The newest wave of deals between major labels and AI companies points in the same direction. The future they describe isn’t a machine writing hits. It’s fans customizing, covering, and remixing licensed catalog inside sanctioned tools, with every interaction licensed and accounted for.
Spotify’s forthcoming AI product will let fans make licensed covers and remixes from participating artists and songwriters as a paid add-on for Premium subscribers. Universal Music Group and ElevenLabs are building a separate platform for remixes, mashups, new interpretations, and personalized vocal experiences.
The industry calls these superfan apps, remix and cover apps, or active listening. The question hanging over all of it is simple: how many people actually want to do this?
The West isn’t starting from zero. Hook already operates a licensed social-music platform built around fan remixes and mashups, with more than 1 million licensed songs from 18,000 artists. The company says users have created more than 100,000 Hooks, generating more than 300 million views. In August, Universal Music Group announced a broader licensing partnership with Hook after two years of collaboration across more than 30 artist campaigns.
Hook isn’t exactly the same product the industry is now building around generative AI. The company explicitly distinguishes itself from platforms that generate entirely new tracks, though AI is part of its technology stack and fuller generative remix tools are on its roadmap. But it is already testing one important part of the thesis: whether Western fans want to manipulate licensed recordings instead of simply listening to them.
China offers a different kind of evidence, because its mainstream music apps started putting AI-powered creation tools and licensed voice experiences directly in front of enormous audiences years ago.
What China’s AI Remix Experiment Has Proven
It started with the voice. By late 2022, Tencent Music Entertainment had already released more than 1,000 songs with human-mimicking AI vocals, one of them with over 100 million streams, according to the company. The consumer product followed.
Kugou, QQ Music, and Tencent’s karaoke app Quanmin K Ge — known internationally as WeSing — all added AI cover features.
On Kugou’s AIK, a user could record about 60 seconds of their own voice, allowing the model to recreate their timbre. They could then choose from songs available in AIK’s library and have the model “sing” them in their voice, including notes they might not be able to hit themselves. Contemporary reporting put the price of a full AI cover at 10 yuan and a clip at 5 yuan.
Then came the licensed celebrity voice. In February 2024, AIK reached an authorization deal with Karen Mok, one of Mandopop’s defining voices, letting users generate duets with an authorized model of her voice. More than 10,000 duets were made within 24 hours, and more than 30,000 people took part in the first week. Kugou followed with G.E.M., Yisa Yu, and Yoga Lin, and QQ Music and Quanmin K Ge shipped their own star-duet features.
In July 2024, Westlife used AIK to release a Mandarin cover of Jason Zhang’s “Love + Courage” exclusively through Tencent Music, with the band saying they personally oversaw the process. The recording passed a million streams in its first week. This is one version of the product Western labels are now pushing further: a consented, licensed, artist-sanctioned AI voice offered directly to fans.
The listening numbers were bigger still. Datouzhen Official, an AI singer created by Kugou using Tencent Music’s Vemus technology, had released more than 4,000 tracks by August 2026. Chinese tech press reported more than 56 million listeners on Kugou and 32 appearances on the platform’s TOP 500 chart.
On May 15, an AI R&B rework of “999 Roses,” a 1993 ballad by Tai Zheng-xiao, topped Kugou’s TOP 500 hot-songs chart, while videos under the song’s related topic on Douyin — China’s version of TikTok — had accumulated more than 1.2 billion views.
There is some evidence the creation tools make money, too. On its Q2 2026 earnings call in August, Tencent Music specifically called out AI creation features — including Kugou AIK and QQ Music’s AI songwriting tools — for producing “very good commercial returns.” But Tencent has never broken out how much revenue those tools generate.
So will anybody use it? In China, the answer is yes. In its Q1 investor presentation, Tencent said its AI tools serve more than 10 million everyday users and 150,000 professional creators, though that covers its whole suite of AI music tools — not licensed remixes or celebrity duets specifically. Some of those users pay.
What we don’t know is whether that makes active listening a major business or simply a profitable feature inside a much larger one.
