YouTube announced one of this year’s most significant artificial intelligence developments in the music industry this week when it launched its new experimental AI feature that will allow select users to make music clips featuring AI-generated vocals from superstars, including Demi Lovato, Sia and John Legend.
Dream Track officially launched on Thursday as an early beta test, available only to about 100 select creators in the U.S., and YouTube gave no indication if or when the new feature would be widely available. YouTube announced several new AI music tools this week — including one in development that could help turn songwriters’ thoughts into real music just by humming an idea — though voice clones were the most notable feature.
Both the record labels and YouTube itself have been careful to emphasize that Dream Track is an experiment and that it’s not immediately clear how the new feature will develop in subsequent months. But regardless of what happens next, Dream Track marks the first time that some of the world’s biggest artists and music companies have allowed their music to be used for voice clones – one of the most controversial uses of AI in music creation – a significant step that could pave the way for how the industry will deal with advancing technology.
The news brings many other implications and questions, some that can only be better answered in the coming months. Here’s a rundown of what you need to know about YouTube’s new AI music creator.
Voice clones have been going viral all year. Why does this matter now?
Although fans, content creators and professional songwriters have been tinkering with voice clones for months using AI software like UberDruck, these clones are generally unauthorized, and record labels have been pushing the rest of the industry to crack down on AI content, they say. . violates your song. YouTube’s new feature, on the other hand, is a direct collaboration with some of the industry’s biggest copyright holders.
AI has been among the most pressing topics in business this year and has the potential to be the most disruptive technology to hit the industry since Napster and peer-to-peer file sharing in the early 2000s. with AI’s ability to allow people to take artists’ works and imitate them without their consent, as well as to further flood the market with content and make it even harder for real musicians to break through the noise.
But instead of completely rebuking AI, the industry appears to be trying to avoid another disastrous Napster moment and be more proactive in developing the technology in a way that benefits it.
As WMG CEO Robert Kyncl — himself a former YouTube executive — told analysts during WMG’s earnings call Thursday morning, per billboard: “Imagine in the early 2000s if file-sharing companies came to the music industry and said, ‘would you like to try this new tool we built and see how it impacts the industry and how we can work together?’ It would have been incredible.”
YouTube’s new experiment isn’t the first time the industry’s biggest players have cautiously navigated AI. Sony Music hired an executive vice president of AI this summer to help the company navigate its approach around technology. Earlier this year, UMG partnered with generative music app Endel to create ambient tracks using stems provided by the company’s artists.
In August, UMG and YouTube announced a partnership to develop an AI Incubator, as well as a set of principles on how to advance music and AI. “Our challenge and opportunity as an industry is to establish effective tools, incentives and rewards – as well as rules of the road – that allow us to limit the potential downside of AI while promoting its promising upside,” said UMG CEO , Lucian Grainge, when the incubator was launched. .
Who is involved in the new experiment?
The nine artists participating in the Dream Track – Lovato, Sia, Legend, Charli XCX, Troye Sivan, Papoose, Charlie Puth, T-Pain and Alec Benjamin – are signed to Universal, Warner and independent music companies, including EMPIRE.
EMPIRE CEO Ghazi said in a statement to Rolling Stone that “we are always looking for innovative ways for our artists to participate in new technologies,” he said. “Partnering with YouTube Music and Google to develop music-centric AI tools is a positive step forward for our artists and music creators as a whole.”
Benjamin says Rolling Stone in an interview that YouTube invited him to take part in the test just over a month ago, after he told a friend who worked at the company that he had written a song loosely based on AI. He was intrigued by the concept of YouTube and felt that participating would give him the chance to be one of the first artists “to help discover and shape how it is used.”
“It doesn’t necessarily matter how I feel about AI, because that’s where things are going anyway. I don’t think there’s any point in fighting these things,” says Benjamin. “But I’m excited about it. I’m excited to be at the forefront of this new technology. I’m interested in seeing what the AI generates with my voice and what those things sound like, and how creators will interact with the AI version of my music.”
Benjamin expressed curiosity and some nervousness about how creators will use their AI voice clone in the coming months. “What’s cool about it is it gives you a certain degree of immortality when it comes to your voice. Your voice can be used over time, whether you are there to make new recordings or not. There’s something scary about it, but it’s also exciting,” says Benjamin. “What’s also scary is: What if people use my voice in a way that I don’t want them to?
“These are really interesting and scary questions, but I think that regardless of how the application of technology makes me feel, it’s going to happen anyway. I’m sure there will be things that will make me uncomfortable, and I don’t even know what they are yet,” he adds with a laugh. “This seems to be the new wave or maybe not, but we are testing things and about to find out. You can’t stop the wave, so I better ride it.”
So how does this work?
Dream Track uses generative AI trained on the vocals of participating artists. Creators with access to Dream Track can send text-based requests for suggested song ideas and select which artist AI they want to perform the song. Dream Track does the rest, releasing song clips up to 30 seconds long.
Tech YouTuber and journalist Cleo Abram, for example, used Dream Track to create a virtual Charlie Puth to sing about your dog.
YouTube did not provide details about what was used to train the AI artists. But a source familiar with the matter said Rolling Stone that UMG “allowed a limited subset of the company’s content to be trained for a limited time to facilitate the experiment.”
Benjamin, who signed a contract with Warner’s record label Elektra, says Rolling Stone that he had not recorded any new music for the experiment, but he was not told exactly what the content of the training on his AI was. This training almost certainly came from his previously recorded catalogue.
When asked about payment for his participation, he said he did not ask any questions about the monetization process and that these details would go through his label. Neither YouTube, UMG nor WMG provided details about the settlement payments.
Do AI developers really need industry permission to create their AI models? Who owns this song?
AI music presents significant legal questions, from who can own the music produced to what permissions are needed to train the AI.
Not surprisingly, many AI developers believe that training their AI models on copyrighted material should be considered fair use, which would allow them to avoid asking copyright holders for permission and paying copyright holders to use these materials. works. Equally unsurprisingly, music companies – who have built their businesses on decades of intellectual property – strongly reject this notion.
Ownership of AI-generated works is also unclear. The copyright office has ruled that purely machine-generated works cannot receive copyright protection, as humans were not sufficiently involved in the creation of the work. Does a song generated primarily by AI at the specific request of a human overcome this barrier?
