What's new

SPINOFF: Made with AI: Suno, Udio and others - Discussion

How is this a bad software exactly? And besides, what comes first, good music (which current models occasionally achieve) or "Me!"....
He didn’t say bad. It’s technically incredible which doesn’t mean it can’t be morally abhorrent at the same time.
The art of creation (or creation of art if you will) should always come first, creative thinking is one of the defining characteristics of intelligent life.
 
He didn’t say bad. It’s technically incredible which doesn’t mean it can’t be morally abhorrent at the same time.
The art of creation (or creation of art if you will) should always come first, creative thinking is one of the defining characteristics of intelligent life.
Ok - so how this approach to "creation" is bad, once the result is good? I don't touch the whole "stolen" issue here, lets assume it was all legitimately trained...
 
Ok - so how this approach to "creation" is bad, once the result is good? I don't touch the whole "stolen" issue here, lets assume it was all legitimately trained...
Why should I assume when it categorically wasn’t.

Anyway, creation is a much more nuanced process than providing a machine with a set of instructions. You are no longer an artist, the artists are the few who can create such complex ML algorithms for the models to ingest.

Is asking Siri to turn on the light the same as physically turning on the light? The end result is the same the process is not.
Asking ‘insert AI name here’ to produce you some music in the style of Vagner is not a process of creation.
 
Asking ‘insert AI name here’ to produce you some music in the style of Vagner is not a process of creation

I never suggested "I am creating" the music via AI, that would be silly. I am merely managing the process, but the good results is what matters in my book.
 
Why should I assume when it categorically wasn’t.
Yeah it is a bit hard to give AI creation credit while it emulates mathematically all of the music available on the internet, sometimes blatantly using voices of famous singers, dead or alive.

One think is the writing part, and one thing is the "production value", the sound. The latter is very very good surely. The writing is hit and miss and mostly miss I find - for now.....
 
I never suggested "I am creating" the music via AI, that would be silly. I am merely managing the process, but the good results is what matters in my book.
That’s a fair view and when everyone can do exactly the same with the minimum effort and zero actual skills, what then will set you apart? What makes you worth paying vs the next person?
 
That’s a fair view and when everyone can do exactly the same with the minimum effort and zero actual skills, what then will set you apart? What makes you worth paying vs the next person?
Sorry, its seems you are not aware that AI does need expertise and effort for actual work, and results vary wildly between users, immediately revealing their level.
 
Sorry, its seems you are not aware that AI does need expertise and effort for actual work, and results vary wildly between users, immediately revealing their level.
AI will learn from experts and quickly allow everyone to obtain similar results with much less work and knowledge . So that " actual work " is mostly needed for machine learning .
 
AI will learn from experts and quickly allow everyone to obtain similar results with much less work and knowledge . So that " actual work " is mostly needed for machine learning .
This is not how it works. Even meeting the best robot in the world - you would still have to explain him WHAT you need. Direct brain connectivity?? The results will start to be tenfold more varied between people.
 
This is not how it works. Even meeting the best robot in the world - you would still have to explain him WHAT you need. Direct brain connectivity?? The results will start to be tenfold more varied between people.
You gave a brief to a machine. Thus, YOU created fuck all, regardless of how skillful your brief(s) may be.

If you gave the same brief(s) to a human composer who then created a piece of music for you, who was the creator of that piece of music? Certainly not you….
 
You gave a brief to a machine. Thus, YOU created fuck all, regardless of how skillful your brief(s) may be.

If you gave the same brief(s) to a human composer who then created a piece of music for you, who was the creator of that piece of music? Certainly not you….
Except the line is fuzzy. People work with loops, arpeggiators, and various other algorithmic content generators all the time and yet that generally does not call into question the status of the resulting "composition."

At some point the brief becomes so explicit that it verges on being a text version of a DAW project. I'm not saying this is possible now or that it will ever really be possible, but it is conceivable and so it's bound to lead to a place of utmost confusion.

I don't think we ever did decide culturally about other forms of algorithmic composition and whether and to what extent they should indeed count as "composition" just because someone had the sense of setting up the algorithm and running it. Is setting up such algorithms really that different from giving a brief to an AI?

I find AI poses interesting thought problems of this sort even if I'd rather not have to confront them in reality.
 
I don't think we ever did decide culturally about other forms of algorithmic composition and whether and to what extent they should indeed count as "composition" just because someone had the sense of setting up the algorithm and running it. Is setting up such algorithms really that different from giving a brief to an AI?
Setting up a generative algorithm would be the equivalent of inventing an ai not giving simplistic instructions to an existing one that will end up producing very similar results for all users . Generative tonal music implies some musical knowledge if you want to avoid a cacophony or a " modern " contemporary classical music from the 50's type of sound .
 
