AI disqualifies anyone as a musician! It's like playback.

Explore how Machine Learning and AI can expand musical creativity while keeping the human in the creative workflow. This forum is dedicated to respectful dialogue where diverse perspectives are welcomed.
Post Reply New Topic
RELATED
PRODUCTS

Post

ksandvik wrote: Sat May 02, 2026 12:23 am Wouldn't it be faster to sit down by a keyboard or guitar and just write music than write endless prompts? Sounds boring for me.
Y not both? ;)

It is not either or.
“The biggest crime of a musician is to play notes instead of making music.”
Isaac Stern

Post

Some of the reactions here imply that people criticizing AI have not used it for more than 5 minutes, if at all. The black and white discussion here is: people without any artistic vision prompt AI-composed songs with a one-liner, and the results are robotic songs without soul and tons of artifacts. This might habe been the situation two or three years ago, but the current state of development is quite different. Just a couple remarks:

1) Composition: You *can* feed your own songs to AI, so the basis of everything generated is your own composition. In that sense, it's much more like the "accompaniment" tool that was mentioned earlier here, or a cover band performing your song (or, depending on how close you keep it to your input, just a different singer, or a different take of the very same performers). Suno allows this, and you can generate even without any prompting.

2) Control: If you know what you are doing, you can control the generation - for example in Suno - on a granular level, down to individual elements. You can control style elements, the amount and type of musical elements added by the AI (or not), and the audio influence. With inline prompts added to your lyrics, you can directly control individual segment generation, even down to bars, and what's added or deleted. If your audio is the basis, the AI actually can be controlled by the human user and does add only what the user intends to add. So you can double voices with a background choir, add or delete bass, drums, whatever. This is not "AI controlled", but human choice then. And this can be a lot of work, sweat and tears, a tedious process - that some seem to equate with music production or 'emotions' in music here.

3) Artifacts: Earlier versions had a lot of artifacts, like a prominent hiss, but this has been reduced greatly. Also, for elements where the AI has a lot of training material, the artifacts are minimal. For example, vocals have improved immensely (as there are probably a lot of soloed voices in the training material, or segments with soloed voices). Some other elements are still lacking, as there is maybe not enough 'pure' training material (yet), so the generation seems to be based on instruments in context, but not individual tracks. The improvements here are just a matter of time, and something that we will see in future versions for sure - artifacts will disappear.

4) Stemming: Many artifacts mentioned here to be due to "AI" being "bad" are based on the stemming used when remixing generated songs (as programs like Suno generate 'complete', mixed songs). For example, Suno Studio seems to apply a post-fact stemming to mixed material, and the resulting stems do have similar artifacts to all the stemmers on the market (i.e. residuals from other tracks, weird wobbly artifacts, degrading audio quality). It does not re-generate the individual tracks (yet). If I was Suno - who has a contract with Universal now - I'd train my model on existing stems in Universial's archives, in conjunction with mixed songs, to improve per-track quality and generation. You could come up with killer stemming that way, as the AI based stemming would be vastly superior to traditional forms of stemming - it would be able to regenerate tracks without artifacts then. Also, this could be used for audio-restoration: You could bring back old recordings as re-generations with perfect audio quality.

5) Authenticity: It is implied here that any music done by AI has no soul, as it's done by a 'machine'. This disregards the functioning of most systems out there, which are based on large training data sets of *human* made music. The AI therefore replicates human patterns, it does not apply 'cold artificial logic' - in that sense, AI systems are more 'human' or ''emotional' than pattern generators that apply mathematical logic (arpeggiators, jammers etc.). You may argue about AI producing 'more of the same' or 'the reign of the mass' - that's fine (I would disagree, though) - but it has nothing to do with 'no soul'. The 'intent' argument above implies that no human involvement steers the generation of the AI (i.e. simple prompt, no artistic vision at all) - but an AI production can have as much intent as any composition, if the the human 'composer' steers the AI as an 'orchestra' or 'studio musician', feeds his/her own audio to the AI, does inline prompting, post-production etc.

6) Specialized tools: Some seem to equate Suno or similar systems with AI. First of all, as discussed above, the current version of Suno allows for a lot of granular control over generation, and therefore develops more and more towards a studio tool (if you want to use it that way). And then there are specialized tools for specific tasks, like Ace Studio or Synth V, or AI mixing tools etc. I really don't see how to make a clear cut differentiation of ethical vs. unethical AI use if everything is lumped together as 'AI is bad'. If you use Suno as a cover band or studio vocalist by feeding your own audio into it, you can still be the composer - but you can also be the lazy person that prompts a cheap Taylor Swift clone with a one liner prompt. It depends on the use in a specific stage of the production process.

So in short: If we continue to talk about "AI" as a catch-it-all term for anything being remotely conntected to some form of ML being applied, or AI being equal to prompting a one-liner in Suno, then this becomes a pretty futile discussion. Like all the discussions that complain about any tech development being evil, as it's a disturbance of the status quo.

I actually - surprisingly! - agree with Bones on many things here. At least, he's speaking from experience, whereas a lot of the arguments in this thread are based on prejudice, or 5 minutes experience with AI tools in an early stage of development.