What China’s AI Remix Experiment Has Not Proven
The making side of these licensed fan experiences appears much narrower. Thirty thousand people in a week is real engagement for a single feature, but it’s tiny next to the audiences these apps serve. The biggest numbers aren’t from fans individually remixing, but rather from people listening to someone else’s AI transformation.
The licensed products also face unlicensed competitors. Unauthorized celebrity voice clones have circulated openly on Chinese platforms for years. By 2023, AI re-creations of Stefanie Sun, Faye Wong, JJ Lin, and Jay Chou were already attracting millions of views. The licensed and unlicensed versions can satisfy much of the same appetite, but an unauthorized version isn’t limited to artists and catalog that have opted in and doesn’t require a star’s approval.
The unlicensed market isn’t just a copyright problem. Tencent Music says it’s becoming a competitive one. On the company’s Q1 2026 earnings call, CEO Ross Liang said rival streaming services were using AI to quickly fill their music libraries and draw users onto their platforms. Management blamed a slight quarter-over-quarter dip in membership revenue mainly on that competition, particularly among free and ad-supported listeners.
Price friction arrived immediately. Users complained about paying 10 yuan (about $1.50 USD) to generate a track and another 29.9 yuan to download it (around $4.50). Kugou also blocked screen recording of AI covers, citing copyright restrictions. That meant a user who wanted to make and keep a song could end up paying roughly 40 yuan (around $6) each time.
Taken together, China has shown that people will use these tools, pay for some of them, and listen to the results at enormous scale. What it hasn’t shown is that creating those results will become a mass behavior itself.
The audience is massive. The number actually doing the remixing is much smaller.
The Lesson for Active Listening in the West
China isn’t the only place to hint at this divide between making and listening. The aforementioned Hook provides one Western example; Japan offers another, older one.
Japan has spent two decades building music around licensed voicebanks (we talked about this back in May). Fans can buy recordings of real singers’ voices and use them to create entirely new songs. IA, built from the voice of singer Lia, had several songs pass a million views on niconico within six months of launching in 2012, and voicebanks like it are now being rebuilt as AI models.
But Vocaloid culture developed around a familiar imbalance: a dedicated community made the music, while a much larger audience simply listened to what that community produced.
South Korea hasn’t gotten as far. HYBE subsidiary Supertone has talked about a fan remix product since 2024, but it still hasn’t launched. Meanwhile, unauthorized AI covers of K-pop idols continue to circulate widely online.
That leaves China as the clearest large-market test so far of putting licensed AI voices and creation tools directly into mainstream music apps. And what it shows is less “everyone becomes a remixer” than a much smaller group creating material that a far larger audience is willing to hear.
As the rest of the world bets on active listening, China shows what shape the bet actually has: enormous audiences for AI reworkings of familiar music, millions of people using AI creation tools, and much thinner evidence that masses of listeners will repeatedly pay to manipulate licensed catalog or authorized celebrity voices.
That matters because the Western market is about to get crowded fast.
Spotify, Universal Music Group, Warner Music, Udio, KLAY, Suno, ElevenLabs, Hook, GRAI, and others are now building variations on the same basic idea: licensed music that fans can do something with instead of merely stream.
The products differ. Spotify plans to let Premium users make licensed covers and remixes. Universal and ElevenLabs are building their own platform. Udio has licensed fan-creation experiences planned, while KLAY has agreements with all three major labels around what it explicitly calls active listening. Suno is developing opt-in experiences with participating artists and songwriters.
Smaller platforms are already testing the concept too. GRAI, which recently acquired Hangout, is building a social music service around licensed covers, remixes, and other interactive experiences. Starchild, the only platform in this list that’s already live, lets independent artists authorize fans to remix their songs and purchase licenses for the resulting versions.
They are different products aimed at different audiences, but they are all making some version of the same bet: that music fans want creation layered onto listening.
China suggests that bet may be only partly right.
There is strong evidence that people will listen to AI reworkings of familiar music. There is also evidence that a meaningful minority will use AI creation tools themselves. What China has not yet demonstrated is that licensed fan creation can become a mass behavior on anything like the scale of ordinary listening.
If that pattern carries over, these companies may not be competing for hundreds of millions of would-be remixers. They may be competing for a much smaller group of highly engaged creators whose work is consumed by everybody else.
