JSUT (Japanese speech corpus of Saruwatari Laboratory, the University of Tokyo) corpus

JSUT数据集是由东京大学Saruwatari实验室创建的日语语音数据集，旨在支持端到端语音合成研究。该数据集包含10小时的阅读风格语音数据，涵盖了日常使用日语字符的所有主要发音。数据集通过精心设计，包括多个子集，如基本5000、计数后缀26等，以覆盖不同语言和发音特征。创建过程中，数据集从Wikipedia和TANAKA语料库中收集句子，并手动补充缺失的发音。该数据集适用于学术和非商业研究，特别是针对日语的端到端语音合成技术，旨在解决日语语音处理中的复杂性问题。

The JSUT dataset is a Japanese speech dataset created by the Saruwatari Laboratory at the University of Tokyo, aiming to support end-to-end speech synthesis research. This dataset contains 10 hours of read-style speech data, covering all major pronunciations of daily-used Japanese characters. The dataset is meticulously designed with multiple subsets, such as Basic 5000, Counting Suffix 26, etc., to cover diverse linguistic and phonetic features. During its creation, sentences were collected from Wikipedia and the TANAKA Corpus, and missing pronunciations were manually supplemented. This dataset is suitable for academic and non-commercial research, especially for end-to-end speech synthesis technologies for Japanese, and aims to address the complexity issues in Japanese speech processing.