We often think that automation can only take low-skilled jobs, and those working in art or science, for instance, are completely safe. But in reality, machines seem to be coming for art as well. We previously reported about an AI that created original paintings and images from text input — but what we’re going to tell you next may blow your mind.
Recently, one of the biggest music streaming service providers in China, Tencent Music Entertainment (TME) revealed that an AI-based voice synthesis system sang more than 1000 songs released by the company in the last three months. The AI called Lingyin Engine can mimic the voices of human singers and then produce original songs in those voices. Not only were the songs embraced without suspicion — but they were wildly popular.
The power of Lingyin Engine
So imagine, you are a songwriter, who always wanted one of their songs to be sung by the Late rockstar Elvis Presley. Lingyin Engine can make this happen — sounds crazy, but it’s true. The AI will sing your song in the synthetically produced voice of Elvis, just like the way you want.
Unfortunately, not everybody can get their hands on the Lingyin Engine. TME has patented the technology and now it’s using it to release songs in the voice of both deceased and currently active Chinese singers, who are popular among the listeners. The company also claims that one of its tracks titled “Today” has become the first AI-voiced song to cross 100 million streams, and has generated an estimated revenue of nearly $350,000 so far.
We couldn’t really track or verify Tencent’s claims, which come from an investor release but if it’s true, it could spell a big change for the music industry. These statements should be taken with a grain of salt, but they’re not that far-fetched, especially when you consider that earlier this year, an AI-powered rapper was signed to Capitol Records, and South Korea is also using AI to produce music “sung” by human artists.
What’s more fascinating about the Lingyin Engine is that once it has mimicked a singer’s voice, it can create numerous songs using that voice in any language and style. To companies like Tencent Music, which spent millions of dollars every year on singers, this may look like the best business opportunity as now they can create songs without singers. It would also open up a whole new range of songs that would have never been possible before — but at a cost.
Voice synthesis technology poses a serious threat to the livelihood of singers and voice artists. Take the case of Tencent’s 1000 AI-voiced songs. The company would have had to hire hundreds of singers to record the 1000 tracks if Lingyin Engine had not been available.
With the rise of AI singers, soon human singers might start copyrighting their voices because technologies like Lingyin Engine could make them obsolete anytime. Once it has a singer’s voice sample in its database, it won’t ever need the singer to produce a song.
Also, Tencent is not the only company that has thought of using AI for creating human-like synthetic voices.
AI voice synthesis is already in demand
Supertone, an audio tech startup in Seoul, South Korea claims to be the world’s first AI-powered voice synthesis technology. It was acquired by Hybe Corporation in October 2021 for a whopping $32 million. Since Hybe also owns the popular K-pop band BTS, there are rumors that Hybe could use Supertone’s AI to create songs in BTS’ voice, while the members of the band are away completing their compulsory military tenure.
Popular video hosting service TikTok also has plans to develop advanced machine learning models that could perform audio and speech synthesis. A company representative told MBW about TikTok’s plans to invest in “cutting-edge R&D in areas like speech & audio, music processing, natural language understanding, and multimodal deep learning”.
Currently, TME is busy releasing tracks in the synthetic voices of trending Chinese singers. Plus, it has multiple songs and ideas lined-up to be processed by the Lingyin Engine. The engine is probably an expert at mimicking singer voices because before the company made an official announcement about its 1000 AI songs. Out of its 587 million active users, nobody ever posted or reported that they were listening to a synthetic voice. The users were clearly unable to differentiate between real voices and AI voices.
The success of TME’s AI songs is a strong sign of technology’s dominance over humanity. You shouldn’t be surprised if you see an AI like the Lingyin Engine disrupting and reshaping the entire music industry in the future.
