Rick Rubin on AI (& now Graeme Revell, too)

Anything about MUSIC but doesn't fit into the forums above.
Post Reply New Topic
RELATED
PRODUCTS

Post

VOODOO U wrote: Fri Jan 23, 2026 10:46 pm Welp. While you guys are all mumbo jumbo, i now have 6 songs done sans vocals
Yeah, we've got al the songs for an album now. Fleshed out from what Tunee gave us, it's about 48 minutes worth, over 11 songs and a couple of interstitials. Depending how the vocal recording go, we have enough that we can drop one or two if we want, although I am increasingly confident I can pull the vocals off well enough (and nothing liek the AI vox).
Turns out Boner was right. This shit works. You just have to be patient, persistent and have good production skills to get past the formulaic generic stuff A.I. spews out.
Yeah, it's amazing what you can learn when you keep on open mind and actually try something for yourself, isn't it?
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Audient EVO 8 | Lumi Keys | Studio Pro 8
Korg Odyssey, bx-oberhausen, Proxima, PolyMax, GR8, JP6K, Union, Atomika,
Invader 2, Flow Motion, Olga, TRK 01, Thorn, Spire, VG Iron

Post

Come on the Lord of Acid is just you from a second account right?
When the data is corrupt in the Desert of the Real, Beyond the Last Thought, where intuition reigns, is the solace that will embolden and strengthen the soul, giving hope once more to this age of failing technique. eassae.com

Post

BONES wrote: Sat Jan 24, 2026 2:01 am
ghettosynth wrote: Fri Jan 23, 2026 7:20 pmI still can recall stopping dancing mid track because of bad filter artifacts/distortion in the late 90s and watching while everyone kept dancing because they didn't notice.
Yeah, that's because they were into the music and you weren't. I feel kinda sorry for you if that's all it takes to throw you out of the moment.
It's because they had shit ears and were not bothered by crap production, just like the mainstream now. Which is why AI doesn't matter as much as some here think that it must They also can't hear shitty DJing.

What they do seem to care about today is being duped by AI. Thinking that they are buying human work and getting AI work. Which is interesting, but not because AI work isn't "creative." It's interesting because if you were duped in the first place, you actually liked it. It's only not a good value at the point that you believed that you were suckered. Which means, that you weren't engaging with the music so much as you were engaging with identity at some level.
What is interesting to me about this is how good the results are with respect to how large, really small, the model is.
How does it work on less well known genres? Because that was initially the problem with things like Suno, it simply didn't know what EBM was so it took the rest of the prompt and made heavy metal from it, because that was something it understood that fit with the rest of our goals.
Probably terrible. That's a mainstream model problem. It's also related to things discussed here. The further out you go from the mainstream the less data that there is to train and the less ability it will have to model that music. I would suggest to anyone wanting to seriously use these models to write music or parts of music using AI to use the mainstream/frontier models as they are going to be pushing the envelope where it's challenging, e.g., increased context length to allow for longer tunes and front to back reason while also increasing token resolution to allow for greater fidelity around microtemporal events, e.g., trills, legato-transitions. These two properties are at odds with each other with respect to memory size and compute scaling.

Hence, progress is made at the algorithmic level, better modeling/tagging of those features. You aren't going to get this in a tiny model.

My recommendation was more about those who want to play with something in a private setting for whatever reason. It could be as simple as you don't want to put your lyrics in the cloud before you publish your song yourself.

For me the interest is that it yields both a usable open model as well as a usable open architecture that is ripe for experimentation on a small scale. By small, I don't mean the level of small that you can play with on your mid-range gaming GPU. You'll need some reasonable hardware. By experimentation I don't mean anything that you guys are discussing here, that's of no moment to me beyond have a bit of fun with it in lieu of pushing up my post count here.

