K-pop label HYBE uses AI to overcome language barriers.
In a dimly lit recording studio in Seoul, the K-pop music label responsible for BTS is embracing artificial intelligence (AI) to create a multilingual song. HYBE, South Korea’s largest music label, utilised AI to blend the voice of a South Korean singer, “Midnatt,” with native speakers in five other languages. The resulting track, titled Masquerade, was released in May in Korean, English, Spanish, Chinese, Japanese, and Vietnamese. While some K-pop singers have previously released songs in English and Japanese alongside their native Korean, this simultaneous six-language release is a global first, according to HYBE, and could potentially pave the way for other popular acts to adopt the technology.
Lee Hyun, also known as “MidNATT,” recorded the song in each language despite only having limited proficiency in English and Chinese alongside Korean. Native speakers performed the lyrics, which were later combined seamlessly with the singer’s voice using HYBE’s in-house AI music technology, named Chung.
AI’s influence on the music industry is on the rise, coinciding with the Grammy Awards’ implementation of new rules for AI technology usage and the prevalence of AI-generated mash-ups on social media. The technology used in Midnight’s song employs deep learning powered by the “Neural Analysis and Synthesis (NANSY)” framework, developed by Supertone. This approach results in a more natural sound than non-AI software, according to Supertone’s chief operating officer, Choi Hee-doo.
HYBE acquired Supertone for 45 billion won ($36 million) in January, and they plan to make some of the AI technology used in MIDNATT’s song accessible to creators and the public, though specifics regarding potential fees were not provided.
While AI music technology is not entirely new, its application in creating multilingual songs is an innovative approach. Valerio Velardo, director of The Sound of AI, a Spain-based consulting service for AI music and audio, highlights that not only professional musicians but also a broader population can benefit from AI music technology in the long term, comparing it to Instagram’s impact on picture creation.
Presently, HYBE’s pronunciation correction technology requires “weeks or months” to function effectively. However, as the process becomes faster, it could serve broader purposes, such as interpreting in video conferences, according to Choi Jin-woo, the producer of MidNATT’s Masquerade, known as Hitchhiker.
About The Author
