Music and Audio Computing Lab

Main menu: Home | People | Research | Publications |

Singing Voice Conversion and Synthesis

The goal of this project is modifying singing voices or generating them from score information using various signal processing and machine learning methods. Specifically, we are interested in modeling natural expressions of a target singer including vibrato, note timings, dynamics and other singing articulations.



Here are the lists of publications.


Singing Voice Synthesis

  • Korean Singing Voice Synthesis Based on Auto-Regressive Boundary Equilibrium GAN
    Soonbeom Choi, Wonil Kim, Saebyul Park, Sangeon Yong, and Juhan Nam
    Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2020 [pdf] [website] [code]
  • Conditional GAN 기반 가창 합성 시스템
    최순범, 남주한
    한국 음성 학회, 2019

Singing Voice Conversion

  • 가창 음원의 음악적 표현 이식 시스템
    용상언, 남주한
    한국 음성 학회, 2019
  • Singing Expression Transfer from One Voice to Another for a Given Song
    Sangeon Yong and Juhan Nam
    Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018 [pdf] [demo]

Funding

We have received the following funds to support this research.

  • Industrial Technology Innovation Program, Ministry of Trade, Industry & Energy , 2017-2020
  • Development of Smart Korean Traditional Music Editing Tool, Korea Creative Content Agency, 2016-2018