South Korean pop music label HYBE removes the language barrier with the help of AI, creating the world's first six-language debut of a new song
CTOnews.com, July 19 (Xinhua)-- producers of HYBE, South Korea's largest pop music label known as BTS bulletproof Youth Corps, are using AI technology to remove the language barrier, Reuters reported. In May, the company successfully released a new single "Masquerade" for pop singer MIDNATT in six languages, including Korean, English, Spanish, Chinese, Japanese and Vietnamese.
▲ Tu Yuan HYBE official YouTube account some South Korean pop singers have previously released songs in English and Japanese in addition to their native Korean, but according to HYBE, the application of the new technology to release six languages at the same time is the first in the world and may pave the way for it to be popular with more audiences in this way.
"We will first listen to the reactions and voices of fans and then decide what to do next," Chung Wooyong, head of HYBE's interactive media division, said in an interview at the company's studio.
MIDNATT, 40, speaks only limited English and Chinese except Korean, but he successfully recorded songs in six languages with the help of AI technology. He said native speakers in all languages read the lyrics first and then seamlessly combined them with his singing and melody with the help of HYBE's internal AI music technology.
"We divide a sound into different components-pronunciation, timbre, pitch and volume," said HYBE's technical director. "We studied the pronunciation associated with tongue movement and used our imagination to try what results our technology can produce."
In a demonstration, the word "twisted" in the English lyrics was added with elongated vowels to make it sound more natural, while the singer's voice did not change significantly.
HYBE bought Supertone in January for 45 billion won (CTOnews.com Note: currently about 256 million yuan), which developed a neural analysis and synthesis (NANSY) framework that supports deep learning and makes songs sound more natural than songs written using non-AI software.
