Roy started Wubble with his co-founder, Shaad Sufi, in 2024, from a small office in Singapore’s central business district. Since then, his platform has generated tunes for global giants like Microsoft, HP, L’Oreal and NBCUniversal. They’re even used on the Taipei Metro, where AI-generated tunes soothe harried commuters. 

Generative AI has been a controversial subject in the creative industry: Artists, musicians and other content creators worry that companies will train AI on copyrighted materials, then ultimately automate away the need for human creators at all.

Roy, however, thinks Wubble is a way to fix a music sector that’s already broken. Artists are awarded micro-payments on streaming sites like Spotify, which only works for the most famous artists. 

Roy spent almost two decades at Disney, where he oversaw operations at its networks and studios in major cities like Tokyo, Mumbai and Los Angeles. He said his time leading Disney’s music group opened his eyes to the tedious process of music licensing.

“So many licensing deals were not going through because of the quantum of paperwork, the amount of red tape, and how expensive, complex and convoluted the entire process was,” he says. Yet, the incumbent music firms “don’t have a lot of motivation to streamline processes.”

Wubble is trying something different, collaborating directly with musicians and paying them for the raw material used to train Wubble’s AI. “If we’re looking at Latino hip hop, we’ll go to a recording studio in Buenos Aires or Rio de Janeiro, and tell them we need ten hours of Latino music,” Roy says. Wubble then negotiates a deal and offers a one-time payment for their work, at rates Roy argues are more competitive than other companies offering music streaming services.

He admits that a one-time payment isn’t a perfect solution, however, and adds that he’s currently exploring how technologies like blockchain can uncover new ways to compensate musicians for their help training Wubble’s AI models.

David Gunkel, who teaches communication studies at Northern Illinois University in Chicago, thinks training AI from artist-commissioned material is a smarter business move than just trawling the web for copyrighted content.

Production companies like Disney, Universal and Warner Bros., for example, are suing AI companies like Midjourney and Minimax of copyright infringement, arguing that users can easily generate images and videos of protected characters like Star Wars’s Darth Vader. 

“If you’re curating your data sets, and compensating and giving credit to the artists that are being utilized to train your model, you won’t find yourself in a lawsuit,” he explains. “It’s a better business practice, just in terms of your long-term viability as a commercial actor.”

Text-to-speech generation

Wubble currently offers just instrumental music and audio effects, but Roy thinks voice is the next step. By end-January, Roy says his platform will offer AI-generated voiceovers created from written scripts, to cater to clients who require narrative-led audio tracks. “So, the entire audio content workflow for a business can be housed on Wubble,” he concludes proudly. 

AI music startups are popping up around the world, hoping to use the powerful new technology to make the process of creating tunes and songs easier. Some, like Suno, cater in generating full songs, while others like Moises offer tools for artists.

In Asia, too, Korean AI startup Supertone offers voice synthesis and cloning, using samples to generate new vocal tracks. The startup, founded by Kyogu Lee, was acquired by HYBE, the entertainment company behind K-pop sensation BTS, and now operates as its subsidiary. Supertone even debuted a fully virtual K-pop girl group, SYNDI8, in 2024. 

At Fortune Brainstorm AI Singapore last year, Lee said he saw musical artists as “co-creators,” not just in terms of licensing their voices, but also asking for their help in refining the technology. 

AI “will democratize the creative process, so every creator or artist can experiment with this new technology to explore and experiment with new ideas,” he told the audience.

Roy, from Wubble, also sees AI as a way to make it easier for more people to get involved in music creation.

“Music creation has always been a privilege. It’s been the domain of those who have the time and resources to learn an instrument,” he says. “We believe that every human being should be able to create—and AI enables that now.”

Is Spotify spreading its royalty pool across too many artists?

One music industry strategist thinks so, arguing in a provocative new essay that the streaming giant should implement a 250,000 monthly listener threshold to concentrate payments among professional musicians who can earn a sustainable living.

