Thank you for the support. The hope isn't to put people out of work. The hope is to provide tools to empower creation.

Art has always followed a similar formula. Great concepts followed by refined execution. The tools that are used to complete that path dont always have to be a struggle. If I want a guitar part on a song im working on. I could spend the years time learning how to play guitar. Or I reach for a sample because im not fluent in guitar and I want to get that idea down in that moment of inspiration. I could hire a guitar player or i could reach for a kontakt patch that hits my idea. The intent isnt to make one less guitar player out of money.

Lets approach the perspective of the guitar player. They might only be proficient in playing guitar. Their 10,000 hours was put toward mastering their craft. They feel the daunting side of recording, mixing, mastering, and need a backing track for a melody they wrote on guitar? They can use their guitar riff and explore the genre possibilities by putting that riff into a AI platform and get a whole backing track that fits vibe of the idea they have in their min This could take them from just being a guitar player to creating a tangible song.

Ive seen this with many things in this new tech era. Children's books being created, business plans being written, Plugins being coded, even people picking up instruments now that the internal dialogue of "Im not a musician" is being silenced.

d.

 

I agree Human minds aren't binary. Neither is this subject. Have you applied your human mind and cultural relevance with any of these new tools? Not just "make me a song about my dog" type of use. Like work with the tool to hit the target of a creative idea you have? Its kind of mind blowing how you can take an idea and evolve it into reality.

If anything its a fun thing to try. Upload one of your songs, could be a work in progress song and try to imagine it in any genre. Even humming, beatboxing, or singing an idea into suno covers is pretty mind blowing the results you get.

None of these have to be the final product but they can be a voice note on steroids of your idea. And thats just one of the many use cases.

Yes there will be countless numbers of people creating AI slop music. As a musician you have the knowledge to use a tool like this in a way that could make the music you have always wanted to make.

Simple example to spark creativity:
https://www.youtube.com/shorts/ftP9eB_do_4

 

My memories of those periods are quite different. Internet radio was huge for sports. No one cared about 128k when it brought a new form of access. 320kbps was the norm within 2 years of audio streaming (and still is for many uses), preceding the ipod which was also wildly popular from launch and selling millions of units annually within 2 years. I remember spotify taking over as many peoples preference from pandora. I don't recall much flack other than people wanted explicit control of the playlist, which is what spotify gave them. Napster happened before all of this, birth of internet radio aside.

My feelings on "ai" music have nothing to do with the technology. Calling a llm ai is another matter, but .. focus danmit!

The first response mostly focused on intellectual theft and the enrichment of the few on the backs of the many. I have no problem with the technology. I use Dreamtronics Synthesizer-V. Dreamtronics pays their data points (ie the vocal performers). The for profit company pays for the fruits of their labor. Everyone happy.

The majority of the incorporated for profit media llms the public can name do not. Using 'Fair Use' claim as a shield is just legal stalling while they get 'too big to fail.' A key part of Fair Use, especially weighed in commercial ventures, is the enrichment to the source. A movie review that quotes a couple lines of the script is Fair Use. Movie gets promotion that doesn't use the very movie itself as the vehicle driving traffic to the reviewing. Reviewer makes profit via paycheck. Taking whole songs as data points with no acknowledgement, payment, or credit entirely fails one of the essential checks. "But Transformative....!!"....still requires passing the above check. LLMS are taking the whole movie. Arent they? Well, see, they don't actually know. Because while creating 'ai' they are also incapable of ...lemme check my notes again... ah, yes... spinning up a database to store network history to see just were any particular defined data came from. Seems like a pretty convenient way to 'not know' where your data comes from....

I have zero doubt it will be popular with a segment of people and those people will find enjoyment in engaging with it. Other than being concerned about a sudden rise in deaf people and its societal causes, their enjoyment doesn't impact me... let the people have their fun. Not opposed to that either.

I'm not particular about how one constructs music. Use whatever tool you want. I've also never seen someone play guitar hero and then transfer that skill to an instrument or notation or recording or a method of creation that conveys a thought of any sort. Doesn't take aware their bitchin gh flair. Doesn't mean they can write a melody either.

