Artificial intelligence (AI) audio corporations ElevenLabs unveiled a new AI dubbing model, "Dubbing v2," on the 16th.
The model addresses a key limitation of existing AI dubbing—loss of emotion and delivery. It analyzes the speaker's actual expressive style, including the original voice's emotion, tone, intonation, and pauses, and naturally reproduces it in more than 90 languages.
It also supports multilingual dubbing that goes beyond simple translation to preserve the immersion and character of the original content through automatic voice cloning, automatic voice timing adjustment, and context-based translation.
The company expects Dubbing v2 to support the global expansion of Korea's drama, films, webtoons, games, and YouTube content. An ElevenLabs official said, "Korea has powerful content assets that captivate viewers worldwide, but until now, translation, scriptwriting, voice acting, audio editing, and timing adjustments for global expansion required massive production processes and expenses," adding that Dubbing v2 will help shorten that work time.
Hong Sang-won, ElevenLabs' Korea lead, said, "Making translated voices feel as if the original speaker actually spoke that language has long been one of the biggest challenges in AI dubbing," adding, "The newly introduced Dubbing v2, with precise emotional recreation and excellent sync, will be a powerful bridge that helps domestic creators, marketers, and content corporations communicate more deeply and naturally with people around the world beyond language barriers."