The Institute of International Speech and Language Technology of Nagoya Institute of Technology and Techno Speech Co., Ltd. announced that they have succeeded in jointly developing an AI singing voice synthesis system that can reproduce human voice quality, habits, and singing style with unprecedented accuracy.
Nagoya Institute of Technology and Techno Speech, which have jointly worked on research and development of voice synthesis and singing voice synthesis technology, have so far used voice synthesis on commercial karaoke equipment "JOYSOUND" and voice creation software "CeVIO Creative Studio".・ We have been promoting the introduction of singing voice synthesis technology.
In this study, by applying statistical machine learning technology such as deep learning, that is, AI technology, to a singing voice database of about 2 hours by a specific singer, the voice quality, habits, and habits of the singer We have developed a system to learn how to sing.When synthesizing the learned singing voice, it is only necessary to enter a score with arbitrary lyrics, and it is said that the singing voice of an ultra-high quality virtual singer that is indistinguishable from humans has been realized.The results of this research will be announced at the 2019 Spring Meeting of the Acoustical Society of Japan, which will be held in March 3.
Techno Speech is a venture company established with the aim of popularizing the world's most advanced voice-related technology developed mainly by Nagoya Institute of Technology.The company and the university are aiming to further increase the value of voice-related technology provision and services based on the results of this research.