✘ Streaming services are not Calm & FluffyAnd: Songs are going viral faster than ever; Should we still be using the term 'new media'; Electronic music royalties under the microscope; Has the AI bubble burst yet
AI developments are moving fast, yet it feels incredibly slow. The recent update from Deezer that 50k tracks fully generated by AI are uploaded every day is shocking. It’s no surprise that more tracks get uploaded and get through whatever streaming services have in place to prevent this slop. Or, do they actually want to prevent this? What if the DSPs are just trying to find ways to pay less royalties. It certainly fits in the podcast, audiobook, all audio strategy. Take the the recent developments at Tidal, where you can now do direct uploads. It’s this part that stands out:
It’s easier than ever to create something that sounds like music and streaming services are becoming subscription-based personal music players. In light of this Bas, known around here as the founder of MUSIC x, wrote in his newsletter Calm & Fluffy about the systemic issues underlying the current issues we see with streaming services and AI: the aforementioned 50k AI tracks on Deezer; the Dutch Spotify viral chart being taken over by far-right AI ‘songs;’ Billboard’s chart having an AI-generated track hit number one. It’s an excellent read, and exactly the kind of engaged thinking we need around the topic. I’m publishing it here, too, because I think you should all read it. Suddenly, Spotify’s viral chart for the Netherlands saw itself populated by AI-generated far-right music. Songs with titles like “Fuck you! Leftist fuckers!”, “Just call me far-right”, and “If you don’t like it then leave!”. This happened against the backdrop of the recent Dutch parliamentary elections following the fall of a populist right-wing government. In the run-up to these elections, anti-asylum seeker demonstrations erupted in violence multiple times, with one in The Hague leading to torched police cars and the smashing of the party office windows of the Netherlands’ liberal democratic party, D66.
Those far-right AI songs? Not against the terms of service, according to a statement by the music streaming market leader, reports Dutch newsletter AD (translation mine):
This definitely made me scratch my head for a moment, at the risk of entrapping my fingers in my luscious though oft-tangled locks. Last week, another music streaming service, Deezer, reported receiving over 50,000 fully AI-generated tracks per day. At this rate, that would mean nearly 20 million per year, or 38 per minute. With anyone able to generate tracks with a few clicks, streaming services will have to reckon with moderation, lest they become comment sections. Popular music = economicsMusic has undergone massive changes over the past 150 years. Popular music, as in music that people in your community are all likely to know, was folk music. Songs that travelled. Songs that a person in your community was best at performing. Some songs would be new and specific to your community, and others might be centuries old. Then the recording came. And mass consumption. Instead of a participative music culture, most music was primarily consumed: through records, then downloads, then streams. Music as a format shifted, too: from albums, to single-track downloads and iTunes playlists, to infinite personalised recommendation-based radio stations. Music creation is also a tale of consumerist culture and economics. Big bands got dethroned by pop and rock bands, who in turn lost their place at the center of popular culture to hiphop (or pop influenced by the genre): economically much more efficient with drum machines & microphones. And then electronic music blew up globally, which can be made with just a laptop and performed by DJs travelling with not much more than a USB drive. This is not intended to read like a tale of cynical greed, but rather one of systems and incentives. The system of consumerism rewards economic efficiency until every last cent of profitable return has been squeezed out. In the process of the economic squeeze, a few genies have been let out of the bottle, undermining the system built by the recording industry itself. Genie |
 |
The other genie that had been let out of the bottle was that almost everyone could now make music. Got a PC? Got an internet connection? Now you can have a small studio on your laptop, whether by purchasing music production software or donning a virtual eye patch and sailing the high seas (which is how many producers of that generation got started).
No longer did you have to get an A&R at a record label to ok your project for it to have a chance to be recorded and heard by people. Just do it yourself.
The 2008 documentary RiP: A Remix Manifesto, featuring mash-up artist Girl Talk, perhaps captures this cultural moment of the late 2000s best.
Slowly, active participation in pop culture returned to the mainstream through remix and mashup culture, which many corners of the record industry vigilantly tried to fight until it found ways to monetise it through content identification systems that redirect digital revenue streams to rightsholders.
Online, popular music started behaving a bit like folk music again. Participative. Communal. And context-specific.
Over time, these remixes could behave like memes. Music catalogues made their way into Instagram Stories and TikTok. If you’re too young to remember the internet before that era, it’s hard to make clear how restrictive things were.
Music creators, who could now really be anyone, started gaming the existing structures. Some through nefarious means like botting streams, or by now-common industry practices like releasing loads of short tracks, so you rack up higher streaming counts (and royalties), or releasing so regularly that you never fall off of people’s Release Radar playlists.
The result:
Many tens of thousands of tracks are added to streaming services daily.
And this was before AI.
We should not be surprised that music has lost many of its 20th-century characteristics.
Entire industries have focused their energy on ensuring that music is ever easier and ever cheaper to make and distribute, whilst hyperpersonalising it to ensure people have something to listen to all the time. “Music for every moment”, as Spotify’s marketing campaign slogans rang.
Generative and adaptive music was always going to happen.
Music as language
In a way, we’ve come full circle. Music has historically been one of humanity’s most important venues for storytelling. In recent centuries, music has gone through an extraordinary process of professionalisation. In the 20th century, it became incredibly difficult for the average individual to tell stories through music that sounded on par with music supported by record labels.
