PJS: phoneme-balanced Japanese singing voice corpus
收藏资源简介:
PJS是一个由明治大学和东京大学信息科学与技术研究生院创建的日语歌唱语音平衡数据集,包含100条语音数据。数据集通过使用语音平衡的语音数据集来保证音素平衡,并采用CC BY-SA 4.0许可,确保数据可自由用于研究和商业用途。数据集不仅包含歌唱语音数据,还包含平行的语音数据和音乐信息,如旋律构成描述,适用于歌唱语音合成和自然语言音乐信息处理研究。此外,数据集还解决了现有数据集存在的版权和数据不平衡问题,为机器学习提供了易于训练的小型数据集。
PJS is a balanced Japanese singing speech dataset developed by Meiji University and the Graduate School of Information Science and Technology, The University of Tokyo, containing 100 speech samples. It ensures phonemic balance through a phonetically balanced dataset design, and is licensed under CC BY-SA 4.0, allowing free use for both research and commercial purposes. In addition to singing speech data, the dataset also includes parallel speech data and musical information such as melody composition descriptions, making it applicable to research in singing voice synthesis and natural language-driven musical information processing. Furthermore, this dataset resolves the copyright and data imbalance problems existing in current datasets, providing a compact and easily trainable small-scale dataset for machine learning.

- 1PJS: phoneme-balanced Japanese singing voice corpus明治大学和东京大学信息科学与技术研究生院 · 2020年



