Supertone, the AI voice company owned by K-pop giant HYBE, has launched a new AI text-to-voice tool that it says generates “high-quality, expressive audio content.”
Dubbed Supertone Play, the new tool is available in English, Korean, and Japanese. It will be expanded within the year to include service in Spanish and Chinese, Supertone said in a statement on Tuesday (February 18).
Supertone Play has a library of 150 voice characters designed for multiple content creation uses, including natural conversation, professional narration for news and audiobooks, and “trend-driven meme voices.”
It also includes a voice-cloning feature that allows users to create a voice character from their own speech “in just 10 seconds,” according to a press release. The voice character is customizable to mimic different emotional tones.
For those worried the technology could be abused to create superfakes, Supertone offers assurances: Its new tool incorporates safeguards such as ‘Voice Captcha’ to ensure a creator’s voice – and not someone else’s – is being used to create custom characters, as well as ‘inaudible watermarks’ to make Supertone-created vocals traceable.
The company says the tool generates voices of sufficient quality to use in video games, commercials, videos, and podcasts. The tool is available now on the Supertone website, and is free to try until April 16.
“Play marks a major step forward in making AI voice technology more accessible and versatile for creators across the globe,” Supertone President Kyogu Lee said.
“With its highly expressive and natural speech synthesis, Play allows creators to bring their stories to life in new and dynamic ways. Whether for content creation, storytelling, or immersive experiences, we are excited to provide a tool that enhances creative expression through voice AI.”
“Play marks a major step forward in making AI voice technology more accessible and versatile for creators across the globe.”
Kyogu Lee, Supertone
Supertone said that during its beta-testing phase, the Play tool was used in more than 80 countries and generated around 260,000 audio sentences. Of those, around 120,000, or 42%, were downloaded, “a strong indicator that users actively incorporated the generated content into their creative projects.”
HYBE acquired Supertone in 2022, with an eye to integrating its voice-creation tools into music.
Less than a year later, HYBE-signed artist MIDNATT released a track called Masquerade, which was billed as the first-ever track to feature AI-generated vocals in multiple languages. The track featured lyrics sung in Korean, English, Japanese, Chinese, Spanish, and Vietnamese.
Last year, the AI company released Supertone Shift, a tool that enables artists to alter and clone their voices in real time. The tool can be used, for instance, to turn a single artist’s voice into a chorus during a live performance.
Also last year, HYBE debuted a virtual pop group, SYNDI8, with vocals generated by Supertone.
In a demonstration of Supertone’s capabilities, during an earnings call last year, HYBE revealed that the investor presentation given earlier by then-CEO Jiwon Park was actually delivered by an AI voice tool that mimicked the executive’s vocals.Music Business Worldwide
