Suno V6: The music machine is getting better - and that's exactly the problem

 

Suno V6 has been since 9. September outside and technically you don't have to talk about it for long: Things have gotten better. The music sounds more convincing, Instructions are implemented more precisely and generated songs can be edited more easily. There is also a variant with V6 mini, which is particularly fast and intended for large amounts of generation.

From a software company’s perspective, this is consistent. From the perspective of musicians, however, a warning light should go off when the term “high-volume generation” comes on.

Because Suno doesn't just sell a new instrument. A synthesizer doesn't write me a complete song at the push of a button, sings it, arranges it and then delivers the next variant. Suno automates exactly the work steps, with those musicians, producers, Composers, Studio musicians and singers have spent a significant amount of their time so far.

Of course, a musician can use Suno creatively. But that doesn't change the basic economic logic behind the product.

The faster Suno can produce music, the better this music sounds and the less human work it requires, the more attractive the system becomes for people, who don't really care about music as art.

And that's exactly where the slop factory begins.

Suno doesn't need a new prince

In my opinion, the biggest danger is not that, that Suno will eventually produce an artist more brilliant than humans.

Suno doesn't have to do that at all.

Music is enough for a profitable system, which is good enough.

A little lo-fi for learning. Ambient to fall asleep to. Workout-Beats. Piano-Music. Trap. Pop. Soul. Anything, which doesn't bother you for three minutes and is then replaced by the next track by the algorithm.

A human musician can perhaps seriously complete an album a year. An automated pipeline has no such limits. It can produce variants, Try out genres, Generate titles and covers and then look, what works.

At some point it no longer reminds you of a record label.

It's reminiscent of SEO spam with audio files.

When the cost of producing a single song almost disappears, the success rate must be tiny. Von 10.000 Tracks can 9.900 go under completely. If the remaining hundred generate enough streams, the business can still pay off.

That's exactly why V6's improvement isn't automatically good news for music.

Better slop remains slop. It's just harder to spot.

Where does musical knowledge actually come from??

Despite all the enthusiasm about technical developments, it should not be forgotten, that the AI ​​music industry has sparked a massive dispute over training data.

The crucial question is not just, whether a model can create a convincing soul song. The question is also:, how it learned, what soul actually is.

After all, generative music models don’t just fall out of the sky. You learn musical structures from existing material: Arrangement, Harmonic, instrumentation, Production, Life, Song structures and countless stylistic connections.

That's exactly why the arguments between Suno were, Rights holders and the music industry in principle.

The music industry has produced decades of human labor. This creates a system, that can partially automate human music production.

And then musicians may have to pay a subscription, to be allowed to use this machine.

You don't have to be a technophobe, to recognize this as a pretty serious conflict of interest.

Deals are being made now

The situation has now changed. Suno has agreements with Warner Music Group, BMG and Believe closed. It is available via licensed use, Opt-in systems and participation were discussed.

This is better than a permanent legal vacuum.

But it still doesn't answer the most important question for musicians:

Who gets how much?

“Revenue sharing” sounds fantastic on a press release. However, musicians now know such formulations quite well.

Streaming was also supposed to be the great democratization of music. Today independent artists are discussing it, how many hundreds of thousands of streams they need, so that a usable amount comes together.

That's why it's not enough, to announce, that artists are involved.

We need numbers.

How high is the share??

Who determines the value of a voice or a musical style?

What does an artist get?, when its material is released for a model?

Can he withdraw his consent??

What happens to models that have already been trained?

Who can check the billing??

And what proportion remains with the platform, Rights holder, Publishers and other intermediate stations depend?

As long as such questions remain unanswered, there is no reason, to celebrate the new relationship between AI companies and the music industry as a victory for musicians.

Suno deserves more music. Not necessarily musicians.

For me, this is where the crucial conflict of interest lies.

For Suno, it's basically good, when people generate a lot. More usage means a more attractive product, more subscribers, more data and a larger market.

For the music economy, however, more music does not automatically mean more money.

After all, the number of human ears does not grow with the number of tracks generated.

One day continues 24 Hours.

When a limited amount of attention is distributed across more and more music, Attention becomes even more valuable.

And it is precisely this attention that is already the real problem for independent artists today.

We have no shortage of music.

We have a shortage of people, that they hear.

Now we build machines, which can theoretically expand the already gigantic music catalog without end.

You can at least ask, who this is actually supposed to help.

The independent artist bears the risk

A large company can experiment with such changes.

A major has huge catalogs, Data, Marketing structures and bargaining power. When a new market develops with generative music, a corporation can try, to earn money from it.

A small artist does not have this position.

He pays for admission, Plugins, Distribution, Promotion and possibly musicians or producers. At the same time, it competes for the same playlists, Search results, Recommendations and ultimately the same ears.

When additional industrial quantities of generated music appear there, competition will not be fairer.

That's why I think the oft-heard statement “AI democratizes music” is at least incomplete.

It democratizes possibility, Generate audio files.

Whether they too Music industry democratized, is a completely different question.

The opposite could just as easily happen: Production becomes less valuable, while platforms, Models, Data and reach are becoming increasingly valuable.

Then power was not democratized.

Only production became cheaper.

Of course you can still use Suno

You still have to stay clean here.

When a rapper throws his own tune into Suno, generated crazy instrumentation, takes eight seconds out of it, everything chopped up, rearranged, raps about it and then creates his own song out of it, I don't think this is the same as an automated content farm.

Technology has always been a part of hip hop.

Samplers were technology. Drum machine were Technology. Auto-Tune is technology. A DAW is a huge piece of software.

But there is a difference between a tool, with which a person makes music, and a system, whose economic strength lies precisely in this, to make as much human work as possible unnecessary when making music.

This difference should not be argued away, just because the results are now pretty impressive.

And streaming could become a dump

Spotify, Apple Music, Deezer and other services will have to deal with this development.

Because millions of additional tracks do not automatically mean millions of additional cultural works.

When automated operators can easily generate huge catalogs, Fake artists come, Streamingbetrug, Recommendation manipulation and content farming inevitably come to the table.

Then labeling becomes interesting.

Upload limits are getting interesting.

Fraud detection is getting interesting.

And at some point possibly also the uncomfortable question, whether every automatically generated track should have the same right to compensation as music, which actually has an artist behind it.

However, this must be prevented, that small musicians of all people are suffering from new rules, while large rights holders get special agreements.

That would be the worst possible outcome: The industry solves the AI ​​problem and the independent pays the bill.

Maybe the answer lies outside the streaming race

That's exactly why Bandcamp, own shops, Websites, Newsletter, Concerts, physical releases and direct fan relationships more interesting.

Not because we're suddenly back 1996 should play.

But because a direct relationship between musician and listener is harder to flood with slop.

When someone buys your album, he has made a decision.

When someone comes to the concert, also.

When someone orders a tape, reads your newsletter or follows a small label for five years, there is a connection, which a recommendation algorithm cannot simply replace with the next generated track.

Perhaps this will be one of the most ironic consequences of the AI ​​revolution.

The more artificial and unlimited digital music production becomes, the more valuable it could become again, that there are actually people behind a record.

For me, Suno V6 is neither some harmless new plugin nor the downfall of music.

It is a powerful commercial system, whose interests are not automatically the same as the interests of musicians.

We shouldn't forget that, just because the technology is fascinating.

Suno wants, that as much as possible is generated.

Want streaming services, that as much as possible is heard.

Rights holders want, that as much as possible is monetized.

And musicians should pay damn attention, that in the end, between these three interests, someone will get paid for it, To actually make music.