in Korea, an artificial intelligence can now modify the voice to translate songs into all languages

The software, developed by Supertsone and acquired by Hybe Studios, has stored thousands of voices in dozens of languages ​​and then uses AI to correct the pronunciation of its customers.

Have you always dreamed of a career in a K-pop group, but your Korean level is close to zero? This is no longer a problem thanks to the software developed by South Korean studios. They dominate the pop industry with their boy bands and girl bands like BTS or Blackpink: their albums are top sellers in Asia, but also in the rest of the world, even if the titles are in Korean and sometimes in English .

>>Music, plastic surgery and politics: how South Korea conquered the world with its K-pop

But now, Seoul studios want to release albums in multiple languages ​​simultaneously, and for that, they can count on artificial intelligence. Using software, the artist can record their song, which is then translated perfectly into the language of all major entertainment markets.

A song simultaneously translated into six languages

And this software was developed by Hybe studio – the production company of BTS group. Two years ago, Hybe bought the company Supertsone, which specializes in voice design using artificial intelligence. This start-up has stored thousands of voices in dozens of languages ​​to identify sounds, accents, intonations practiced all over the world, and then uses artificial intelligence to correct the pronunciation of its customers.

The Korean artist who played the game is Lee Hyun. He was part of a boy band in the late 2000s and then had a more up and down solo career. He aspired to make a comeback, and therefore accepted Hybe’s proposal. His song is therefore simultaneously translated into English, Spanish, Japanese or even Vietnamese, in addition to his native language, Korean.

Concretely, he sang, in the studio, in phonetics, in the six languages ​​so that the tone of his voice was faithful, but he did not need to really work on these foreign languages. Then, Supertone’s software corrected its pronunciation so that it more closely resembles that of natives of all the markets targeted by the company. His song even includes parts sung by female voices: again, it is artificial intelligence that has feminized the singer’s voice.

A mixture with a more natural look

But why bother with a singer who has to be paid if artificial intelligence can do everything? This is the question that critics or fans ask themselves. Hybe replies that the mix between the human voice and the machine enhancements gives a more natural result than a fully synthetic voice.

We can already technically create titles without a singer or singer, but the rendering is different. Especially here, the artist is recognized, with a pre-existing fan base, which promises even more sales than a virtual star, to whom it is more difficult to become attached. For his part, Lee Hyun insists that he was able to faithfully communicate his art and his message. Its title is called masquerade“masquerade” in French.


source site-9