AI and music technology

[host body]

Now with the AI revolution in full swing we're seeing more and more plugins and even hardware that are at least partially run with ai algorithms. For example, there's a pedal where you can write (or say) prompts and the pedal will generate an effect for you. I watched some demos and the FX were all pretty shit, but it's definitely a direction music tech is going in. Similarily you have a lot of free plugins for DAW using AI, like for example you can transcribe audio into midi notes (presumably with things such as velocity and other parameters), something that was very rudimentary without AI.

I suspect in the future more and more gear will use some sort of AI "under the hood", for example a reverb plugin or pedal will use AI to generate the different types of reverb you get, but you won't really even notice that the tech is different from a traditional, pre-programmed reverb.

How do people here see this change? I'm personally very against using AI in any generative form, but a tool that would allow me to play a stringed instrument and transcribe it to midi notes, to run thru my synths... Well, it would save me 10 years of time, learning to play keys as well as I can play bass, not to mention stringed instruments allow some neat tricks that aren't really possible with a midi controller. It is tempting, and I'm sure I'll at least try it out.

[Leewar]

Im sure that much like the pearl clutching that went on when samplers became commonly used, people will over time begin to use it like much like any other tool. From then on it will get interesting as im sure it can be 'misused' IE Manipulated and fucked about with in the way that every other musical tool, or instrument can be to produce the kind of sounds that apply more to the kind of music this forum is all about.

Kind of like the leap from analogue 4 tracks to free home studios and plug ins on your PC. We misuse both to gain the results we need.

[Balor/SS1535]

The problem with AI is less what it does to the object of its application than its impact on the persistent user of it.  I teach college students, and the drop in ability (both intellectual and practical) has been massive on just a term-by-term scale.  There are interesting things being done with AI by people like Harmony Korine, Salitter Workings, and---dare I say it?---even Merzbow (when put in the broader context of how he has worked in the past).  Aside from them, I just encounter artistic garbage.

[Hyena]

I see it used in muzak a lot. In good noise, not so much. It's great for a Google on steroids and for amp sim captures. I don't see it making good art because it's not a sentient being.

[tiny_tove]

 AI expands possibilities on many aspects. Beeing heavily using generative software before Language models became accessible thanks to friends who are nerds. Crazy experiments for creating pattern that never repeats themselves, spund reactive visual textures and shapes.  AI and non-AI generatuve patches in programs like Max have been fantastic companions to build 100% original atmospheres that can be destroyed visa analog. Fluid Corpus Manipulation has been a good ally in that.

Don't find appealing working on noise with SUNO and similar. I am using it in my daily job to generate my compay's gingles or orchestral parts, a good alternative to stealing samples and having yuour own melodies replayed with different options. I love SUNO's shimmer reverb and the chaotic feature that can turn a sample into something out of this world, but I much prefer it for the work purposes than for noise fow which I prefer working with actual machines.

I am all about pushing ai-oriented visuals and hacking censorship behind the main models. It's like a new form of collage... Obviously to haved decent effects and avoid ai slope you must mix your own sources and find your own ways to break barriers, push to the limits, ecc.

Just remember to record and save everything :)