Menu
What's new
By:
Filters

Suni ai and Udio...the end of us?

P

pulpfiction

Active Member
Like many of you, I've recently discovered suno ai and udio and am a little worried about the future of our industry/area of interest?
Is there anyone out there who has a bit more insight and can perhaps allay some of our fears? :)

Questions that are on my mind at the moment:

If the internet is now flooded with countless AI songs, is there any way we can still gain a foothold in the music sector?

If countless songs are now being produced with the click of a mouse, isn't there a very good chance that the songs made by us without Ai will already exist in a similar form? (which could lead to legal disputes...)

Even if this is curbed a little with laws in the foreseeable future....for example...only things made by humans can be protected with copyrights.....how will you know the difference? It's already the case that every fart, at least in the top charts of the mainstream, is being sued for with lawyers....because even without AI many things sound too similar...


I'm currently working on a small project that I think is really not bad and has potential.... Out of curiosity, I tested what Suno ai or udio can do on this topic with a few commands....I was shocked.....t didn't convince me so much now, but who knows what it will look like in half a year....there were already a few moments in the Ai song that blew me away...

How do you deal with it?

How do you manage to continue pursuing your passion, even though in a few months it might all be unnecessary because AI will eventually overtake us in all areas?
Click to read more...
 
No offense to be made, but I think this topic already exists and is quite active:

Made with AI: Suno, Udio and others

I've spent the past 24 hours being quite astonished at Udio, which I already consider vastly superior to any AI music generator I have heard to date. To be clear: I share this for the purposes of awareness and conversation, not endorsement. The lyrics I gave it were random text relating to a...
vi-control.net vi-control.net
 
pulpfiction said:
Like many of you, I've recently discovered suno ai and udio and am a little worried about the future of our industry/area of interest?
Is there anyone out there who has a bit more insight and can perhaps allay some of our fears? :)

Questions that are on my mind at the moment:

If the internet is now flooded with countless AI songs, is there any way we can still gain a foothold in the music sector?

If countless songs are now being produced with the click of a mouse, isn't there a very good chance that the songs made by us without Ai will already exist in a similar form? (which could lead to legal disputes...)

Even if this is curbed a little with laws in the foreseeable future....for example...only things made by humans can be protected with copyrights.....how will you know the difference? It's already the case that every fart, at least in the top charts of the mainstream, is being sued for with lawyers....because even without AI many things sound too similar...


I'm currently working on a small project that I think is really not bad and has potential.... Out of curiosity, I tested what Suno ai or udio can do on this topic with a few commands....I was shocked.....t didn't convince me so much now, but who knows what it will look like in half a year....there were already a few moments in the Ai song that blew me away...

How do you deal with it?

How do you manage to continue pursuing your passion, even though in a few months it might all be unnecessary because AI will eventually overtake us in all areas?
Click to expand...
I don't know if this thread will be merged with another but I'll reply here.

I'm just a beginner/amateur with no real aspiration to earn money and A.I. has demotivated me a bit to learn and improve. I will not stop completely though and will still listen to my own stuff. I would not listen to pieces I generated with A.I., at least not in its current form because we don't have enough control on the result.

I will still make my own pieces the old way for now but hopes of sharing them eventually have diminished a lot.
I don't want to "compete" with those who'll claim to have "created" a great piece by just typing prompts. Yeah yeah your piece sounds better than mine but it is not really yours.

Also I would not want to share a piece I worked hours on and be asked "What prompt did you write ?" It's also this devaluation of human work that is demotivating.

One thing is sure though, I was buying way too many libraries and it has now stopped. I guess it's a good thing for my wallet :)

If I was a pro I would be worried for sure.
Especially if I specialized in underscoring.

The impact on amateurs is not to be underestimated though.
I'm pretty sure it will demotivate many to learn music theory.
In 100 years, how many humans will know music theory ?
Probably less than today.
 
Last edited:
CD Baby is the largest independent digital music distributor. With over 20 years in the industry, I have been her partner for some time and she does not allow songs created by AI:


"Artificial Intelligence

You will not be able to distribute A.I.-generated content. Even if the A.I. generator you are using permits commercial use of what you made, there's no way to ensure unique sounds or rights compliance."

If all distributors were following this rule, the AI will not prosper.
 
My take is, try not to let it get you down and put you off music. Sure it's a bit scary, with how rapid AI is progressing and all... But don't let some ones and zeros put you off your own creative journey to explore and create, learn / grow with other people.

Those who 100% rely on using AI to "create" something will have no idea how to piece something together for themselves, pretend they're "musicians" and never develop their own unique sound. Which is what makes music awesome, being unique. Sitting back and listening to your own work, after spending years learning, will be infinitely more satisfying than typing a prompt.

