• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

AI and Future of Music Production

New version of an original song of mine from the 80s generated with the new YuE2 AI Music model running locally on my pc in WanGP. I used the original recording as a reference track for the AI model to follow. It did a pretty good job following the original melody and replacing the original real male voice with a generated female vocal. YuE2 generated a little better quality audio than Ace-Step 1.5, but is much slower. This track took almost 30 minutes to generate.

 
A new song. I was trying for a baroque chamber music style with piano, violin, cello and viola and female vocals. I used the YuE2 model running in WanGP on my pc. The cover image was generated by Grok.

 
What do you think?

I think Clanker crud should be nuked from orbit. As a person who has spent decades honing a take on music and then bring that take or perspective to the people, I take HUGE exception at ai scraping it and adding it to the LLm so some !@#$$%$^ little !#$%@#@#$ can then create their own version, addition, what ever with out doing ANY of the work that I have done to crawl my way through absolute HELL to get to this point. I have lost friends, I have survived and I am still here trying to create 100% human art.

Fuck Ai.

If you wanna chew the fat, I will buy you a REAL beer at the REAL dive bar and we can turn our DEVICES OFF.

I know my stuff gets scraped, I know it gets chucked into the LLM. Without my explicit permission.

NO.
 
I suppose it's only a matter of time before a sadder but wiser AI agent writes a 12-bar blues about The Hugging Face Incident...
 
Love it.

Watch out Howlin' Wolf, Sonny Boy Williamson et al.
Them damn blues robots are coming for ya!

Lol.
 
I've been figuring out how to use the ABC music notation system. The new YuE2 music model can take an ABC file as input. I found a native Linux App called EasyABC that lets you see the sheet music as you are writing the notes, etc. in ABC. Pretty slick.

I wrote simple melody lines for these two instrumental only songs. They were generated with YuE2 running in WanGP on my pc. I had Grok generate the cover images.


 
For anyone interested, this is what the EasyABC editor looks like:

1790661071960.png
 
As someone more involved in AI/machine learning in general, I think AI for music is going to be a big nothing burger at least in the foreseeable future.

Consider this:
  • AI model use to create new music given a prompt is only as good as the prompt's content. This implies creativity needed in order to get a creative output, which doesn't change much with respect to whether new tools increase or decrease creativity.
  • Intelligence models are only ever as good as their training data. If you try to infer data outside of their training set, then it doesn't work. The reason we don't notice this with our human brains is because we have thousands to millions of years of model training experience along with several decades of multimodal information collection.
  • Training new models requires inordinate amounts of computing power, and until some big advance comes along (analog photonics processing!) and makes this cheaper and faster, model training is super expensive and time consuming to get a quality result.
  • AIs don't actually "know" anything about music, specifically music theory rules. The same is true of other AIs solving math problems. There is currently research here to figure out how to add concrete rules and proofs to models, but it's not clear whether this is realistic or possible yet.
  • If you're concerned about AIs replacing musicians, then just get into live music because AIs will never be able to replace that by definition.
Basically, until something big happens, AI is similar to the change brought by the ability to record individual tracks instead of a band playing live.

As a life-long classically trained composer and music producer your perspective is closest to mine in regards to the above comments. No doubt AI will produce amazing advances in medicine, materials science, mathematics and numerous other fields. But in the fine arts? Probably not. Here's why: Making music is about sensing, feeling, observing, imagining and evaluating sound and ideas. This all requires personhood, consciousness, sentience. Since AI feels nothing, values nothing and can't even experience the taste of dark chocolate, how in the world would it feel the difference between a meaningful (subjective) musical experience and one that leaves one empty or bored? It cannot.

And there's another equally important reason why AI's role in the fine arts will be very limited. When I work on a musical composition, whether it be a symphony, song, or soundtrack, I am making thousands of decisions. The joy and challenge of creating music is in the making of these many decisions. Turning over the decision-making process to a machine would be like, at least for me, having someone else eat my food or embrace my loved ones on my behalf. It's the experience itself which is of ultimate value. AI cannot experience life and art is about life. Even if you feed AI all the "rules" of harmony and counterpoint, even if you teach it to expertly imitate the style of a particular composer, there will be something profoundly lacking--the experience of a sentient person having something to say in music.