As to it the music being godawful... Lots of really well produced music by humans is godawful too! I have no doubt the sonic qualities will continue to improve. They're already pretty good to excellent in many areas. Wasn't my sticking point (but I didn't make that clear either and it wasn't the obvious take). I have never used or heard a generative tool and thought... wow, this sounds fantastic! From the most basic midi random or euclidean generators to the "fancy smancy dressed up TI calculators hooked into a database of oppression, woo baby lets milk vc and call this shit ai" it all comes off as drivel. I've certainly heard generators (of any flavor or tech level) put out a riff that catches my ear. I've never heard one take that data and build it into something that didn't immediately kill my interest in what it was doing.

When a generator can actually do that the force behind it will either be that a single llm is consuming all the worlds power supply and tears so it can finally run enough permutations to finally write something that doesn't sound like dogshit coffehouse pop or a machine will have become greater than its sum and the era of AI has started....for realsies this time. I have no opinion on an actual AI creating music. I'd be too busy both marveling and shitting my pants.

To sum....Technology moving forward - yay!(mostly). The people building explicitly for profit companies on theft of labor - boo!(guillotine). Generative music in a wholly generative state, regardless of technology used or its source material if any, - generally dogshit!(have no aside for this one, but you know, continuity). The end consumer using generators - Whatever floats your boat(so excited for the new 7 hour ambient jam drop!) Generative tools not built off stolen labor and tasked as tools that serve the composer/composition - Sure, why not (typing an idea in a text box and getting a result is fine, just different interfacing - if that result is a whole 'song' - something was created sure, but I will have zero doubts it sounds godawful).

 

I really think you should play around with suno more and get a better understanding of it and its power as a creator. I think that will change a lot of this opinion. Especially with the new model and features that dropped a few weeks ago. When you first use or experiment with any new tool you cant make excatly what is in your mind until you get a better grasp on the tool.

You put in your music, or musical idea and iterate on it and you will see the views on sound quality, performance, it "STEALING" arent represented in the current perspective on it that you have.

 

Real good watch for all of us in this thread. I think they all represent all of our perspectives!

 

These models don't (well, shouldn't) spit out discernible prior performances of the training data. That isn't the theft I've been fairly specific on. I know very well how this technology does its mojo under the hood. They cannot function without this data. You cannot have a full handle on how this technology works and also claim no theft is occurring, with explicit reference to SUNO, when we know SUNO trained on protected works. I iterated before, I am not a ludite. I just told you I use a product whose core functionality depends on similar training. I like tech. I fuck with it. Tech fucks with me. We fuck each other (ok, I'll stop). My problem isn't and never has been with the tech. Its with the cocksuckers about an hour south of San Fransisco. That the tech also produces godawful music when left to its own accord is just incidental and fun to say. Sonically its some good sounding godawful shit, compositionally its still just godawful generative shit.

 

LAWSUIT AGAINST SUNO

Suno does not disclose the dataset used to train its artificial intelligence. On January 21, 2025, GEMA filed a lawsuit against Suno, alleging that well-known works were repeatedly used in the training, but without paying the artists.

 

The American groups started last summer. In the earlier one they already positioned themselves for an affirmative Fair Use (american law thing) defense and stated they A- use any publicly accessible source of good quality B- *if* I did this thing I totally did not do... then I'm sowwwry... and btw since I didnt do this thing have you heard of Fair Use, which doesn't apply whatsoever if I didn't do this thing that I didnt do... but like, you've heard about it right guys? Cause this thing I didnt do would totally be that, but since I didnt do that it cant be that cause it never happened, am i riight fellas?? The legalese for lets drag it out 3 years then settle.

Doenst really matter anyway. They'll settle at some point. Pay a few million to the labels who in turn won't give anything to the artists and the cycle will continue.

Don't know about the GEMA one, but I assume asserting a defense based on American law won't go that well in Germany. I'd also guess though that Germany has a version of the law.

