Musical artists no longer need to limit themselves to what suits their natural voice. Indeed, anyone with an artistic vision can now develop a singing voice to match it.
At last week’s Fortune Brainstorm AI Singapore conference, Kyogu Lee, founder of AI startup Supertone, demonstrated how his AI-powered tools can alter a person’s vocals.
Supertone’s model takes a voice recording and breaks it down into four different attributes: pitch, loudness, timbre, and linguistic content. A person’s timbre is what reflects their “vocal identity,” Lee explained. By isolating it and altering the other three attributes, Supertone can reconstruct different singing voices that are still unique to each person’s sound.
The tools can even work for those without singing experience—like Fortune Asia editor Nicholas Gordon, whose voice was transformed into a K-pop singer’s through Supertone’s tools.
Supertone’s programs can also create singing voices from scratch. Previously, music producers needed to find human artists whose sound perfectly matches their artistic ideas. Now, they don’t have to “resort to human personnel,” Lee said, since Supertone can easily “design unique voices for each of them.”
But Lee said he wants to work with, not replace, musical artists. “We see creators and artists as co-creators,” he explained. Supertone can help artists to experiment with new genres or styles if they don’t have the natural voice for it. The startup relies on artist feedback to refine and improve its technology.
“Listening is believing,” Lee said. Watch him demonstrate Supertone’s model on the mainstage of Fortune Brainstorm AI Singapore.