Of course there will always be those untalented people (who are not aware they have no talent or expertise in music) who will heartily embrace AI. After all, it will support their delusion that they're creating something original, something of value, something that has depth, craft and a sense of intentionality. Pop movies and pop music will embrace AI as well, but nothing good will come of it other than slickly produced imitative drivel.

This being said, AI will possibly find a place in the arts that I am not considering. I am not sure how, but I assume that will happen in ways I am not aware of. I know it won't happen with my own work but every artist is different so I have to leave room for exceptions, as there will always be those individuals who continue to surprise us, rare as they might be.

Jerry
 
Last edited:
As a life-long classically trained composer and music producer your perspective is closest to mine in regards to the above comments. No doubt AI will produce amazing advances in medicine, materials science, mathematics and numerous other fields. But in the fine arts? Probably not. Here's why: Making music is about sensing, feeling, observing, imagining and evaluating sound and ideas. This all requires personhood, consciousness, sentience. Since AI feels nothing, values nothing and can't even experience the taste dark chocolate, how in the world would it feel the difference between a meaningful (subjective) musical experience and one that leaves one empty or bored? It cannot.

And there's another equally important reason why AI's role in the fine arts will be very limited. When I work on a musical composition, whether it be a symphony, song, or soundtrack, I am making thousands of decisions. The joy and challenge of creating music is in the making of these many decisions. Turning over the decision-making process to a machine would be like, at least for me, having someone else eat my food or embrace my loved ones on my behalf. It's the experience itself which is of ultimate value. AI cannot experience life and art is about life. Even if you feed AI all the "rules" of harmony and counterpoint, even if you teach it to expertly imitate the style of a particular composer, there will be something profoundly lacking--the experience of a sentient person having something to say in music.

Of course there will always be those untalented people (who are not aware they have no talent or expertise in music) who will heartily embrace AI. After all, it will support their delusion that they're creating something original, something of value, something that has depth, craft and a sense of intentionality. Pop movies and pop music will embrace AI as well, but nothing good will come of it other than slickly produced imitative drivel.

This being said, AI will possibly find a place in the arts that I am not considering. I am not sure how, but I assume that will happen in ways I am not aware of. I know it won't happen with my own work but every artist is different so I have to leave room for exceptions, as there will always be those individuals who continue to surprise us, rare as they might be.

Jerry
I can understand why making all those decisions yourself is such a big part of the enjoyment. But I think that’s a separate question from whether someone else could find AI-assisted music meaningful.

Someone using AI may still put plenty of creativity, judgment and effort into shaping the result. I wouldn’t assume they lack talent or are fooling themselves simply because they use a different tool.

I’d rather hear what they’ve made before deciding whether it has value.
 
I can understand why making all those decisions yourself is such a big part of the enjoyment. But I think that’s a separate question from whether someone else could find AI-assisted music meaningful.

Someone using AI may still put plenty of creativity, judgment and effort into shaping the result. I wouldn’t assume they lack talent or are fooling themselves simply because they use a different tool.

I’d rather hear what they’ve made before deciding whether it has value.

Which is why I wrote the last paragraph:

This being said, AI will possibly find a place in the arts that I am not considering. I am not sure how, but I assume that will happen in ways I am not aware of. I know it won't happen with my own work but every artist is different so I have to leave room for exceptions, as there will always be those individuals who continue to surprise us, rare as they might be.

AI isn't merely just "a different tool". It's not like any musical instrument that has existed in the past or present, including the desktop computer. Why? Because it's a tool that makes artistic decisions. I've worked with MIDI on a full-time basis for over 40 years. But even with the most sophisticated DAW, synthesizers and sound libraries, these tools do not make musical decisions, not regarding melody, harmony, voice-leading, counterpoint, orchestration, structure, length, dynamics, tempo, meter, rhythm or texture. Do these tools influence the sound? Of course they do, this is true for all mediums. But AI is radically different.

