No, I say that harmony means "how humans hear music" so you can't abstract harmony from the human experience.
You say that you like to hear AI music using harmony but completely stripped away any aspect of the human experience, which is impossible because of what I've just said. It has nothing to do with AI using harmony per se, but you want it to use harmony without any human aspect of harmony, which is an oxymoron. I'm not gonna repeat myself a fourth time

 

this would be fun. So no training, just a deep understanding of how music works??

 

The big studios have always catered to artists with record label support. What has changed is the easy access to digital tools such as DAW & VST, and industry outsiders have spent a lot of time learning how to master them. This has skewed the market, and the old studios can't compete against these new low cost competitors who operate these barebones low-cost studios.

 

Last edited: Jul 25, 2024 at 12:14 AM

AI is not a tool for making music. Artificial intelligence is a clever way to steal from artists and their works of art. Unfortunately, the judges who deal with copyright matters do not have sufficient knowledge of music and are not able to recognize these works. If the judges and courts know enough about the art, they should try the creators of artificial intelligence. Because they don't really do anything special and are thieves in the form of developers. Even those who use artificial intelligence are criminals because they knowingly or unknowingly fall into the hands of technology fraudsters and make their fake economy flourish.

 

What's different now is that no humans are actually making the music.

 

It still amazes me how everyone still refers to all these current mainstream devices as "AI", that simply mimic true artificial intelligence with a finite collection of simple algorithms. True artificial intelligence is not limited in scope to a finite set of rules, and is capable of learning over time to the point where it actually improves upon human intelligence. The fact that none of these mainstream "AI" based devices are achieving anywhere close to autonomous task performance capability, all the while still being termed as "AI" mainly for sales and marketing purposes, is really quite comical.

The best result one can hope for with these commercially hyped "AI" devices is a random collection of events dictated by a set of pre-programmed algorithms operating sequentially (not concurrently) to fulfill a set of human requested parameters. These results being not dissimilar to a random number generator, or word guessing app, the likes of which have been around for decades and were not termed "AI". Hence, we now have the mad commercial dash to "AI" nirvana wherein the slapping of an "AI" label on anything raises the potential profit margin of a product by anywhere from 20-40%.

If you're able to flow-chart code that you're calling "AI", then it ain't really "AI". Basically you'll have a decision diamond that everything else on the chart eventually points to called simply "artificial intelligence"...

 

Last edited: Jul 25, 2024 at 8:47 PM

I said exactly the same. Modern lies and fraud in the form of new names. When people do not have enough information about something, they are fooled by new names. This technique of deceiving people has been done for thousands of years and unfortunately, as we can see, it still has enough power to deceive people. People should not be fooled by new names. Artificial intelligence by its worshipers has become an idol, and as we know, idols are finally broken.

 

If the distribution supply chains consider music as "content", then the only way to change this is to convince the audience that music is special. That means making better music. AI is not better music. It is regurgitation of what already existed prior. And if AI creates content from this, then what existed prior is AI and it will generate more music from itself. The downward spiral continues.

Unless the audience is willing to pay real money to real musicians, then real studios will never survive. And unless musicians want to (gasp) leave the bedroom and go out and make music a social thing, there's no need to have a destination to go to for recording it - assuming that there is something worthy of being recorded. The economics will dictate that if the musician doesn't consider their art special, they won't devote funding to having a 3rd party record & mix it. And if the distribution supply chain doesn't want special - they just want quantities of content for the hypnotic "like & subscribe" crowd, then the downward spiral can never be reversed.

The demise of the recording studio is not about technology. It is about the audience not demanding to have something special to listen to.

 

I would suggest that there isn't a music industry anymore. The power of being able to produce content on your laptop and an audience that refuses to spend money directly to the artist has removed the need for any industry. Record labels don't make records, so they have become simply marketing agencies. They prefer to be considered "artist management" agencies, but in truth they are just there to sell. They monetize and make a profit because they have legal ownership to back catalogs that they can milk for revenue for sync licensing, sue the Spotify's of the world, etc. But they don't have budget to punt on the career of an artist now. The artist has to first generate 10 million TicTok followers and then maybe they will devote some of their marketing staff to the cause.

So there is no industry. We are the industry. With this, however, comes a lack of curation. That has been given to the algorithms to do. The curation is machine learning of our listening preferences, and it suggests what we should be listening to. There could be an argument that this is superior to the way music was curated in the past, but the one advantage that past curation (ie. record label deals, etc.) is that it was driven by economic success and some A&R guy or gal was punting that your music would sell records, and would be allowed onto radio playlists, etc. Now anything is fair game, although the A&R guy or gal is now some code trying to work out what you might enjoy listening to.