I'd say, currently the only markets that will be mostly affected are ones where any old-music is slapped on something. For example; low budget commercials (where stock music is usually heard) - things that require more thought and input - films, games, TV shows - maybe not so much.

I don't think library developers should worry either. Most of them will have a huge catalogue of session recordings, If I were them (and wanted to start using AI) I'd machine learn all those recordings and update sample engines to better utilise the sounds they have to make the experience more fluent and straight forward for those of use looking to use them to write our own music.

Theory helps, sure. But it's not a requirement. Do you need theory to hum a melody? Do you need theory to write a tune? Of course not. The more you learn and practice an instrument, be it guitar, vocals, whatever - the better you'll get at getting ideas out of your head into the real world.
 
Last edited:
pulpfiction said:
Even if this is curbed a little with laws in the foreseeable future....for example...only things made by humans can be protected with copyrights.....how will you know the difference?
Click to expand...

Since you wanted a little bit of hopeful news: this point is actually something were the interests of composers, musicians and listeners somewhat align with those of GenAI companies. They don't want their models to train on AI-generated output (garbage in = garbage out problem), so they are looking for ways to watermark AI-generated stuff.
 
Someone with an in-depth knowledge of music theory and editing of music is going to get far better results than an average user. These tools have alot of limitations, and for something like film scoring, I'm not going to say writing the music is the easy part, but so much of it is dealing with the director, really understanding the scenes and characters and the overall intent of the film. So its hard to see it making inroads into this in the immediate future. Also I feel most directors would be fundamentally opposed to AI music, given their industry is potentially next on the AI chopping block so to speak.

Game music on places like itch.io where people are posting free or games for a couple of bucks, definitely going to wipe out most of that starting point for video game composers imo. Library music also depending on the copyright issues is likely to become AI dominated.

Pop music will also be affected, with film music the structure is dictated by what is occurring on screen. Pop music is much more formulaic with an easily copiable structure by AI.

But I certainly wouldn't lose heart, no one can predict the future and AI is likely to run into serious limitations similar to self-driving cars to get the final 5-10% and if you play around with these models for a while, the limitations become more and more noticeable.
 
Trash Panda said:
Just out of curiosity, have you actually tried using it to create something that isn't pure novelty/comedy/meme music? That fear will wear off real quick. :laugh:
Click to expand...

Kind of. Yeah, the current output is still pretty bad but it is leagues ahead of last year's output. The (well-founded, imo) expectation is that these models will continue to improve in the next years.

The same thing is happening with video. That one Will Smith eating spaghetti video was March 2023. Now compare that with Sora and draw a line into the future.

Edit: I'm not saying it is the end of music or something, but it is worthwhile to keep these developments in mind when planning your career.
 
pulpfiction said:
If the internet is now flooded with countless AI songs, is there any way we can still gain a foothold in the music sector?
Click to expand...
Although AI-generated music might seem good, compared to the work of any composer it is still very basic. This is mainly because the models they are trained on are often musical pieces and not single instruments, there is no way to fully customise the output nor do they generate elaborate pieces yet.

With text, which is easier to train on, generating content is a lot easier, and you have thousands of AI-generated books. Authors who churn out books ridiculously fast likely use AI too, for a large part of the books they release. Even so, this is no replacement for quality work by a human, there's always a 15-20% that needs to be put in to reach an average standard for someone who is well-read. Nothing extraordinary, just average. Music generation is nowhere near this.

pulpfiction said:
If countless songs are now being produced with the click of a mouse, isn't there a very good chance that the songs made by us without Ai will already exist in a similar form? (which could lead to legal disputes...)
Click to expand...
This is nothing new, disputes existed even without AI, classical composers are known to lift bits and pieces into their work. Arguably, this is even more common today where you can reach a lot more people via the internet.

If someone uses the same model to generate music, similarities could happen, just like with two human composers subconsciously or consciously using motifs from other works.

pulpfiction said:
Even if this is curbed a little with laws in the foreseeable future....for example...only things made by humans can be protected with copyrights.....how will you know the difference? It's already the case that every fart, at least in the top charts of the mainstream, is being sued for with lawyers....because even without AI many things sound too similar...
Click to expand...
There are efforts to watermark and detect AI-generated songs, just like with AI-generated text or images. Currently most AI-generated audio has a little bit of fuzz too, which is a tell-tale sign of an AI being used.

pulpfiction said:
How do you manage to continue pursuing your passion, even though in a few months it might all be unnecessary because AI will eventually overtake us in all areas?
Click to expand...
If it's a passion, you possibly do it first and foremost because you like what you are doing. AI can't take that away, no matter how advanced it becomes in the future.