The economics pursued by the record industry ultimately undermined their cultural monopoly by weakening their gatekeeper position. If anyone can create & distribute music cheaply, creators don’t have as much need for record companies’ financial resources. So the record industry went from taste-making to taste-chasing, supporting artists at slightly later stages of their careers, often after they could show traction with fans or the ability to go viral.
Radiohead’s Thom Yorke once called Spotify “the last desperate fart of a dying corpse.” His critique came in the context of a hopeful era of the web. He thought music didn’t need another intermediary, and certainly not one as closely aligned with the major labels’ interests. Ultimately, the battle for the web was won by middlemen who sit between artists and fans: Meta, TikTok, Google / YouTube, Apple Music, the list goes on. For what it’s worth, I think it’s good that at least one of these intermediaries has its roots in music. But I digress.
Thom Yorke called it. He thought the recording industry was losing relevance compared to its role in the previous century.
Now we’re seeing that play out.
If everyone can express their ideas through music with just a few clicks, what role do the recording industry’s streaming services have?
The choices streaming services face
The Netherlands’ viral charts on Spotify turned into your typical 2025-era internet comment section. This was only changed after the distributor of most of the far-right AI tunes, DistroKid, became aware and decided to pull the plug on them.1
What’s worse is that 97% of listeners struggle to distinguish between AI-generated and non-AI music. That means the emotional impact of generated songs will, at least upon discovery, be similar to ‘human music’.
I believe that music in the 20th century was an anomaly. It was a product of an era of mass media and a time when few people had the means to produce and distribute music. As we moved into networked media, music could become small again: relevant only to specific contexts. This can be a national political context, but also a particular WhatsApp group. Memes already behave this way, because they’ve long been easy to create with image editing. Now it’s music’s turn.
And this is not new. I used to be on hiphop forums where people would record tracks to battle each other, for an online audience of maybe a few hundred people. This was in the early 2000s, when computers didn’t come with built-in mics or sound cards that made recording easy.
The future already happened, and now it is taking hold of popular culture.
At the start of 2024, I published a piece titled Will AI correct the anomaly of the 20th century music business? in which I concluded:
“We might comfort ourselves with the thought that we’re not there yet, but if the last 2 years have shown us anything it’s that things can move fast. Lightning fast.
In the introduction I mentioned the question “what happens when anyone can just make a song with a text prompt?” but I believe this question to be irrelevant, because before we can formulate a meaningful answer, the above will have already come to fruition.
Even with the above just around the corner, we should act like we are already there. What we believe to be true in the future, is true now. If we hold it to be evident, inevitable, the best thing we can do is to behave like it is already the case, at least part of the time.”
Streaming services will have to pick a strategy.
Are they going to be comment sections? Will they be vestiges of the type of recorded music that emerged in the 20th century? Will generative music services like Udio (now licensed by the world’s largest major label) and Suno (still under legal pressure) become the platforms for this type of music instead?
There are many, many roads.
LINKS
☣️ Songs Are Going Viral Faster Than Ever — But Fewer Are Becoming Streaming Hits (Jubran Haddad)
“A few years ago, virality was chaotic and unpredictable. In 2020, a bedroom dancer could turn a song into a global hit overnight. Today, it’s a formula that labels and artists have learned to engineer. Hooks are designed to loop, lyrics are trimmed for catchiness, and release campaigns are built around “pre-save” teasers and influencer seeding. But no algorithm can force genuine interest. If users don’t choose to make it part of their content, the song simply won’t take off. Fans know when the hype is genuine or manufactured.”
✘ It’s this part that stands out for me. It’s mostly about the shifting nature of virality on a platform like TikTok. In 2020, most artists going viral were unsigned and unknown, now it’s Justin Bieber and Sabrina Carpenter. The game of discovery is already moving elsewhere.
🆕 is new media a term we should still be using? (Sean Monahan)
“‘Gatekept’ media was a story about production costs. The owners and the editors programmed culture. Complaints about bias on pre-Elon Twitter and post-Elon X follow the same script. There is one clearing house that sets the cultural agenda. One ring, one platform to rule them all. As with the twentieth century printing presses, technically the system is open to everyone and the best media output rises to the top. But it’s dishonest to insist owner and algorithms have no effect on which topics trend.”
✘ Yeah, media goes through cycles. It’s good to be reminded of that every now and then.
🔬 Electronic music royalties: Under the microscope (Rufy Ghazi et al)
“Electronic music presents unique challenges for traditional royalty systems. Whilst dual registration with both PRS (for composition rights) and PPL (for recording rights) is required across all genres where artists create and perform their own music, electronic music producers face additional barriers. Unlike singer-songwriters who often identify naturally with these dual roles, electronic music creators historically perceive themselves as ‘producers’ rather than ‘songwriters’—a cultural disconnect that has created decades of non-registration.”
✘ A longstanding issue in DJ and club culture that needs this type of research and understanding. Fantastic practical recommendations, too.
🫧 Has the AI bubble burst yet?
✘ nuff said
MUSIC
Sub-bass tremors make your eardrums tingle in delight. The Bug vs Ghost Dubs delivers a truly awesome set of tracks exploring the deepest sides of bass. Listen to Implosion now.
Implosion, by The Bug vs Ghost Dubs The Bug vs Ghost Dubs |
MUSIC x is free today. But if you enjoyed this post, you can tell MUSIC x that their writing is valuable by pledging a future subscription. You won't be charged unless they enable payments.
| ||||||||||
