VESUS

该数据集包含了来自12,593个文件中的253个短语，这些文件涉及多位说话者，专为情感语音合成而设计。此外，该数据集还包括10位说话者，其情感内容已被标注，并在实验中作为风格标记使用。该任务的目的是进行带有情感调校的文本到语音合成。

This dataset contains 253 phrases extracted from 12,593 files involving multiple speakers, and is specifically designed for emotional speech synthesis. Additionally, the dataset includes 10 speakers with annotated emotional content, which is utilized as style tokens in experiments. The goal of this task is to perform emotion-tuned text-to-speech synthesis.