Hand-made objects, human-narrated audiobooks, paintings, or any other art is still appreciated by many people all around the world. AI just gives access to anyone to generate an idea, but it can be only as good as the used model and even then it currently needs a lot of system resources. We're not going anywhere 😀
 
Brendon V said:
Someone with an in-depth knowledge of music theory and editing of music is going to get far better results than an average user.
Click to expand...
Not really. That is delusional thinking. Trying to hold on to a raft that is not really there.

If Bernstein and I inputted a prompt it's up to Mr. AI/chance to determine how good or not the output will be.

That is the whole point of the tech.

Music theory knowledge or any musical knowledge or lack of is completely redundant.
 
thorwald said:
Although AI-generated music might seem good, compared to the work of any composer it is still very basic.
Click to expand...
I dispute this unfortunately.
I would say that many Udio "compositions" or whatever we should call them, are much better written, better sounding, better produced, even more memorable then a lot of human made music I heard lately. Sorry. I feel sick while writing this.
(Not the most original surely, that's for now saving us)

Then there is the fact that to achieve this, the various AI developers spit in the face of all the musicians in the world, living or not, and just straight up robbed them in full daylight, hoping to just go on like nothing happened.

It's terrible to say - but in my last week exploration of AI music, in order to "know the enemy", I have stumbled on a few results that are still in my head - meaning well written, memorable themes.
Also very versatile, I heard perfect proper blues tunes that sounded like they were recorded in 1955 on a raft on the Mississippi, and right after some proper very well made traditional Indian tracks.
Not many human composers and writers can pull that off.

A solution to fight this madness has to exist, but denying it's achievements isn't one of them - in my opinion.
 
AudioLoco said:
I would say that many Udio "compositions" or whatever we should call them, are much better written, better sounding, better produced, even more memorable then a lot of human made music I heard lately. Sorry. I feel sick while writing this.
(Not the most original surely, that's for now saving us)
Click to expand...
I think originality is the keyword here. It is a lot easier to prompt-engineer good-sounding music from models that have well-produced tracks. If you compare that to, say, library music where one of the goals is to create as many tracks as possible in a reasonably short amount of time, the tracks strive for a specific standard, they are going for an effect as it were.

AudioLoco said:
Then there is the fact that to achieve this, the various AI developers spit in the face of all the musicians in the world, living or not, and just straight up robbed them in full daylight, hoping to just go on like nothing happened.
Click to expand...
This is something I absolutely agree with, but I feel it's way beyond the scope of this topic.

AudioLoco said:
Also very versatile, I heard perfect proper blues tunes that sounded like they were recorded in 1955 on a raft on the Mississippi, and right after some proper very well made traditional Indian tracks.
Not many human composers and writers can pull that off.
Click to expand...
I think this is a bit unfair. AI-generated music is going to be as good as the used model. If the orchestration, instruments, setting, etc, were outstanding and they are part of the used model, the end results will reflect that to an extent. Expecting the same from a human orchestra or from modern instruments is like expecting a sample library to replicate a live concert. An AI has more data to work with.

But on the other hand, it is still an AI, working from a finite amount of options and parameters, as opposed to a human. It might fool people, just like sample libraries do, but ultimately it is as good as what people put into it. I am sure someone could build a ridiculously huge sample library with this in mind, but it hasn't happened so far.

AudioLoco said:
A solution to fight this madness has to exist, but denying it's achievements isn't one of them - in my opinion.
Click to expand...
I use AI every day. For audio restoration, for describing images (because I am blind), for orientation when I'm on the move, to read text on a computer, etc. I train models when there's a need. I have a quite good idea of how AI works. If I denied its achievements I would be the biggest hypocrite on Earth. No, AI can certainly do things that would be impossible to do otherwise, some of which I'm even grateful for, but fearing that it will replace a human is like saying that sample libraries are bad because they completely replace human musicians.
 
AudioLoco said:
I would say that many Udio "compositions" or whatever we should call them, are much better written, better sounding, better produced, even more memorable then a lot of human made music I heard lately.
Click to expand...
I have to agree, unfortunately, from what I've heard from pop radio lately (although, in my opinion, pop these days is a very low bar). There are songs it generated I'd actually listen to unironically. Shocking and genuinely scary how far this has come.
 
There's another aspect that keeps coming to mind recently:

As our forum suggests, this is about virtual instruments and there are also many professionals in the field here. I've been feeling old-fashioned with my virtual instruments lately, now that I know that there is technology that can generate any instrument and any voice almost perfectly, without the need to use a huge and expensive sample library that still requires many good musicians to produce.
Even though I recently acquired Synthesizer V with a few voices....

In your opinion, does it make sense to invest more money in VST libraries in the near future (especially if you want to start with music....)? Wouldn't it make more sense to advise a beginner to wait a few more months because things are coming onto the market soon that might make VST instruments irrelevant?
 
You must log in or register to reply here.
Top Bottom