"AI is going to have a profound impact on your business and your art"

[ras.s]

I was reading about the new Developer Challenge and the grand prize is a trip to Boston to visit a tradeshow in September, the A3E (Advanced Audio + Application Exchange). There's also the donation prize fund, so let's bump that up as well..

So I went to read about what is this A3E about and that lead me to their website and their Facebook page. And further on from the Facebook page to a mailing list post/press release, which is here and I'd like to talk about that.

Here's the content, basically:

"Artificial Intelligence (AI) is no longer a Sci-Fi myth, but a core component in the creative tools that are redefining the Musical Instrument landscape. A3E will explore how Artificial and Algorithmic Intelligence is being implemented in audio technology today, and how the brightest innovators see it used in the future. There is no question that the creative role of machines is changing dramatically, and the future of production will require us to adapt and embrace the machine as a creative partner. Whether you're a developer, an artist, an educator, or a manufacturer, AI is going to have a profound impact on your business and your art" offers Doug DeAngelis, A3E Conference Chair.

A3E is assembling some of the brightest minds in computer & music technology to discuss where the industry is heading in the next 10 years, and the effect it will have on how music is composed, produced, and performed. In recent years, there is a rapid movement toward 'intelligent' applications, where music and audio software replace the need for human expertise in audio engineering, arranging, composition, production, and performance ability. What will be the global impact on the industry if artificial intelligence is one day capable of creating art?

What to think of that?

We've already passed the point where it's hard to tell whether a recording is "real" or made with a computer (that is, the sound of the recording). But artificial intelligence -- will it be impossible to tell whatever a recording itself is made by a human or a computer? I don't doubt it will, and I wouldn't be surprised if it already is.

I'd feel dirty about it. Sure, every now and then I would really appreciate having a drummer that I could command to please my musical quirks, and I might as well (occasionally) be content with it being a computer program, but to have a computer algorithm compose music for me altogether? It's quite common here to bash people about making "generic EDM" that sounds like it is both sonically and compositionally cut-and-paste kind of stuff, I really don't have much to say about that -- everyone to their own. But to have a computer program compose that music is something that to me sounds just wrong. Why do any of this if computers can do it too -- and without doubt, they can be programmed to be experts of any genre, and play instruments without the years of learning it takes to learn to play instruments.

There are probably uses for music made with AI, but I don't know what they could be. Games, commercials? Even then it would be away from actual musicians and composers. Having a practice partner might be nice, though. Will AI be just another instrument in our musical arsenals?



What are your thoughts on making music alongside an algorithm, or even letting the algorithm compose your music for you? What uses can you think for AI in music? What do you think what kind of impact AI will have in making music in the future?

[jancivil]

I think the notion of AI composing viable music is wildly overconfident. You bring up an AI drummer: well a primary aspect that has to be sussed is predictive algorithms for human timing and I don't even think it's been studied intelligently. I think the wrong people for the job are the ones doing it, I think there is a lot of carts pulling horses.
Humans think and respond to a musical climate and context, vis a vis other humans, with history and opinion and imagination. You aren't going to make a machine sentient any time soon. Unless we have that, this is going to be quite limited in scope. I don't have much to say about the potential use for it, albeit there is a lot of music made today that doesn't seem to require much of these human capabilities, maybe teach some people a lesson about their place in the thing.

I think these claims are spurious and self-serving, academics creating interest in the things they write about, justifying their position: 'A3E Conference Chair'.

[jancivil]

We've already passed the point where it's hard to tell whether a recording is "real" or made with a computer (that is, the sound of the recording). But artificial intelligence -- will it be impossible to tell [whether] a recording itself is made by a human or a computer? I don't doubt it will, and I wouldn't be surprised if it already is.

Impossible to tell what recording was made by human or computer? There are a lot of things that may as well have been all machine because the human couldn't be arsed to do more.

The reason people are convinced my music made with the help of the machine is 'real music' is a result of MY INPUT; my timing convinces, I know from years of experience about give and take and push and pull, the dynamics. Knowledge, in the fullness of that word. For that to be replaced by a machine, you have to build a personality, a full personality and create a history of memories, just to even begin. Then a history of scenarios where this set of things that makes a person is applied to events, meaningful events. It's enormously complex. I bet they aren't talking about any of this.

While there are algorithms which are clearly used, and useful for certain components of a *workflow*, extrapolating that to 'creating art' is a great leap in my estimation and I think a person ready to say that must have a rather different definition of 'art' than I do, or that I'm very interested in entertaining.

[camsr]

Marketing will have to keep up with these new trends. Imagine "AI sucks" t-shirts everywhere.

[Gamma-UT]

What are your thoughts on making music alongside an algorithm, or even letting the algorithm compose your music for you?

Why not? It already happens – there are tools out there that will do it to do things like four-part harmony or rhythm variations. The main issue with these tools is that nine times out of ten, it's easier and quicker to do it yourself than to configure the tool and then discover it didn't really do what you wanted.

What uses can you think for AI in music? What do you think what kind of impact AI will have in making music in the future?

I can think of a lot of uses for AI in music – gameplay-dependent music in games is the most obvious. But let's be realistic, proper AI is still out there somewhere. We're really talking about algorithms.

