Big Ocean’s first subunit, Big Ocean JJ, unveiled its English single “BUCKET HAT” at the AI for Good Global Summit 2025 in Geneva, Switzerland. Layering the voices of members PJ and Jiseok over English guide vocals, “BUCKET HAT” was brought to life through SUPERTONE’s AI voice conversion technology. Just as Big Ocean expanded its stage through the release of English singles and international tours, SUPERTONE’s technology has garnered attention for broadening the boundaries of creativity in the entertainment industry and enhancing the sense of immersion. We spoke with Studio Team Leader Youngguk Lee at SUPERTONE, who oversaw the Big Ocean JJ project, about the recent collaboration and the future of AI voice technology in entertainment.
In producing “BUCKET HAT,” the focus appears to have been on preserving the artists’ unique vocal tones while ensuring their English pronunciation sounded natural. Which SUPERTONE technology was most critical in achieving this?
Youngguk Lee: Our work on “BUCKET HAT” marked our first collaboration with Big Ocean JJ. So, we began by training on their vocal data and building a new model, which then served as the basis for applying SUPERTONE’s voice conversion technology. Because the song is composed entirely of English lyrics, our goal was to preserve the members’ unique vocal character while enhancing the clarity and naturalness of their English pronunciation. Using SUPERTONE’s Controllable Voice Conversion (CVC) technology, we carefully calibrated the output so that, while it drew on the pronunciation of the guide vocals, it still retained the distinct tone and individuality of the members.
How does SUPERTONE’s self-developed foundation model NANSY, which also underpins CVC technology, perform voice conversion?
Youngguk Lee: SUPERTONE’s foundation model NANSY (Neural Analysis & Synthesis) is essentially a framework that separates the voice into four controllable elements: timbre, linguistic features, pitch, and loudness. The MIDNATT project with Lee Hyun, which we unveiled two years ago, shared a similar approach—preserving the artist’s unique timbre while enhancing only the clarity of foreign-language pronunciation where needed. Over the past two years, NANSY has become far more refined, enabling us to capture vocal expressiveness and energy with even greater precision.
*Foundation model refers to an AI (artificial intelligence) model that is pre-trained to perform various tasks.
In his interview with Maeil Business Newspaper, SUPERTONE President Lee Kyogu said, “Our work is about helping people push past their limits in creative expression.” In that sense, this collaboration seems to go beyond simply converting the artists’ pronunciation—it feels like it’s about expanding their creative potential. How do you see SUPERTONE’s technology supporting artists like Big Ocean and others?
Youngguk Lee: I believe the significance of SUPERTONE’s technology lies in its ability to make possible what was once difficult to attempt. This collaboration with the members of Big Ocean JJ, too, was an opportunity to provide creators with a meaningful experience through SUPERTONE’s new technology. Looking ahead, our goal for CVC technology is to support artists who face challenges such as vocal nodules or changes in timbre, enabling them to preserve their own voice and continue expressing themselves in recorded music.
SUPERTONE has been expanding its collaborations beyond K-pop, working actively in film, drama, and more. Going forward, how do you envision SUPERTONE working with the entertainment industry?
Youngguk Lee: Among the many elements that make up content such as music and video, I believe voice and audio are especially important in creating a sense of immersion. That is why at SUPERTONE, we work to realize the intentions of creators while enriching their work in the process. Through our technology, we aim to offer creators new ways of expressing their stories, always keeping their perspective in mind as we research and apply our tools. Moving forward, we hope to continue developing these resources so that creators can make use of them to their full creative potential.
