AI and Art

[Somebody613]

The tech for music, language, and images is fundamentally the same. It requires consuming a huge body of other works and imitating that with variation.

Just as it doesn’t just randomly assign pixels, which given infinities would produce a great work of art but in a world with limits would make colored visual static, randomly creating sound waveforms from the basis of pure tones (which is how all sound is fundamentally made) will also in practice only ever create noise. Literally radio static fuzz. Given infinities, music will emerge, but there are not infinities. It’s not even AI if it’s just doing it randomly, anyway.

All AI music is training on real music and then fitting to stylistic convergences exactly the same as visual art. It is equally theft or not theft.

I disagree.
You could tell AI to use random sounds as starting points and then proceed in specific patterns that would produce more or less palatable sequences.
This would be much harder to do with text, but still somewhat possible to randomly cram random words into predefined positions for "verbs" and "nouns" separately.
And this would be impossible to do with pixels without feeding AI any actual ready picture drafts, which unlike music are NOT made via mathematic sequences.
This is the crucial difference:
Drafts for music and text are "rules" that only define TYPES of input data (closeness of sounds to each other, and word categories).
But pictures DO NOT follow any such "rules" that could be defined on the PIXEL level - if you want anything more complex than simple geometry shapes (which ARE math based).
This still doesn't mean that it's easy to do any of the first types - most of the results would still be garbage, but it's possible to randomly come upon something that SOMEONE would like.
Whereas in the case of pixels, it's nearly impossible to program a pattern that WOULDN'T already "steal" basic shapes (as "elements") or more.
Yet that would still only account for "kindergarten" level of "skill" - simple lines in schematic patterns.
But if you expect AI to be capable of painting even a simple apple in any "realistic" shape and color tone from scratch - yeah, nope.
Even be allowed to use "basic geometric shapes" would NOT help it to do so, lol.

 

[Hygro]

Music also isn’t defined at the pixel level. A note is not a pixel. A note is as complex as a brush stroke with its myriad “pixels” conforming to the pattern of the medium and technique. Music has proved more difficult to make convincing than visual art, compelling AI music is further behind compelling AI images.

Text is also way ahead of music.

All three rely on the humongous accumulation of existing human made material and its training on that material.

Random tones is as useless as random pixels, and randomness isn’t “AI”.

The likelihood of spontaneous random sound waveform generation creating anything other than radio static is as unlikely as spontaneous random pixel color producing a worthy painting.

Generative sound, image, and text AI relies fundamentally on the same principles and have similar challenges.

There doesn’t exist a sound generating AI that can exist in a training vacuum and produce anything musically coherent.

 

[Birdjaguar]

Are there any AI generated short stories yet? Novellas? Poetry? + Music?

 

[Angst]

Art for this thread can include: painting, drawing, movie scripts and images, books, TV shows, advertising, any published writing, speeches, etc.

music?

music IS math

no

The tech for music, language, and images is fundamentally the same. It requires consuming a huge body of other works and imitating that with variation.

Just as it doesn’t just randomly assign pixels, which given infinities would produce a great work of art but in a world with limits would make colored visual static, randomly creating sound waveforms from the basis of pure tones (which is how all sound is fundamentally made) will also in practice only ever create noise. Literally radio static fuzz. Given infinities, music will emerge, but there are not infinities. It’s not even AI if it’s just doing it randomly, anyway.

All AI music is training on real music and then fitting to stylistic convergences exactly the same as visual art. It is equally theft or not theft.

now that i've got you here hygro (also saw the other post), any good suggestion for AI music?

i think questionable content/deathmøle made a music program that generated a groove about the time of myspace (!) that sounded... boring but servicable, but i haven't really been presented anything convincing. after the breakout of ai image and text work i've seen a few attempts shared on youtube, eg emulating mozart's style or whatever, just in pure midi (stuff that has a HUGE backlog and could be iterated/farmed) that was atrocious. not atrocious-sounding (it's simple midi, so i was not disappointed there) but atrocious-composed, and not mozart at all

 

[Angst]

Are there any AI generated short stories yet? Novellas? Poetry?

that one harry potter chapter was nothing short of brilliant, but mostly because of its pure chaotic element of it. was absurd and made no sense

on poetry, ai usually only does good modern poetry when it glitches, but then it can somehow stumble into some really god damn cool stuff.

some danish writers have worked with releasing some ai works recently, i haven't read them yet. but there was a publishing some months ago. i think it's still just a novelty.

 

[Hygro]

music?

no

now that i've got you here hygro (also saw the other post), any good suggestion for AI music?

i think questionable content/deathmøle made a music program that generated a groove about the time of myspace (!) that sounded... boring but servicable, but i haven't really been presented anything convincing. after the breakout of ai image and text work i've seen a few attempts shared on youtube, eg emulating mozart's style or whatever, just in pure midi (stuff that has a HUGE backlog and could be iterated/farmed) that was atrocious. not atrocious-sounding (it's simple midi, so i was not disappointed there) but atrocious-composed, and not mozart at all

Mostly no haha.

Microsoft or someone had an "AI" music generator that was hilariously bad but sort of pulled it off. Like key and bpm at least. That would have been 15-20 years ago. There are great videos. Ok I just found it https://knowyourmeme.com/memes/songsmith

The most compelling AI music in the Dalle/Chatgpt full generative sense has been trap instrumentals. The worst I have found is techno, as the essence of techno (and I would imagine moreso for one of techno's two parents, funk) is deeply removed from its component sonic tropes. Instead of straight generation, I have however read about people describing their chords to ChatGPT and asking for suggestions and having good responses, that they then play themselves.

The specific-person voice emulators are really interesting, so if you can sing you can fairly convincingly swap your voice for someone else's.

There's a lot of plugins using machine learning. I have two drum sequencers / sample aggregators that map your samples against the other samples based on similarity in sound and get color coded based on whether they seem like kicks, snares, etc.

Stem splitting, first released by Deezer, is a remix godsend.

WavTool and others are trying to create generative-AI inclusive DAWs.

I have long dreamed of an AI that takes parameters from reading a dance floor and generates dance music, where occasionally based on the mood and behavior of the crowd would go way off the rails into new music territory.

But at this time pure generative AI music is garbo.