Skip to main content


RE: mastodon.social/@pkirn/1164770…

If I could port plugins to Linux, I would reach out to the person behind this. Looks awesome!

#LinuxAudio

in reply to Morten Mosgaard

I was interested at first, but on KVRAudio there’s a long thread about how these plugins were made. Some describe them as “Vibe Coded slop.” I was especially excited about the Ensoniq DP/4 recreation:
Temecula DSP DEEP/4 Plugin
Someone dared me to say that in the thread. After looking, the Vibe Code mob was out in force—torches and pitchforks included—so I backed off.
Maybe they’re good, maybe not; I don’t know. I just wanted you aware that some people are upset about these products.
in reply to Ercanbrack

@ercanbrack there is a special to these vibe-coded plugins, and you can kind of see it on the plugin description alone. The way it's setup, information is organised and so on. So I had my thougts.

I did install the Windows version through Yabridge, and it sounds fine, I especially like having the chorus'.

We see a lot of these products these days.

in reply to Morten Mosgaard

Yeah. I think we are going to see a lot more of these "Vibe Code" programs. The only real difference between "Vibe Coder" and "AI Assisted Developer" is the person doing the prompt engineering. If that person knows what they are doing, AI can be successful. Otherwise, one must rely on reviews from those using the programs. I'll still be watching and listening to everyone (including you) who tries them. The fact that you tried them and they worked is a good start.
in reply to Ercanbrack

@ercanbrack I am a professional software developer with 30+ years of experience designing all kinds of brand new software products across many domains. I use AI agents in my coding work. I believe the pitchforks crowd are doing themselves a significant disservice.

I am making some of the highest quality software I've ever made in my career with these tools. It very much comes down to who is steering. They truly are just amplifiers of intent and vision.

in reply to Michael9

@fortifieduniverse
I don’t disagree at all—you’re right, it all depends on who’s steering.
Vibe coding is a non‑programmer iterating prompts to make AI write code; it may work, but often turns into slop.
An AI‑assisted developer knows the craft, the goal, how to ask, and how to verify—just faster.
I trust AI‑assisted developers like you.

The hard part is knowing what the person doing the steering knows or doesn't know, and what was done to vet the code. 🙂

in reply to Ercanbrack

@ercanbrack Sure... We're saying the same thing.

On the flip side... just because something was built by a human in no way guarantees it's not a piece of smoldering hot garbage.

I just find the pitchfork crowd to be a little non-thinking. AI-assisted coding is a massive benefit to making music (and a great number of other pursuits) on Linux.

I am not a fan of Big Tech in any way. And there are complications with these tools. But ultimately I believe they will be a great liberator.

in reply to Michael9

@fortifieduniverse
I truly believe that it is largely misunderstood for most people. One solution is to help them to understand the tool better.

There are groups that are currently designing "AI Free" logos.

mastodon.online/@mastodonmigra…

I've been pushing the difference between a "Vibe Coder" and an "AI Assisted Developer" with everyone I speak with.


Fedi Hive Mind - AI Free Label

OK, we have a winner. For the last few days we've been discussing a label for content and code attested to be free of generative AI (see QP in second reply). Yesterday's poll showed a clear preference for the simplest of the four options.

Seems like what we need next is someone with actual graphics ability to take this rough concept and create usable graphics. If you are interested please read on to the first reply.

#AIFree #AI #AIFreeLabel


in reply to Ercanbrack

@ercanbrack @fortifieduniverse
Replied to this already here:

mastodon.online/@mastodonmigra…


@ercanbrack

While many have opinions on these matters, the purpose of this initiative is not to take a position, or 'ban AI'. Rather it is for those who what to attest there is no generative AI contribution to what they have produced. Simple as that.


in reply to Ercanbrack

@ercanbrack Yep. I'm not really interested in engaging with those groups.

Just like most modern causes... there's not really any interest in nuance or dialogue... you're either in their camp ideologically or you're the enemy.