The proposal comes as Spotify moves in the opposite direction for podcasters, slashing its Partner Program eligibility requirements by half this week to make it easier for video creators to start earning money.

On Wednesday (January 7), Spotify announced that video podcasters can now qualify for monetization with just 1,000 engaged audience members over 30 days, down from 2,000.

The company also reduced the required watch time from 10,000 hours to 2,000 hours, and slashed the minimum episode count from 12 to just three.

For music artists, however, Spotify maintains a different standard: tracks must reach at least 1,000 streams in the previous 12 months to generate royalties on the platform. The threshold, introduced in April 2024, also includes an undisclosed minimum number of unique listeners to prevent gaming the system.

Writing on Substack, music industry strategist Joel Gouveia argues that artists below his proposed 250,000 monthly listener threshold are already receiving payouts “so small they change nothing.”

By redistributing Spotify’s $10 billion annual royalty pool among far fewer qualifying artists, Gouveia calculates that per-stream rates could increase from around $0.004 per stream to somewhere between $0.0065 and $0.008.

“When everyone gets paid, no one gets paid enough,” Gouveia writes, arguing that the current system is “quietly killing the middle class of music” through extreme dilution.

Under his proposed model, Gouveia suggests an artist generating one million monthly streams could see their monthly income jump from $4,000 to between $6,500 and $8,000 – transforming streaming from supplemental income into a sustainable livelihood.

Gouveia acknowledges the harsh reality: “Any artist under 250,000 monthly listeners is currently getting $20 here, $50 there, $100 if they’re lucky.”

He argues that these micropayments don’t fund growth, change strategy, or enable sustainability, making them functionally meaningless while draining resources from the struggling middle class of artists who have real audiences but can’t earn enough to sustain careers.

In 2023, Deezer and Universal Music Group launched their ‘artist-centric’ royalty model, which applies a double-weighting multiplier to streams from artists with more than 500 monthly listeners and over 1,000 monthly plays. Artists below those thresholds see their per-stream royalty worth half as much as more popular acts.

Believe, parent company of TuneCore, criticized that model at the time, arguing it represents an unfair system “centered around taking compensation from rising artists to allocate it to top and established artists.”

Gouveia’s 250,000 monthly listener proposal (with a threshold 500 times higher than Deezer’s) would likely intensify such concerns.

Gouveia draws parallels to other creator platforms, noting that YouTube requires 1,000 subscribers plus significant watch time before monetization, while TikTok demands 10,000 followers and 100,000 views in 30 days. “Most platforms already run a threshold system,” he writes. “You don’t earn until you’ve proven demand.”

As pointed out by MBW in September 2023, such barriers exist at other tech giants, too.

The Meta platform, for example, won’t monetize in-stream ads on your videos until you have 10,000 page followers plus 600,000 total minutes of video watched in the last 60 days.

Spotify introduced its 1,000-stream music threshold with a specific economic rationale: the company said it was targeting tracks that generate less than five cents per month on average. These were micropayments, it argued, that were “being destroyed by being turned into fractional payments” that often never reached artists due to distributor withdrawal minimums.

The threshold was designed to reallocate that money, which Spotify said represented 0.5% of its royalty pool, or approximately $40 million annually, to tracks exceeding 1,000 streams. The platform also implemented an undisclosed minimum number of unique listeners to prevent artificial streaming fraud.

Meanwhile, Spotify is now lowering barriers for podcast creators as it expands its video podcast offerings, creating a marked difference in approach between its two content types.

The divergence highlights competing visions for streaming economics: Should platforms maintain accessible barriers that encourage any artist to participate? Or should they implement dramatically higher thresholds to concentrate revenue among established acts, even if that means excluding millions of tracks from monetisation?

Gouveia believes the answer is clear: “The industry doesn’t need more participation. It needs more sustainability.”

Music Business Worldwide