There's a natural assumption that the only use for algorithms is to replace things that humans already do well, which doesn't make a lot of sense. Machines are good a following rules, not throwing the curveballs to the listeners that the listeners enjoy. Anticipation and surprise are two of the main ingredients for good music. An algorithm could do the first, surprise is a bit trickier.

But they do let composers explore different spaces. When things like Noatikl first came out I was a bit disappointed at how hard it was to do conventional things with it. Then, on reflection, I thought maybe that's not the point – maybe it's for making music that humans find hard to do, such as proper randomness. If you think about it 'stochastic music' as envisaged by Xenakis, Cage etc isn't really random – it's more about musicians' reactions to each other. A computer lets you throw proper randomness into the piece that people would find hard to simulate, so it does open a new space.

[ras.s]

jancivil,

I mean that part "whether a recording is "real" or made with a computer" more about the use of samplers, etc. It's relatively easy/common to fool a person to think a drum part is actually recorded in a studio, or that the string section is actually made of people's blood and sweat. It no longer takes years of playing an instrument to make a convincing recording with that instrument. So that "fooling" is already possible. Whatever it's going to be possible to use an algorithm to compose those parts convincingly (or in a manner that atleast produces interesting results), I don't think it's impossible, and some of it is already done, I think.


Another thing about AI as a musical vehicle is that it can be used to create music that isn't based on currently existing genres or musical traditions (or the history and personality of the composer). Take xoxos stuff for an example; he has made drum sequencers that produce drum parts that are independent of genres, they sound like their own personality and as such, are very creative. Whatever those results can be merged with others results is still something that is up to the composer.

[tetsuneko]

Is this recent AI hype related to the Passing of the Turning test that happened a while ago btw?

I for one welcome our new AI producer overlords. Generic me-too genre music scene is going to change dramatically - If you thought Beatport et al is getting crowded by new releases, you haven't seen nuttin yet We will be literally flooded with music to "go through"

Having a DAW that learns from your actions is only going to make you repeat yourself even more, but you'll be working hella faster.

Sign me up for a decent AI engineer tho, I can just make all the "creative" decisions and the AI can do my edits and all other project management. Something like that I could certainly appreciate.

[fese]

Is this recent AI hype related to the Passing of the Turning test that happened a while ago btw?

Nope, that only happened recently. And it was harshly criticized, as it seems they modified the rules so the software could pass the Turing test more easily. ("Passing of the Turning", while being a typo, sounds good. Like an album title from a nineties heavy band.)

I think one has to differentiate between AI and rule-based algorithms here. Music is mostly based on rules, creating and composing music based on given rules (creating melodies, arranging in different styles, playing instruments) has been done, but that doesn't qualify as AI in my opinion.

AI would be creating something new based on existing rules, like mixing heavy metal guitars with hip hop beats and thus creating a new genre. A computer would need to learn if and how that works, maybe fail a lot an learn from those failures. Humans do that. Except sometimes they don't learn, like when mixing heavy metal bands with a symphony orchestra...

[Touch The Universe]

Artificial Intelligence (AI)

[LimboLoves]

I find it hard to believe that AI can release any kind of music, except the generic EDM styles. Can't add lyrics and probably won't have a human element of passon in it either.

Art, however, should be scared.

[Arglebargle]

for artists who create music for fun not profit, it doesn't really matter. could it take away jobs, theoretically, but humans have a need to express themselves, so it's not like everyone is going to drop their instruments.

[rcat]

A.I. won't be realized until quantum computers are fully developed. This is not going to happen for quite some time, no matter what anybody says. In the meantime try Rayzoon Jamstix.

[Hink]

for artists who create music for fun not profit, it doesn't really matter. could it take away jobs, theoretically, but humans have a need to express themselves, so it's not like everyone is going to drop their instruments.

well said

[ras.s]

Haha @ Rayzoon Jamstix.

I guess there's a lot of hype going on in that press release. There's certainly a distinction between full-on artificial intelligence and things like machine learning and algorithmic composing.

But stuff like what tetsuneko described there, having the computer act as sort of an personal assistant, that's an interesting idea. Sort of having the computer work with me than having it work for me. Though I'm not sure what I'd allow it to do on its own. Years ago I did a four month stint as a trainee in a music studio, and what I basically did was I made coffee for four months. :P I did learn quite a lot though during that time, very neat experience (and not just about making coffee).

And the press release in the OP does give a ten year time frame and I'm sure things will be different in the year 2024. What are "'intelligent' applications" it mentiones? And the whole question of computers making art is an interesting one, .. I mean, if some program already passed the Turing test (albeit questionably, but still) -- is inducing emotional experiences next?


Well I guess we'll wait and see. I hope they do some sort of KVR video specials from the show as there seems to be some sort of affiliation going on.

[Rajah]

I'd feel dirty about it. Sure, every now and then I would really appreciate having a drummer that I could command to please my musical quirks, and I might as well (occasionally) be content with it being a computer program, but to have a computer algorithm compose music for me altogether?


What are your thoughts on making music alongside an algorithm, or even letting the algorithm compose your music for you? What uses can you think for AI in music? What do you think what kind of impact AI will have in making music in the future?

I guess you have never toyed with Jamstix?


http://www.rayzoon.com/jamstix3_vid.html