MBW’s Stat Of The Week is a collection during which we spotlight a single knowledge level that deserves the eye of the worldwide music trade. Stat Of the Week is supported by Cinq Music Group, a technology-driven file label, distribution, and rights administration firm.
The usage of synthetic intelligence-created music simply moved up a gear.
We’re not speaking about AI in mere instrumental music manufacturing, however using machine studying to really mimic and even recreate human vocals – rendering the necessity for an actual singer out of date.
MBW first explored this matter final March, during which we analyzed the long-term implications of HYBE’s funding into (and subsequent acquisition of) Korea-based Synthetic Intelligence firm Supertone – which claims that its AI tech can create “a hyper-realistic and expressive voice [not] distinguishable from actual people”.
Now, over in China, issues have reached the subsequent stage: Tencent Music Leisure (TME) says that it has created and launched over 1,000 tracks containing vocals created by AI tech that mimics the human voice.
And get this: one in all these tracks has already surpassed 100 million streams.
In the course of the three months to finish of September, TME rolled out what it refers to as “patented voice synthesis know-how”, the Lingyin Engine. This tech, says TME, can “rapidly and vividly replicate singers’ voices to supply authentic songs of any fashion and language”.
A part of TME’s preliminary work utilizing the Lingyin Engine concerned growing “artificial voices in reminiscence of legendary artists” such because the late Teresa Teng, and the late Anita Mui.
(‘Resurrecting’ the voice of a deceased star is one thing HYBE’s Supertone gained quite a lot of media consideration for final yr: The corporate used its personal tech to recreate the voice of South Korean folks celebrity Kim Kwang-seok.)
Cussion Pang, TME’s Government Chairman, defined to analysts earlier right now (November 15) that TME used the Lingyin Engine to “pay tribute” to Anita Mui by “creating an AI code primarily based on her [voice]” for a brand new observe – Might You Be Handled Kindly By This World [English transation] – launched in help of the New Sunshine Charity Basis in China.
“[This track] has turn into the primary tune by an AI singer to be streamed over 100 million instances throughout the web.”
Cussion Pang, Tencent Music Leisure
Teresa Teng’s voice was recreated by TME/the Lingyin Engine to guide the observe Letter Not Despatched [English translation], launched earlier this yr to mark the anniversary of the Taiwanese star’s dying.
TME additionally confirmed right now (November 15) that – along with “paying tribute” to the vocals of lifeless artists through the Lingyin Engine – it has additionally created “an AI singer lineup with the voices of trending [i.e currently active] stars reminiscent of Yang Chaoyue, amongst others”.
As talked about, by the top of September, TME says it had created and launched over 1,000 songs with human-style vocals manufactured by the Lingyin Engine.
A kind of tracks has set the usual for recognition: TME’s Cussion Pang confirmed this morning to analysts {that a} model of 1 tune, which seems to be referred to as As we speak (English translation), “has turn into the primary tune by an AI singer to be streamed over 100 million instances throughout the web”.
The place might all of this go subsequent?
For one factor, the thoughts inevitably wanders to the truth that over 100,000 tracks at the moment are being uploaded to main world music streaming companies each single day.
The place might that determine scale as much as if limitless tracks at the moment are being born with uncanny human-esque AI vocals?
It’s additionally price remembering what Choi Hee-doo, the COO of Supertone – that’s the Korean AI voice-creation platform – stated final yr, when pondering how the tech may evolve.
“For instance, BTS is actually busy as of late, and it’d be unlucky if they’ll’t take part in content material as a result of lack of time,” the exec advised CNN.
“So, if BTS makes use of our know-how when making video games or audiobooks or dubbing an animation… they wouldn’t essentially need to file [that audio live] in individual.”
Apparently, Okay-pop firm HYBE’s greatest natural income driver in Q3 2022 was its Artist ‘Oblique-involvement’ enterprise line, which sees the title and likeness of celebrity artists reminiscent of BTS utilized in different areas like video games and promoting with out requiring the band’s lively participation.
HYBE has now doubled down on its AI-generated voice plans, by totally buying Supertone in October in a $32 million deal.
Certainly, when HYBE confirmed that BTS could be enlisting within the military final month, HYBE CEO Jiwon Park, breaking down HYBE’s technique with out its top-earning act, stated that the corporate’s newly acquired AI voice startup will “function a key piece of the know-how sphere we intention to create”.
He added: “HYBE plans to unveil new content material and companies to our followers by combining our content-creation capabilities with Supertone’s AI-based talking and singing vocal synthesis know-how.”
Along with HYBE and TME, there’s additionally one other large of the know-how and music world that appears to be betting large on AI: TikTok and its mum or dad firm ByteDance.
Again in July 2019, ByteDance acquired Jukedeck, a UK-based AI Music startup that specialised in creating royalty-free music for user-generated on-line movies.
In Might, ByteDance launched Mawf, a machine-learning pushed music-making app that analyses incoming audio alerts after which “re-renders” these alerts utilizing what it says is machine studying fashions of musical devices. ByteDance additionally lately launched a music creation app in China referred to as ‘Sponge Band’ in response to Tech Planet.
This yr, as first reported by MBW, the corporate has been doubling down on its AI-powered music-making ambitions through a hiring spree for AI music specialists.
TikTok is particularly (and at present) hiring for a Analysis Scientist in Speech Synthesis in California. TikTok says that this individual will “lead analysis to advance science and know-how in Pure Language Processing and Speech Processing (e.g., Speech Synthesis, ASR)”.
They may even “analysis, mannequin, design, develop and consider novel machine studying fashions and algorithms”.
TikTok says that this workforce’s focus is “on cutting-edge R&D in areas like speech & audio, music processing, pure language understanding and multimodal deep studying”.
Might TikTok – which runs its personal artist distribution service SoundOn, and is reportedly making ready to increase its Resso music service into extra markets – launch tracks within the close to future (similar to Tencent Music) both totally created by AI, or that includes ‘AI artificial voices’?
If it did, what wouldn’t it imply for its relationship with the music trade?