Would it come as a surprise, if I refer to people entirely using loops and arpeggiators as ‘playing with Lego’ ? In my book they’re as much of a musician as a DJ…

A musician can either sing or play an instrument. I want to hear the actual musical skills of the performer. A super skilled ‘prompter’ still only provides text, no matter how you angle it. I don’t buy it.
 
Setting up a generative algorithm would be the equivalent of inventing an ai not giving simplistic instructions to an existing one that will end up producing very similar results for all users . Generative tonal music implies some musical knowledge if you want to avoid a cacophony or a " modern " contemporary classical music from the 50's type of sound .
That doesn't really answer the problem of the fuzzy boundaries. And I don't think that setting up a generative algorithm is necessarily more complicated than devising a prompt for an AI. Or rather, it's perfectly possible to set up a very simple generative algorithm on par with a simple prompt for an AI. It is likewise conceivable that a very complex prompt, on order with a more subtle, more complicated generative algorithm, could be given to the AI. Indeed one could more or less give the same prompt to the AI that you'd program for the generative algorithm. Not to mention which that cacophony of 1950s and 60s music did not exclude it from the domain of "composition," though it did raise the question for some, I suppose, of whether it was indeed music.

The point is that because we have come to allow all sorts of algorithms in music composition (and at this point we can hardly exclude them) it will be hard to ever draw a line that effectively excludes AI, because what is your ground going to be. It can't be because the machine is doing it.

There is of course the legal side, and copyright that holds some promise of arresting the use of AI. But only some promise.
 
Would it come as a surprise, if I refer to people entirely using loops and arpeggiators as ‘playing with Lego’ ? In my book they’re as much of a musician as a DJ…
That may be so, but you are very much in the minority on this, and there is a boatload of commercial music that leans heavily on such devices.

In any case, unless you completely unplug your music making from the box, at some point you will hit the machine and it doing something automatically and on your behalf that you have not personally programmed.
 
That may be so, but you are very much in the minority on this, and there is a boatload of commercial music that leans heavily on such devices.

In any case, unless you completely unplug your music making from the box, at some point you will hit the machine and it doing something automatically and on your behalf that you have not personally programmed.
Arpeggios and loops have been used in " modern " music long before the box became a thing . Samplers and synths have always been seen as instruments by most and some always thought that it was cheating : synths generate music automatically , sampler are used to steal other people's music ... that is a very different debate .... Now the " it has always been this way " argument does not work here , it would be hard to tell musicians : " you have used drum loops In your music in the past , you now have to accept AI because it is the same thing "
 
Arpeggios and loops have been used in " modern " music long before the box became a thing . Samplers and synths have always been seen as instruments by most and some always thought that it was cheating : synths generate music automatically , sampler are used to steal other people's music ... that is a very different debate .... Now the " it has always been this way " argument does not work here , it would be hard to tell musicians : " you have used drum loops In your music in the past , you now have to accept AI because it is the same thing "
I'm making no claim that you have to accept AI. In fact, I've been pushing back against it, suggesting composers need to stop dreaming of using AI to replace other musicians and creative labor in their music making. I am making the claim that because algorithmic devices run deep in contemporary music making, the lines are fuzzy and so it will be hard to form a cogent argument against AI. Arguments against AI inevitably disqualify other practices that are perfectly common. So what do we do about that?
 
That may be so, but you are very much in the minority on this, and there is a boatload of commercial music that leans heavily on such devices.
Consider the song that's been #1 on the charts for the last few weeks.... "The song is built entirely around a sped-up sample of Rodney O and Joe Cooley’s “Everlasting Bass,” a 1988 synths-and-scratching classic that’s been sampled dozens of times by everyone from Lil Wayne to Three 6 Mafia. ... [using] this well-worn sample in 2024 is, I’d argue, a statement in itself: The beat is legit, but the lyrics are what matter."






In the intro the producer did chop and rearrange parts of the loop to create a dramatic rhythmic effect, and in other places also layered in some (by now fairly conventional) fast trap hi hat rolls over top of the sped up drums from the loop, as well as mixing in a processed sung vocal loop that rhythmically interlocks with the gaps in the drums, and doing some more (more or less creative) processing with compression and other effects. And arguably it's all primarily a backdrop (or a soundtrack) for the original vocals and the lyrics.

Obviously, AI generated loops can be used similarly.... And depending on how fine-grained the chopping is, it can be used to create new melodies as well as harmonies and interesting new combinations of sounds, concepts, rhythms, etc.
 
Top Bottom