English as a Second Language TTS dataset (ESLTTS)

该数据集专为研究在显著口音条件下评估说话者自适应的文本转语音任务而设计，由纯粹的非母语英语发言构成。数据集中的发言者涵盖了31种不同的母语背景，为每位发言者分别提供了适应集、测试集和开发集。该数据集规模包括134位非母语英语发言者贡献的37小时、41,000条语音。其任务定位于说话者自适应的文本转语音合成。

This dataset is developed for research on evaluating speaker-adaptive text-to-speech (TTS) under conditions of prominent foreign accents, and it is composed exclusively of non-native English speech. The speakers in this dataset span 31 distinct native language backgrounds, with separate adaptation sets, test sets, and development sets provided for each individual speaker. The dataset totals 37 hours of speech and 41,000 utterances contributed by 134 non-native English speakers. The core task targeted by this dataset is speaker-adaptive text-to-speech synthesis.