And beyond that, they’ll still be competing with unauthorized tools, so a sanctioned product has to offer something better than legality alone.
Price is part of that equation. If every experiment requires another payment, while an unauthorized alternative is free, the friction starts working against the licensed product. The stronger model may be to bundle creation into something fans already pay for rather than put a tollbooth in front of every remix. Spotify hasn’t announced a price for its add-on, and other services are still developing their models.
The West spent two years arguing over whether AI could use the catalog. China spent that time finding out what people do once it can.
So far, mostly, they press play.
🆕 New AI Music Releases
Got a new release? Submit it here.
Albums/EPs
All Is Wild - Can You Hear The A.I. In My Music? (Pop) and Pastoral (Folk rock)
Common Damage - No Clinical Significance (Rock / Alternative)
Leea - Reticências... (Jazz)
Shindig Productions - Chrome Hearts, Analog Souls (Electronic / Dance)
Silke Winther - HUMAN (Electronic / Dance)
We’ll Always Have Pari Pari - Classic Lady (Hip-Hop / Rap)
Zonne Artanne - Sol𝑬laN - Songs of Nature (Acoustic Folk)
Singles
All Is Wild - “What Grace Can Bring” (Gospel folk rock)
Born Ethereal - “The Crown Falls” (Symphonic Death Metal)
Common Damage - “As Prescribed”
Honest Bob - “Count Me In (explicit)” and clean (Hip-Hop / Rap)
Ione - “Tears of Brass & Bone” (Blues)
Jyndal - “Wise Words” (Post-punk / Folk)
Lowan Wolf - “Lowan’s Gospel” (Folk Gospel)
Noah & Felix - “IN BETWEEN” (Rock / Alternative)
Rick Dallas - “Drive” (Electronic / Dance)
Silas T. Morrow - “The Clock-Eyed Prince” (Rock / Alternative)
SimbaXp - “Flicker” (Chillstep / Future Bass)
Single Station - “Grace Found Me” (Gospel rock)
SYRA_404 - “Offline Again - BOOTLEG_404: Room Unknown” (Electronic / Dance)
Music Videos
Leea - Reticências (Jazz)
🎧 Song of the Day: “Half A Wife” by Detour Soundscape
Detour Soundscape's "Half A Wife" earned Viewer’s Choice honors at last week's Fresh Friday Festival for its deeply moving, cinematic storytelling that captures the feeling of living with chronic pain. In the video, a box represents feeling trapped in her own life, unable to fully participate, until her husband tenderly climbs inside to stay with her rather than trying to fix it.
Read Professor Jacket's full Week 38 recap on Zinstrel.com to dive into the winning video — the very first installment in his new weekly series chronicling the best in AI music videos.
💬 Last Word
“Fans and creators aren’t just promoting songs anymore. They’re turning music into memes, skits, stories, edits, dances, lifestyle content and entire pieces of culture.”
— Chukwubueze Okolie (Head of Operations, Kings Music Group), via LinkedIn
MORE ZINSTREL:
Zinstrel.com | @zinstrelculture on IG | AI Underground on Discord
Zinstrel #118: The Udio Remnant
Zinstrel #117: UMG Is Targeting Both Ends of the AI Music Pipeline
Created by Marcus Lawrence
Copyright 2025-2026 Zinstrel, LLC - All rights reserved
Special thanks to our paid subscribers: Blue Hermit, Bukar Gadzama, C.Y. Lee,
Charles S, Chez, Daniel Drolet, Daryl Dekking, Dave, Dexter Garcia, Drew Thurlow,
Ghost Ryder, Itai Leibowitz, Jacci Cenci-McGrody, Lenny Skolnik, Lynn Clapp,
Marc Wolinsky, Matthew Marturano, Matti Kuha, Mete Dibi, Rosanna, Sheila Spence,
Steve Clapp, Steve Grossman, Tim Burley, and Tracy Maddux.
No need to look at China, with its very different fan behavior from the West. Licensed remixes powered by AI have been available in the US since "years ago" on Hook. Including a full catalog UMG license. They just have not cranked up their PR machine and they don't call themselves an AI company. Check it out!