MuST-C

该数据集名为MuST-C，是一个用于训练和测试同声传译模型的 多语言语音翻译数据集。其中，验证集包含了开发集（分别为英语到德语的1423对，以及英语到西班牙语的1316对）；测试集则包含了tst-COMMON（分别为英语到德语的2641对，以及英语到西班牙语的2502对）。该数据集的规模涵盖了408小时（英语到德语）和504小时（英语到西班牙语）的语音数据，任务类型为同声传译。The dataset named MuST-C is a multilingual speech translation dataset designed for training and testing simultaneous speech translation models. Its validation set comprises development subsets, with 1423 pairs for English-to-German and 1316 pairs for English-to-Spanish respectively. The test set includes the tst-COMMON subset, which contains 2641 pairs for English-to-German and 2502 pairs for English-to-Spanish respectively. The dataset spans 408 hours of speech data for English-to-German and 504 hours for English-to-Spanish, and its core task is simultaneous speech translation.