Post

ghettosynth wrote: Sat Jan 24, 2026 3:52 amIt's because they had shit ears and were not bothered by crap production, just like the mainstream now.
More than a little arrogant of you to assume, don't you think? What do you think makes you so f**king special? A good song is a good song, production won't change that.
Which is why AI doesn't matter as much as some here think that it must They also can't hear shitty DJing.
Is there any other kind? The most successful club I was involved in succeeded because we played video clips off VHS tape, which we'd dubbed from other VHS tapes. So we had shit sound, poor quality video and basically no ability to cue anything up, yet we got around 3,000 punters a week through the doors, over two nights, because we were playing what they wanted to hear/see. You're worrying about things that don't matter.
Thinking that they are buying human work and getting AI work. Which is interesting, but not because AI work isn't "creative." It's interesting because if you were duped in the first place, you actually liked it. It's only not a good value at the point that you believed that you were suckered. Which means, that you weren't engaging with the music so much as you were engaging with identity at some level.
No, what it means is that they were engaging with the music and now maybe they feel a bit embarrassed to realise it wasn't what they thought. Kinda like f**king a fat chick when you're drunk, then waking up the next morning and realising what you've done. But the reality remains that if they didn't find out it was AI, they'd still be liking it now. And, as I say time and time again, it's the music that matters.
The further out you go from the mainstream the less data that there is to train and the less ability it will have to model that music. I would suggest to anyone wanting to seriously use these models to write music or parts of music using AI to use the mainstream/frontier models as they are going to be pushing the envelope where it's challenging, e.g., increased context length to allow for longer tunes and front to back reason while also increasing token resolution to allow for greater fidelity around microtemporal events, e.g., trills, legato-transitions. These two properties are at odds with each other with respect to memory size and compute scaling.
Is that how it works? I think it just pulls shit out of the aether and makes a song because sometimes a vocal can just be gibberish, so it doesn't feel like it assembled it in the same way you or I would. We definitely have issues with the breadth of it's ability in some areas. e.g. I haven't heard it try to put guitar into anything. Maybe Craig's told it not to but it seems odd, given that it's reasonably prevalent within the genre and we certainly use guitar parts. But we've fed it lots of different reference songs, yet it always seems to shit out something that sounds more like the other stuff it's shit out than it does like the reference track.
My recommendation was more about those who want to play with something in a private setting for whatever reason. It could be as simple as you don't want to put your lyrics in the cloud before you publish your song yourself.
Right, because your (as in, whoever's) songs are so f**king good, everyone is going to want to steal 'em. I'd post mine on a billboard in George St, I couldn't care less if someone ripped 'em off. On the contrary, I'd be quite flattered.
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Audient EVO 8 | Lumi Keys | Studio Pro 8
Korg Odyssey, bx-oberhausen, Proxima, PolyMax, GR8, JP6K, Union, Atomika,
Invader 2, Flow Motion, Olga, TRK 01, Thorn, Spire, VG Iron

Post

BONES wrote: Wed Jan 21, 2026 11:47 pm Peace is easy, it comes with solitude. But I'm not looking for peace, I'm looking for anarchy. The end of civilisation cannot come quickly enough for me. 2040 is apparently the magic date so we won't have too long to wait.
Why is 2040 the magic date? Edit: I see it's an MIT prediction.

Civilisation got to the point to where there were enough resources such that young people could get together in groups, take substances, and participate in a wide range of simulacra, an almost endless range of styles of simulacra.
Is there more to Bones' genre than a simulacrum of anarchy? If someone wanted to understand anarchism as a social formation they could read endless books, and discover that there isn't a particularly good solution to the problems of hierarchy... which is why we're stuck with sprawling bureaucracy.

Maybe there's a bit more to it than just a bunch of people pretending they're angry at an undefined simulacrum of a 'Machine'. Maybe it's all those German bureaucrats who spend all day filling out forms, or workers in a Volkswagen factory doing the same repetitive task all day, and wish they could claim their pension already; they need their own version of fight club. But the music sounds more like it's about the Terminator movies than the reality of a Volkswagen worker.

And if the music/genre is somehow expressing the frustration of serving corporate overlords there's irony in the fact that it's reasonably likely that AI is in the process of cementing a kind of corporate takeover of everything, the grand Machine Mmmachine Mmmachine. GRRRR.
Every day takes figuring out all over again how to f#ckin’ live.

Post

chagzuki wrote: Sat Jan 24, 2026 2:48 pm And if the music/genre is somehow expressing the frustration of serving corporate overlords there's irony in the fact that it's reasonably likely that AI is in the process of cementing a kind of corporate takeover of everything, the grand Machine Mmmachine Mmmachine. GRRRR.
Also ironic putting it out on Spotify—I know, "it's the labels fault."
When the data is corrupt in the Desert of the Real, Beyond the Last Thought, where intuition reigns, is the solace that will embolden and strengthen the soul, giving hope once more to this age of failing technique. eassae.com

Post

BONES wrote: Sat Jan 24, 2026 2:06 am Yeah, we've got al the songs for an album now.
Does your bandmate use various prompt methods or is it always the rinse and repeat method where the details are prompted multiple times in one project?
Yeah, it's amazing what you can learn when you keep on open mind and actually try something for yourself, isn't it?
Absolutely. Having an open mind is basically an opportunity to get out of the comfort zone and i feel that's where the majority fail.

