚语音频和转录、分割为训练集和验证集，并将音频列转换为音频对象。数据集结构包括训练集和验证集，分别包含2848和713个样本。

This is the Indonesia-to-English dataset for Speech Translation task. This dataset is acquired from FLEURS. Fleurs is the speech version of the FLoRes machine translation benchmark. Fleurs has many languages, one of which is Indonesia for about 3561 utterances and approximately 12 hours and 24 minutes of audio data. The dataset is preprocessed with steps including removing unused columns, removing duplicate rows, merging English transcriptions with Indonesian audio and transcriptions based on the id column, splitting into train and validation sets, and casting the audio column into an Audio object. The dataset is divided into a training set with 2848 samples and a validation set with 713 samples.