 

I bet you are ;)

Maybe we can agree to disagree.

We're in the early days of AI and probably the latter days of humans, who knows what tomorrow will bring? Our new silicone friends might surprise us.

 

That's a valid point and in the current state of AI, what if instead of letting it trawl through actual audio LLMs we only gave it music theory with no human context? We would also need to train AI to experiment outside of these rules we give it. The whole subject is a grey area to the layperson (of which I am one) but maybe some scientists are working on this.

My original point was wondering what AI would come up with if it had no access to human data at all, only the real-world elements that humans manipulate to make music. We didn't create the octave, it's a natural feature and relates to the doubling of frequencies already existing in the natural world.

 

A very bad example.
Birdsong is just birdsong, just like a cat's purr is just a cat's purr.

Just because the word “song” is in birdsong doesn't make birds musicians, because birds simply reproduce existing melodies and sound patterns they have picked up somewhere during their imprinting phase. Once they have learned a few patterns, they string them together. That's it. If that was supposed to be an example of originality in the creation of music that stands out from what machine learning does, then, as I said, it was a very bad example.

 

I didn't mean to suggest that birds can be considered as musicians. They may well be making music without human rules, though some are known to use pentatonic scales (hermit thrush)!

I did a brief Google search on whether birdsong is music or not and... Seems debatable, a grey area. Quite interesting actually. I also asked ChatGPT4 (note: I don't consider this response empirical data, I offer it only for interest sake):

Birdsong has long captivated listeners from all walks of life. The question of whether it qualifies as music is intriguing. Let’s explore:

1. Similarities with Human Music:
   
   • Avian choristers, such as pied butcherbirds, share musical sounds and behaviors with human musicians.
   • They exhibit approaches to repetition, variation, and balance in their vocalizations.
2. Complexity:
   
   • Some argue that birdsong lacks complexity, but consider the brown thrasher with its vast repertoire of thousands of phrases.
   • Pied butcherbirds, for instance, take motifs to expressive limits through intricate combinations and recombinations.
3. Hard-Wired vs. Cultural Transmission:
   
   • While humans are the only primates with vocal learning capacity, songbirds, parrots, hummingbirds, elephants, bats, and marine mammals can learn vocalizations.
   • Birds’ songs result from cultural transmission, manifesting traits passed down over possibly millions of years.
4. Female Birds as Music Critics:
   
   • Ornithologist Richard Prum suggests that birdsong is akin to music because female birds’ preferences shape male songbird behavior over time.
   • If female birds appreciate a male’s song, they reward it with reproductive success1.

In essence, while birdsong may not fit our traditional definition of music, its beauty and complexity resonate across species.

 

Won't happen, because iwhat we call AI is not a real AI.
Our "AI" we use, is just a bunch of algorithms reacting to our prompts.
A real AI would be completely autonomous.

 

You said it won't happen but you're using the current state of AI (actually non-AI, I agree) to prove your point. We can't prove the future based on the present. I stated what I would like to hear, I didn't claim or suggest that it's possible right now or even will be in the future, though I hope it will be.

 

[Rabbit Hole]
About this debate I asked META LAMA 3.1:

I want to ask you why this is not the way that humans create music, and what is differentiating the music coming from ai and humans. And where music comes from.

And his response, very accurate, AI is faster than me, what can i add to this?:

But why humans cant be using this "brute force noise diffussers based on billions of parameters"?

I still have the doubt of why humans can't be just a pile of past events, that are creating music based on billions of parameters as AI's do. But I will pass of asking AI again. (Edit: finally i did it f*ck)

This answers could be influenced by religious and phillosofical points of view?

So AI told me "astute observation" so my ego has enough. LOL

 

Last edited: Jul 25, 2024 at 8:52 PM

Man oh man these romantic musicians seems to not know at all what the future controllers of real AI will gift them then .

 

II have listened to numerous Soul/R&B songs created entirely by A.I. that completely impressed me. Initially, I even believed they were actually made by humans.

 

Devil's Advocate: I agree that what we are now calling "AI" is merely algorithm, and not autonomous or self aware. But I wonder if our own "intelligence" (such as it is) is not simply algorithm brought to a sufficient level of complexity.