oovword/speech-translation-uk-en

该数据集由三个不同的数据集处理而成，用于语音翻译任务。数据集包含音频文件及其对应的乌克兰语和英语文本。数据集的预处理步骤包括音频格式转换、过滤特定性别的说话者、使用Azure翻译服务进行乌克兰语到英语的翻译等。数据集分为训练集、验证集和测试集，分别包含10390、2058和2828个样本。音频总时长分别为10小时45分钟12秒、1小时36分钟7秒和3小时1分钟28秒。

This dataset has been created for the speech translation task, processed from three different datasets. The dataset contains audio, Ukrainian, and English parallel sentences. The audio sampling rate is 16kHz. The dataset is divided into train, validation, and test sets, containing 10390, 2058, and 2828 samples respectively. The total audio duration is 10 hours 45 minutes 12 seconds, 1 hour 36 minutes 7 seconds, and 3 hours 1 minute 28 seconds respectively. The preprocessing of the dataset includes extracting Ukrainian and English parallel sentences from the original datasets, filtering female speakers, and translating using the Azure Translator service.