Universal Music Group and Sony Music plan to use what they call “groundbreaking neural fingerprinting technologies” to detect copyright infringement in AI-generated music.
The two major music companies have each partnered with a research lab called SoundPatrol, which has developed a patent-pending method to analyze music.
SoundPatrol, which originated at Stanford University, is developing what it calls a “forensic AI model for audio-video fingerprinting,” which it claims “represents a step change from existing detection methods.”
The partnership arrives as rightsholders face mounting challenges from AI music generators that allegedly use copyrighted material without permission to train their models. The RIAA has cases going against AI startups Udio and Suno. The majors also filed an amended complaint against the latter company just a few days ago, accusing it of illegal “stream-ripping”.
UMG and Sony Music-backed SoundPatrol was co-founded by Michael Ovitz, the prominent entertainment exec who co-founded Creative Artists Agency, and Walter De Brouwer, a noted linguist and entrepreneur.
Producer Oak Felder and Milk & Honey President, Lucas Keller, were two of the co-founding partners in Soundpatrol, and continue to hold stakes in the platform.
Current music fingerprinting technology looks for exact matches of snippets of music, but SoundPatrol’s patent-pending tech can also detect covers, remixes, and derivatives created by AI, the lab said in an announcement on Thursday (September 25).
According to a press release, SoundPatrol’s tech “employs neuralembeddings that capture and analyze musical semantics in order to identify the influence of original human-created music in fully or partly AI-generated music content”.
A “neural embedding” is a way of turning elements of music (or words or other data) into numbers. Musical elements like chords or melodies are given a numerical representation that makes it easier to recognize similarities between them.
UMG Chairman and CEO Sir Lucian Grainge said: “We’re constantly focused on enabling AI – bringing to market the many commercial and creative opportunities that will benefit our artists while establishing effective tools to protect them.
“Bringing solutions to the table that support the entire industry is at the heart of our relationship with SoundPatrol, who share our commitment to safeguarding our artists’ creative integrity and work.”
Photo: Austin Hargrave
“Bringing solutions to the table that support the entire industry is at the heart of our relationship with SoundPatrol, who share our commitment to safeguarding our artists’ creative integrity and work.”
Sir Lucian Grainge, Universal Music Group
Dennis Kooker, President of Global Digital Business at Sony Music, said AI presents “opportunities for artists and creators” when used correctly.
“We’re committed to navigating this developing landscape by protecting [artists’] work while also exploring the innovative potential of these technologies. Our collaboration with SoundPatrol is about respecting artists’ rights to build a sustainable and equitable ecosystem for everyone,” he said.
The technology could prove to be a boon to music companies that have pursued legal action against AI companies they believe have violated the law by using copyrighted music without permission to train their AI models.
The current spate of lawsuits against AI companies largely rely on comparisons of sheet music or lyrics to make the case that an AI-generated piece of music ripped off a human-made original.
Data from SoundPatrol could back up those claims with algorithmic data. It could also help rightsholders detect AI rip-offs where they hadn’t noticed them before.
“SoundPatrol has answered the long-standing problem of IP theft by creating a frontier lab with neural fingerprinting capabilities that can identify all pipelines of directly transmitted content, whether on its own or intermixed, in real time,” said Ovitz, who serves as SoundPatrol’s board chair.
“This is the first of-its-kind technology implemented to protect all copyright holders and creators of any type of intellectual property.”
He called it a “huge victory for all artists in the creative universe”.
“If we abandon copyright, we risk severing artists from ownership of their own work,” said De Brouwer, SoundPatrol’s CEO.
“It is compulsory to proactively feed deep embeddings of these neural signatures into streaming infrastructures so that owners can maintain control, authenticity, and monetization of their intellectual property in the generative AI era.”
SoundPatrol’s key advisors include Percy Liang, founder of the MARIN foundation models lab, Chris Re of Stanford’s AI lab, and Dan Boneh, director of Stanford’s Applied Cryptography Lab.
“Our collaboration with SoundPatrol is about respecting artists’ rights to build a sustainable and equitable ecosystem for everyone.”
Dennis Kooker, Sony Music
Headed by John Thickstun, Associate Professor of Computer Science at Cornell University, SoundPatrol’s research team includes computer engineers, AI experts and musicologists.
The team includes graduate students from Stanford University’s CCRMA and London’s C4DM, the UK’s leading digital music research group at Queen Mary University. It also includes AI engineers from Carnegie Mellon University, Brown University and the University of California-Berkeley
Aber Whitcomb, the former Chief Technology Officer at MySpace and Jam City, and Frederick Kautz, former Chair of KubeCon and CloudNativeCon, are coordinating teams to strengthen security and innovation, SoundPatrol said.Music Business Worldwide
