Arturia V Collection - Development to Version 12

[swilow11]

I've had ChatGPT give me formulae for use in the Serum wavetable editor based on my description of timbre. It took several iterations before the formula was usable and somewhat close to my description but it worked. Hadn't thought it could make me a full preset though...

[oobesan]

Is this for real? I am a ChatGPT-using fiend. However, I had no idea it could design patches (although I suppose that isn't a stretch since it can write code). But how would it generate the preset files? You must be pulling my leg.

Dead serious, I pay for the $20 a month plan not sure if it's available on the free plan, on a whim I asked it if it understood the DX7 and it said yes and gave a description of one. It then asked me if I wanted it to design a patch and I said yes and it asked me what kind, I said a pad and it spit one out along with a patch sheet of all the parameters then it asked me if it could generate a Sysex file. I said yes and it sounded awesome

I then asked it if it could design an additive patch for the Synclavier. It said yes then I asked it for Sysex and it said it couldn't do one as it didn't use Sysex which is true, then I asked if it understood the Arturia Synclavier V and it said yes so I asked if it could make the patch for it and then asked me if I wanted the file for the patch and it generated it and it sounded again amazingly good. I then asked about a bunch of other Arturia Synths and it said it could generate patches for just about all of them

I was blown away

Wow, that's amazing. However, I feel we will be saying "that's amazing" more and more often as AI becomes increasingly more capable, even if it never becomes consciousness, let alone reaches some God-like "singularity."

Before AI-hating trolls fly off the handle, AI is a LONG way from becoming self-aware. If you ask ChatGPT how it "solved" a math problem, it will describe the approach it took, which turns out to be correct mathematically. But if you look under the hood, that's not at all how the system arrived at the answer, which uses a roundabout, language-based, probabilistic technique requiring a lot of computational power. In other words, ChatGPT has NO IDEA how it solved the problem. Moreover, it's nowhere near as efficient as the human brain even for simple math problems. And it's all like that.

But that doesn't mean AI can't be used to accomplish many useful things, like generating patches.

I'm not going to argue the merits of using an LLM to create new software synthesizer presets for you. But your overly rose-tinted views on the future of AI are very much suitable for your username and are counter to what designers/stakeholders themselves have recently said about our extremely near future trying to coexist with AI. Yes, it is very, very much doom and gloom.

[Hyperbole]

I'm not going to argue the merits of using an LLM to create new software synthesizer presets for you. But your overly rose-tinted views on the future of AI are very much suitable for your username and are counter to what designers/stakeholders themselves have recently said about our extremely near future trying to coexist with AI. Yes, it is very, very much doom and gloom.

LOL. What did I say about “AI-hating trolls” coming out of the woodwork? On what planet did I present “rose-tinted views” about AI? All I said is that AI capabilities will continue to improve and that it will be useful for some things but is unlikely to become a God-like super-intelligence, at least not in the near-term.

You’re the one being absurdly hyperbolic by saying it will all be doom and gloom. The FACT is that no one knows how this will play out.

[SHall1000]

The FACT is that no one knows how this will play out.

I do. I asked ChatGPT.

[IvyBirds]

I'm not going to argue the merits of using an LLM to create new software synthesizer presets for you. But your overly rose-tinted views on the future of AI are very much suitable for your username and are counter to what designers/stakeholders themselves have recently said about our extremely near future trying to coexist with AI. Yes, it is very, very much doom and gloom.

Humanity has for eons been hell bent on self destruction and has had a wide range of social issues and challenges
.
Even without AI or even powerful computers Humans developed the Atom Bomb and had the ability to cause our own extinction in less than a hour by the choice of a single person or two

AI offers the potential threat of new social issues and risks, while also offering the potential to solve many of our existing issues and risks

All of that is beyond my control in anyway, and probably yours as well

Until we are all slaves to our AI overlords, or a madman pushes a button and drops the nukes, I will continue to live my life and make music. Using AI and whatever other tools I can get my hands on

[StartMenu]

I've had ChatGPT give me formulae for use in the Serum wavetable editor based on my description of timbre. It took several iterations before the formula was usable and somewhat close to my description but it worked. Hadn't thought it could make me a full preset though...

How hard is it to use ChatGPT for any software synth you want? Can you give me a general review of what you would have to do to have ChatGPT give tips on making for instance a specific artists Just Intonation scale in ODDSound-ESP from their album or mimicking things they do to do your own take on it? I don't intend to distribute it I'm just curious.

[Hyperbole]

The FACT is that no one knows how this will play out.

I do. I asked ChatGPT.

I suppose this is why Gearspace moderators immediately delete any posts related to AI: Such posts are guaranteed to send the thread completely off the rails almost immediately. Thread participants always end up discussing the merits of UBI, which is everyone's favorite subject.

At the risk of sending us down the UBI path, it's interesting to ask ChatGPT what it "thinks" are the difficulties AI researchers face developing a self-aware AGI.

