vctk

VCTK语料库，包含约44小时的语音数据，由110位具有不同口音的英语使用者朗读。每位说话者朗读约400个句子，这些句子选自报纸、彩虹段落以及用于语音口音档案的引出段落。该语料库支持多种任务，包括自动语音识别和文本转语音，并提供说话人ID、录音、文本转录等信息。数据集采用CC-BY 4.0授权许可。The VCTK Corpus contains approximately 44 hours of speech data recorded by 110 English speakers with diverse accents. Each speaker reads around 400 sentences selected from newspapers, rainbow passages, and elicitation passages used for speech accent archives. This corpus supports multiple tasks including automatic speech recognition and text-to-speech, and provides information such as speaker IDs, audio recordings, and text transcriptions. The dataset is licensed under CC-BY 4.0.