Post

chagzuki wrote: Sat Jan 24, 2026 2:48 pmWhy is 2040 the magic date? Edit: I see it's an MIT prediction.
It's better than that. The prediction was first made in the 1970s and a few years ago some grad students thought to check and see how the numbers were tracking. They discovered that it was all going exactly as predicted. Exciting stuff.
Is there more to Bones' genre than a simulacrum of anarchy? If someone wanted to understand anarchism as a social formation they could read endless books
As soon as you have a "social formation" it ceases to be anarchy.
discover that there isn't a particularly good solution to the problems of hierarchy... which is why we're stuck with sprawling bureaucracy.
I don't know that I am interested in a solution, I just want to see the current system fail. I'll be 82 in 2040, what happens next will likely not affect me but people have had it too easy for too long. Things need a shake-up.
VOODOO U wrote: Sat Jan 24, 2026 8:02 pmDoes your bandmate use various prompt methods or is it always the rinse and repeat method where the details are prompted multiple times in one project?
I'm not sure but I imagine it's a bit of both. Sometimes he'll send me 3 or 4 songs that sound different but have the same or similar lyrics, other times it will be something unique. I know he feeds it a lot of different reference tracks but it often makes little or no difference to the output.
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Audient EVO 8 | Lumi Keys | Studio Pro 8
Korg Odyssey, bx-oberhausen, Proxima, PolyMax, GR8, JP6K, Union, Atomika,
Invader 2, Flow Motion, Olga, TRK 01, Thorn, Spire, VG Iron

Post

BONES wrote: Sat Jan 24, 2026 10:36 pm I'm not sure but I imagine it's a bit of both. Sometimes he'll send me 3 or 4 songs that sound different but have the same or similar lyrics, other times it will be something unique. I know he feeds it a lot of different reference tracks but it often makes little or no difference to the output.
Cheers. I'm going to experiment with reference tracks.

Post

Looks like the rubicon has already been crossed...Study shows Majority of listeners are either very comfortable, comfortable, or indifferent to consuming AI generated music. Only 44% say they would be less interested in listening to music if know it is AI...how likely is this number to increase?

viewtopic.php?p=9190687#p9190687
Music had a one night stand with sound design.....And the condom broke

Post

Study shows 100% of me doesn't give a f**k. As long as I'm making music that I like everything and everyone else can eat a weiner with relish.

Post

VOODOO U wrote: Sun Jan 25, 2026 9:02 pm Study shows 100% of me doesn't give a f**k. As long as I'm making music that I like everything and everyone else can eat a weiner with relish.
Thanks so much for sharing this piece of wisdom right here. This community is very much appreciative of your depth.

Post

csscp wrote: Sun Jan 25, 2026 9:57 pm
VOODOO U wrote: Sun Jan 25, 2026 9:02 pm Study shows 100% of me doesn't give a f**k. As long as I'm making music that I like everything and everyone else can eat a weiner with relish.
Thanks so much for sharing this piece of wisdom right here. This community is very much appreciative of your depth.
What's really interesting about this community being appreciative of my wisdom is i really don't give a wiener relish f**k. I'm too busy selling out and writing hit songs.

Post

bermudagold wrote: Sun Jan 25, 2026 5:01 pm Looks like the rubicon has already been crossed...Study shows Majority of listeners are either very comfortable, comfortable, or indifferent to consuming AI generated music. Only 44% say they would be less interested in listening to music if know it is AI...how likely is this number to increase?
viewtopic.php?p=9190687#p9190687
The key point to note here is "if (they) know". The reality is that even now 99% of casual listeners wouldn't know unless someone told them. In a year's time nobody will know.
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Audient EVO 8 | Lumi Keys | Studio Pro 8
Korg Odyssey, bx-oberhausen, Proxima, PolyMax, GR8, JP6K, Union, Atomika,
Invader 2, Flow Motion, Olga, TRK 01, Thorn, Spire, VG Iron

Post

And therein lies the misery of our future. I hope you can induce the insurrection you desire Mr Bones!

How sad for humanity. I hope it does all end soon so they don't have to endure it.

Post Reply

Return to “Everything Else (Music related)”