ChatGPT will discuss the inherent limitations of Large Language Models (LLMs), emphasizing that these systems currently lack the depth needed for true understanding. It will explain that for AI to genuinely comprehend information, it must first construct a detailed model of the world. ChatGPT will say that this model is crucial for evaluating the trustworthiness of information and making authoritative statements. Additionally, ChatGPT will explain how researchers need to first develop a science of consciousness to determine whether AI systems possess self-awareness. ChatGPT will conclude by saying that the work to build a self-aware AGI is monumental, and simply scaling up existing LLMs is unlikely to succeed. Predictions that AGI could be achieved within one to three years are considered highly optimistic and unrealistic.

So even ChatGPT calls bullshit on most of the speculative claims about AI, or at least near-term AGI.

[StartMenu]

Aside from my other question can someone tell me if AI Assistance is scheduled to help with modular patches and things like artist copying for private listening? In theory artists don't want their work mimicked. But... I sure don't know how to do it on my own.

[IvyBirds]

Aside from my other question can someone tell me if AI Assistance is scheduled to help with modular patches and things like artist copying for private listening? In theory artists don't want their work mimicked. But... I sure don't know how to do it on my own.

I would post this question in the dedicated thread for such questions I started

[StartMenu]

Aside from my other question can someone tell me if AI Assistance is scheduled to help with modular patches and things like artist copying for private listening? In theory artists don't want their work mimicked. But... I sure don't know how to do it on my own.

I would post this question in the dedicated thread for such questions I started

Ok I will.

[martinjuenke]

Humanity has for eons been hell bent on self destruction

Obviously a contradiction in itself.

[swilow11]

I've had ChatGPT give me formulae for use in the Serum wavetable editor based on my description of timbre. It took several iterations before the formula was usable and somewhat close to my description but it worked. Hadn't thought it could make me a full preset though...

How hard is it to use ChatGPT for any software synth you want? Can you give me a general review of what you would have to do to have ChatGPT give tips on making for instance a specific artists Just Intonation scale in ODDSound-ESP from their album or mimicking things they do to do your own take on it? I don't intend to distribute it I'm just curious.

Not sure about its abilities with "any software synth" but seems like well known products are part of ChatGpts training data. My advice would be to login and start prompting.

It still hallucinates a bit though. I was asking for a patch to use on my Argon8 and it started referencing LFO 3 but the Argon8 only had 2 lfos. It also seems to not entirely know the difference bwyeen Serum the synth vs Serum FX and uses the terms interchangeably sometimes.

I'm not at all excited about the integration of AI into the creative industries but there is some creative fun to be had. I requested ChatGPT provide a bunch of chords to create 'wistful nostalgic hopeful and sad' melody and it went ahead and offered up a nice little starting point for me, plus a handy little description of how the chords in question were suitable for my request. And yep, it was definitely 'wistful nostalgic hopefuly sad' but also predictable and kinda boring. But a fun exercise nonetheless. Worth diving in and seeing for yourself.

[StartMenu]

I've had ChatGPT give me formulae for use in the Serum wavetable editor based on my description of timbre. It took several iterations before the formula was usable and somewhat close to my description but it worked. Hadn't thought it could make me a full preset though...

How hard is it to use ChatGPT for any software synth you want? Can you give me a general review of what you would have to do to have ChatGPT give tips on making for instance a specific artists Just Intonation scale in ODDSound-ESP from their album or mimicking things they do to do your own take on it? I don't intend to distribute it I'm just curious.

Not sure about its abilities with "any software synth" but seems like well known products are part of ChatGpts training data. My advice would be to login and start prompting.

It still hallucinates a bit though. I was asking for a patch to use on my Argon8 and it started referencing LFO 3 but the Argon8 only had 2 lfos. It also seems to not entirely know the difference bwyeen Serum the synth vs Serum FX and uses the terms interchangeably sometimes.

I'm not at all excited about the integration of AI into the creative industries but there is some creative fun to be had. I requested ChatGPT provide a bunch of chords to create 'wistful nostalgic hopeful and sad' melody and it went ahead and offered up a nice little starting point for me, plus a handy little description of how the chords in question were suitable for my request. And yep, it was definitely 'wistful nostalgic hopefuly sad' but also predictable and kinda boring. But a fun exercise nonetheless. Worth diving in and seeing for yourself.

Probably figure you could even do song title suggestions and things like fresh idea suggestions. TBH I don't love ChatGPT but it could be one of the next best things in some minds. I would love to see how far it can go in the arena of patch generation instead of patch parameter suggestions.

[db3]

The OP at Audiosex has now removed their first post. Although, later in the thread they posted this image which also included Pure LoFi and SEM. I'm 50:50 on this being fake.

[tappert]

removing a post is usually a sign that it's legit.
No updated ARP2600 would be a no go for me