P6-08: Ethics of Singing Voice Synthesis: Perceptions of Users and Developers

Lee, Kyungyun, Hitt, Gladys, Terada, Emily, Lee, Jin Ha*

Subjects (starting with primary): Human-centered MIR -> human-computer interaction ; Philosophical and ethical discussions -> ethical issues related to designing and implementing MIR to

Presented Virtually: 4-minute short-format presentation


Singing Voice Synthesis (SVS) has recently garnered much attention as its quality has improved vastly with the use of artificial intelligence (AI), creating many opportunities for supporting music creators and listeners. Recently, there have been growing concerns about ethical issues related to AI development in general, and to AI-based SVS development specifically. Many questions remain unexplored about how to ethically develop and use such technology. In this paper, we investigate the perception of ethical issues related to SVS from the perspectives of two different groups: the general public and developers. We collected 3,075 user comments from YouTube videos showcasing various uses of SVS as part of a mainstream variety show. Additionally, we interviewed six researchers developing SVS technology. Through thematic analysis, we identify and discuss three different aspects related to ethical issues in SVS development, highlighting the similarities and differences between the perspectives of the general public and developers: (1) Use scenarios, (2) Attitudes towards development, and (3) Meaning of "Creativity", and (4) Concerns about human rights, intellectual property (IP) and legal issues.

Direct link to video