Singing Voice Synthesis: towards the State of the Art | Dreamtonics Inc. from Tokyo, Japan

Abstract:
Synthesizing realistic singing voice has been the wildest dream for many music producers. Thanks to recent advances in deep learning this is finally becoming a reality, reshaping the way we think of and work with musics. In this presentation Kanru Hua, founder and chief researcher at Dreamtonics introduces to you practical applications of singing voice synthesis, followed by an in-depth exploration through the research that lead to this advancement, and an examination of challenges met in conditional generative modeling for complex distributions.

Bio:
Kanru Hua (华侃如) is a software engineer from Tokyo, Japan. He is the founder of Dreamtonics and is the creator of the Synthesizer V and Synthesizer V Studio voice synthesis applications. He received undergraduate education at University of Illinois Urbana-Champaign. In an interview with vocaloidnews.net (VNN), it was revealed that Kanru first thought of the idea for Synthesizer V in 2015 using inspiration from previous work on the Sinsy voice synthesis engine and the UTAU resampling program Moresampler. He released the first generation of Synthesizer V in December 2018.

Company Website: https://dreamtonics.com/
Product Demo: https://www.youtube.com/@DreamtonicsCoLtd