Not down for any of that.

in reply to Michael9

@fortifieduniverse
I responded:

You’re right—but we’re addressing different layers of the same problem. I’m talking about how AI can improve systems; you’re talking about who gets to decide. “AI‑free” misses the real issue: AI amplifies existing power. The solution isn’t abstinence, but governance—set limits on use (constraints), tie AI to quality not savings (guarantees), protect creative labor (empowerment), and require disclosure and accountability (transparency).

in reply to Michael9

@fortifieduniverse
Agreed. I responded:

You’re right—but we’re addressing different layers of the same problem. I’m talking about how AI can improve systems; you’re talking about who gets to decide. “AI‑free” misses the real issue: AI amplifies existing power. The solution isn’t abstinence, but governance—set limits on use (constraints), tie AI to quality not savings (guarantees), protect creative labor (empowerment), and require disclosure and accountability (transparency).

in reply to Ercanbrack

@ercanbrack Indeed.

Not quite how I would pitch it... Rather than crying about AI... use it to amplify your own vision and your own skills in the marketplace of ideas. These tools can be empowering for everyone... we should be spending energy trying to figure out how to democratize them rather than trying to eliminate them.

The toothpaste isn't going back in the tube.

in reply to Michael9

@fortifieduniverse @ercanbrack I agree with a lot of the good perspectives you come with, and really hope there will be a big strong middle ground soon, who will help democratize the field and not least, bring attention to all the valid ethical concerns about energy consumption, transparency about training data, fair use of others intellectual property etc…

I’m starting to see, that assisted coding already gives a big boost to Linux audio, lowering the effort it takes to port plugins.

in reply to Morten Mosgaard

@fortifieduniverse @ercanbrack I’m in no way blind for the value this can bring for Linux audio specifically, because people with actual skills, like you Michael, get’s to add to the platform, what otherwise would have been too time consuming in a way that made it less realistic to get done in one’s spare time (like the Arthur project).

I do not have a “no assisted” plugin policy, but I tend to move on when a plugin developer is producing too fast and has a completely AI generated website.

This entry was edited (3 weeks ago)
in reply to Morten Mosgaard

@fortifieduniverse @ercanbrack I mainly move on in those situations, since I have no way to check myself, if there is actually knowledge of plugin development behind the profile.

I really like when people are transparent about who they are, what their experience is and if what they make are AI assisted. I get why people don’t already write this, because of some of the reactions, but I really think transparency is key here as part of democratization.

in reply to Morten Mosgaard

@ercanbrack I understand your argument, but I'm not sure how much I agree. I personally have some major work coming out this year that involves AI, and part of the nature of the work requires extreme transparency about it. So, I'm fine with that personally.

But I don't know that I agree with the sentiment that anyone who uses AI should disclose that... should I disclose what text editor I use? Operating system? Judge the work based on the work...

in reply to Michael9

@fortifieduniverse @ercanbrack the problem is: you can't judge the work based on the work, since you don't know how something is made work. And the user of the LLM, can't guarantee fair use of others work, if they don't know what's in there.

Right now the big models don't give any guaranties, if the training material is legally acquired, and they have been caught using pirated materials before.

Transparency helps us understand what LLMs can do, and demand a more ethical marked for models.

in reply to Morten Mosgaard

@ercanbrack You can't judge audio plugins coded by hand any other way than by the quality of the plugin, either! You can't judge whether or not someone cloned another plugin by hand or by ear... it's no different.

And even if someone tells you something is "made with AI" that's nearly meaningless... I personally use MANY different AI tools for many different aspects of my work.

in reply to Michael9

@fortifieduniverse
I agree. You can have bad or great developers with or without AI. AI assistance can produce excellent code—or total slop—just like humans. In the end, the only reliable signal is feedback from people who’ve actually tried it, before you test it yourself.
in reply to Ercanbrack

@ercanbrack And I'm not saying there aren't ethics issues with Big Tech companies... that's true across the board, not just with AI companies... we should aggressively work to use these tools to figure out how to eliminate the unethical companies...

But that's a completely separate concern.