Post

BONES wrote: Sat May 02, 2026 7:45 am
Hipster Bales wrote: Thu Apr 30, 2026 12:22 amI completely agree with this! There are people who are willing to spend a month on 1 project (like me 99% of the time) and then there are people who just want to generate slop for a quick buck.
Clearly you don't know what you're talking about, then. Our current album, all based on music generated by my bandmate (one by me) using AI is going to take us longer to finish than our last album, which we did the old fashioned way. Sure, some people just want to "generate slop for a quick buck" but that's not the only use-case for AI. Like most things, you get out of it what you are willing to put in.
Yes I know that. I was including 1 use case. There are AI assistants to help people, which means there’s still humans massively involved. I’m talking about the Suno/Udio grinders who only care about $$$$

Post

Hipster Bales wrote: Sat May 02, 2026 8:55 am Yes I know that. I was including 1 use case. There are AI assistants to help people, which means there’s still humans massively involved. I’m talking about the Suno/Udio grinders who only care about $$$$
:tu:

Fun fact: A band project I am involved in released a song lately, and we were accused by some guy of being "totally fake" as we used an AI video - and the person said he immediately heard it was 'fully prompted' as it sounded too perfect. Funny enough, I wrote the song three decades ago, and the first version was still recorded on a 4-track-tape. :hihi:

Also lately, I made stripped down version of a fully mixed rock/metal song with Suno, as I wanted to hear how it sounds with just guitar and a female singer. And then people said how emotional and intimate and human it sounds. :o

In both cases, the composition was fully human, though. Go figure! :party:

Post

AI is the new Shadow producer so there is nothing new except that it is your computer doing it instead of using another person to make your ideas into a full song.

Post

D-Fusion wrote: Sat May 02, 2026 9:39 am AI is the new Shadow producer so there is nothing new except that it is your computer doing it instead of using another person to make your ideas into a full song.
Rally car navigator says: "AI is the new driver so there is nothing new except that it is a computer driving it instead of using another person to make it to the finish line.
Intel Core i7 8700K, 16gb, Windows 10 Pro, Focusrite Scarlet 6i6

Post

Was just thinking about the subject:

"AI disqualifies anyone as a musician! It's like playback."

Playback disqualifies anyone as a musician?
That would cover backing tracks, wouldnt it. So bands like Motley Crue, ZZTop, U2, ELO, Pink Floyd and a gazillion others?

Hmmm.

So, what about bands with drum machines or eg Live pusers playing back via loops or MIDI clips? Disqualified too?
Set Theory claim:
"In some cases there is an object called red that contains everything that is red. In much the same way a pot is a plate.
Red is Red and anything that is Red is an object, a class in itself or a real thing if you prefer"

Post

morelia wrote: Sat May 02, 2026 9:52 am
D-Fusion wrote: Sat May 02, 2026 9:39 am AI is the new Shadow producer so there is nothing new except that it is your computer doing it instead of using another person to make your ideas into a full song.
Rally car navigator says: "AI is the new driver so there is nothing new except that it is a computer driving it instead of using another person to make it to the finish line.
Ok. :clown:

Post

D-Fusion wrote: Sat May 02, 2026 9:39 am AI is the new Shadow producer so there is nothing new except that it is your computer doing it instead of using another person to make your ideas into a full song.
Yes, that's also why some people in the industry are so worried and run amok - it will undermine their business model, rather than music production per se.

Post

tq wrote: Sat May 02, 2026 10:22 am
D-Fusion wrote: Sat May 02, 2026 9:39 am AI is the new Shadow producer so there is nothing new except that it is your computer doing it instead of using another person to make your ideas into a full song.
Yes, that's also why some people in the industry are so worried and run amok - it will undermine their business model, rather than music production per se.
That is true and i do understand their worries and frustration.
We all need money to eat and have a good life in this modern world.

Post

Yes, but you cannot fight this. It's like howling at the moon. Change will come, and will affect you... better prepare and adapt, then.

Post

tq wrote: Sat May 02, 2026 1:42 pm Yes, but you cannot fight this. It's like howling at the moon. Change will come, and will affect you... better prepare and adapt, then.
I agree.
Ai is the future and soon we will all be replaced by Ai and robots.
So it is here to stay no matter if we like it or not.

The bigger question is what the big corporations will do when they don't have any use for us anymore.

Post

We all sit on a fake beach and watch an artificial sunset, while in reality, we are inside a tank and serve as a battery. :D

Post

tq wrote: Sat May 02, 2026 2:13 pm We all sit on a fake beach and watch an artificial sunset, while in reality, we are inside a tank and serve as a battery. :D
Depends on the colour of the pill you chose, assuming you ever had the opportunity to choose, of course :)
Mac Mini M4 Pro | 14 Cores (10P/4E) | 48GB RAM | Studio One | Reason | Bitwig Studio | Logic Pro | FL Studio | Cubase Pro | Waveform | Reaper | Renoise | ~1000 VSTs/AUs | ~350 REs

Post

tq wrote: Sat May 02, 2026 2:13 pm We all sit on a fake beach and watch an artificial sunset, while in reality, we are inside a tank and serve as a battery. :D
:hihi: :tu:

Post Reply

Return to “Machine Learning and AI for Music Creation”