 

This is a very dishonest take on AI. You are lying to everyone, including yourself. You are trying to frame it like AI is just a chisel that you use to carve out your idea in the block of marble, but in reality it's more like outsourcing the entire process of creation to a Chinese sweatshop. There's nothing noble about that.

You are also omitting the most important aspect of the problem: nothing is actually "created" by AI, it just regurgitates for you an amalgamation of stolen ideas. The ideas that people embracing AI eagerly claim as their own to feed their egos, but in reality they don't care enough about music to put in an actual effort.

Using AI and calling yourself an artist is equivalent of injecting yourself with synthol and calling yourself a bodybuilder. Both show weak character.

 

Some synths produce horrible stuff too. I don’t complain about them I just don’t use them.

Suno just like all ai tools are trained on open internet data. Unless you really take your time to focus on one thing in the data set you might get a close replica of it.

you can also get some stuff that has never been heard before.

This can also go for gpt etc. I can have it spot back something that has been written before with the right prompting. Or I can use it to write something entirely new with my creative guidance.

 

yep it’s over a trillion dollar law suit. lots of headway daily on all aspects of it. Protecting labels who on the lions share of music that makes money.

“New minute entry in UMG Recordings v. Suno (Music labels sue Suno over training and outputs of music AI): Order on Motion for Leave to Appear”

Link with more info. That bot has some good info:

https://x.com/ai_cases_bot/status/1909945215944974606?s=46

 

I fully agree the labels are forward facing about artist’s rights. But will take all the money when a settlement is agreed upon.

the other race that’s happening apart for legal. Is the tech is getting so much better that. It needs less data to understand how music and genre are made.

Soon a small open source data set can train a great sounding music model.

we saw this when deep seek was able to make a reasoning model on smaller data and compute. These innovations will translate to all machine learning tech.

another thing we’re seeing in the ai video space is. Synthetic data using unreal engine. They can just generate the data they want to train on.

I think this will happen in music. They will be able to just generate the music for a model and train on that no artists data needed.

 

Ive said on this tread and others multiple times. People think using AI is just input output. "make me a jazz song about the new pope" and they call it their new song. Yes some people use it that way. others dont.

Yes it is a chisel for me. Expanding on your body building analogy. For me Its more like pre-workout. I still go and work out. I still put the reps in. I pick up the weights. I might hum a melody into it and get a string line. or i might sing into it and get a orchestra. I take those stems and elements into my DAW add elements from my synth VSTs my Drums and all the other tools ive been using for the past 30 years to make music.

I can now have any element of the musical landscape at my disposal. I use it like a futuristic kontakt library, or even a highly evolved vococoder. Its a faster path with less resistance from mind to music.

I put this link in a earlier post but it shows one of the many ways someone can input their creative ideas in as audio and get elements for their song CHECK IT OUT: - https://www.youtube.com/shorts/TTjfPLuVMec

 

Maybe an actual audio example from me can help you better understand the "chisel" use.

I needed a flamenco spanish guitar for this song im working on. I had an idea in my mind what it should sound like.

So i made a quick voice note HERE:
https://voca.ro/1m2ICSbr0pRt

prompted "flamenco spanish guitar" HERE:
https://voca.ro/1oRqqKpgKMci

for that small loop Did I steal from someone elses work? Did i make someone loose their job? Maybe? But I made headway on my song in a way that fast and efficient for my workflow.

 

Another different example of how any of us producers can use these tools:

https://www.instagram.com/devanibiza/reel/DJZwuNuu-AW/

 

ran my same voice note I did from above through a “black metal guitar” prompt for fun and got this:

https://voca.ro/1h0LhS8vOkKU

 

Last edited: May 15, 2025 at 6:45 PM

Can I pee on it? I'm really into trying piss fetish lately

 

Compared to other 'AI' slop ou there? Yes. Compared to previous Igorrr music videos? No, I don't think so, looks like a downgrade.

 

Might ruin your computer but give it a shot

 

They just typed a prompt, dude. Just like that...

All you need is MeatJourney AI Music Video Generator