Just last year Many experts believed that an AI model capable of generating complete, high-fidelity songs from text prompts would not be available anytime soon, but there is now a growing military of competing music production models that do just that. The competition is on. Suno's v3 model, which was released to the public just a few weeks ago, has made amazing progress, especially in real and humanized vocals. And today, a formidable new competitor has emerged through the just-launched startup Udio. While the output of both companies appears to be roughly equivalent, some early users have noted that, on average, Udio's output sounds crisper than his Suno, potentially betraying the truck's mechanically created origins. This suggests that the sound may be less ambiguous.
Udio's product came together surprisingly quickly after it was founded last December by four former researchers from Google's AI research arm DeepMind. They are backed by various tech heavyweights, including 16z (aka Andreesen Horowitz) and Instagram co-founder and CTO Mike Krieger. “From the day we received the investment, we received tremendous support,” said Andrew Sanchez, one of the company's co-founders. “So the technical co-founders were able to get everything moving pretty quickly so we could start sort of preparing.”
Udio's list of early investors also includes notable music names such as Common, producer Tay Keith, industry veteran Steve Stoute's United Masters, and Will.i.am. In a press release, will.i.am, a longtime evangelist of AI's musical potential, gushed about the company's products, saying, “This is a brand new renaissance, and Udio is the tool for creativity in this era. Yes,” the artist said. , People consulted during product development. “Udio lets you bring your songs to life with AI and your imagination.”
Although neither company directly confirms or denies, there is good reason to believe that both Udio and Suno were trained on copyrighted music without their permission, a practice that Stevie Wonder recently reported It was condemned in an open letter from signatory Artist Rights Alliance. To Billie Eilish. The group reiterated its position in a new statement. rolling stone. Jen Jacobsen, executive director of the Artist Rights Alliance, said: “Using an artist's work without consent, credit or compensation is not only unethical, irresponsible and destructive of a cultural gift, it is also illegal. There is,” he said. (The question of whether copyrighted material can legally be used to train AI is currently being decided in various courts.) If this is not possible, rights holders should promptly take legal action to prevent this.”
in rolling stoneused a pre-release beta of Udio, and the service produced two separate songs in one day that apparently used a voice that was almost indistinguishable from that of the late Tom Petty. I was there. “We're introducing filters to detect cases where a voice is too similar to an existing artist,” says co-founder David Ding. “Of course, this filter isn't perfect. It's also an AI system and we'll continue to iterate on it. But that's clearly not what it was intended to do.” However, Udio plans to use voice cloning more formally in the future. There may be plans to offer it. The company's press release says plans are imminent to give artists “financial control over their vocal likenesses.”
Unlike Suno, which focuses on putting music production tools into the hands of the average consumer, Udio also sees itself as a tool for musicians, and its creative capabilities along those lines , the founders said, should alleviate creators' concerns about the use of training data. . “We have been guided by music people from the beginning,” Sanchez says. “And what that means is, we're making a product that allows those guys to make great music and, frankly, make money” on that music in the future. ” Sánchez emphasized the belief that the company's work is “transformative,” meaning that its models are creating something new from the training data, and that ChatGPT is committed to the use of copyrighted material. This argument is also used in the current court case.
AI blues song rolling stone One created with Suno's v3 went viral last month. Above is his Udio competing work. We used the prompt “Solo Acoustic Mississippi Delta Blues About Sad AI.” Similar to Suno, Udio calls his OpenAI's ChatGPT via API to write lyrics unless you provide your own lyrics. Suno and he also asked Udio to produce a diss track about each other. This is the result.
Suno currently generates two-minute long music clips from prompts. Udio is more customizable, but perhaps less intuitive to use, as it starts with 30-second segments that can be extended to your specifications. The company already offers more granular control to musicians privately, and co-founder Sanchez said the company has reached out to all creators with concerns. “We’re open for business,” he says. “Call us. We'd love to talk and we'll get you there. And you'll be like, 'Oh, wow.' This is actually cool. And now that I feel like I've mastered it, it's less scary. ”