I've been watching several AI "documentaries" on YouTube about early man. It's AI slop, and that's being generous. Why? Because the "creators" don't know what they're doing, or they just don't care. Dialogue is repeated by mistake and they don't catch it. There's no art direction. They give AI commands as to what they want to portray but they don't edit out the mistakes, for example Homo Erectus living some million years ago using a modern metal shovel. Or one face in the tribe looking like Homo Sapiens and another one directly next to him looking like a creature about 1/2 way between an ape and a human. This sloppy lack of attention to detail is common in the several videos I tried watching. Because the people using AI apparently have no training in art direction or dialogue editing or even some basic sense of consistency, they produce nonsense. These works are not ready for publication, but since they have no professional standards they publish any way. If I can't trust the integrity of the production how can I trust the paleo anthropology behind the story they're telling? I'm not even sure the narrator is a real person.

My brother-in-law recently played me a recording of a "band" I had never heard of. Great production values, clean, clear vocals, the music made sense. But there was no character, no originality, the female singer sounded exactly like a generic female pop singer, every chord change predictable. It was AI. At first I thought, oh, they sound professional but like every other band, it had no sound of its own. I was in fact told after I listened it was AI.

None of the composers and serious musicians I know are going to use AI to compensate for a lack of talent, knowledge, skill or training. There are no shortcuts to becoming a good musician. This is true whether your goal is to become a jazz, rock, classical, pop or any other kind of musician.

If having fun is the goal and you get pleasure and meaning telling AI what to do, it doesn't make a difference to me. Keep having fun, maybe something good will come of it.
 
Last edited:
The Suno Studio tools in the paid Premier version are simply amazing for mixing and mastering. I know it can be abused and set to automatically create whole songs in seconds, but as an audio workstation it is fantastic. I could not go back to a traditional DAW.
 
As a life-long classically trained composer and music producer your perspective is closest to mine in regards to the above comments. No doubt AI will produce amazing advances in medicine, materials science, mathematics and numerous other fields. But in the fine arts? Probably not. Here's why: Making music is about sensing, feeling, observing, imagining and evaluating sound and ideas. This all requires personhood, consciousness, sentience. Since AI feels nothing, values nothing and can't even experience the taste of dark chocolate, how in the world would it feel the difference between a meaningful (subjective) musical experience and one that leaves one empty or bored? It cannot.

And there's another equally important reason why AI's role in the fine arts will be very limited. When I work on a musical composition, whether it be a symphony, song, or soundtrack, I am making thousands of decisions. The joy and challenge of creating music is in the making of these many decisions. Turning over the decision-making process to a machine would be like, at least for me, having someone else eat my food or embrace my loved ones on my behalf. It's the experience itself which is of ultimate value. AI cannot experience life and art is about life. Even if you feed AI all the "rules" of harmony and counterpoint, even if you teach it to expertly imitate the style of a particular composer, there will be something profoundly lacking--the experience of a sentient person having something to say in music.

Of course there will always be those untalented people (who are not aware they have no talent or expertise in music) who will heartily embrace AI. After all, it will support their delusion that they're creating something original, something of value, something that has depth, craft and a sense of intentionality. Pop movies and pop music will embrace AI as well, but nothing good will come of it other than slickly produced imitative drivel.

This being said, AI will possibly find a place in the arts that I am not considering. I am not sure how, but I assume that will happen in ways I am not aware of. I know it won't happen with my own work but every artist is different so I have to leave room for exceptions, as there will always be those individuals who continue to surprise us, rare as they might be.

Jerry
Stupid people experience life naively and create banal art. Being human is not the prerequisite for creating meaningful and lasting artistic expression. Ai takes the best of human expression and combines it into something that can exceed the output of human fools. Sublime art will always be from humans. Satisfactory entertainment will be well within the ability of ai. So much of pop music is garbage. Ai pop music isn’t really much worse.
 
New song generated with YuE2 in WanGP on my pc. I wrote the lyrics and used my melody in ABC music notation. I had Grok generate the cover image since Krea 2 doesn't do night scenes as well.

 
A new song generated with YuE2 in WanGP on my pc. I generated the cover image with Krea 2 in Forge-Neo. I wrote the original lyrics for this a while back, but wasn't satisfied with them. I finished them earlier today.


This is a new version of a more metal style song I wrote a while back. The original version was generated with Ace-Step 1.5 xl-sft. This version was generated with the newer YuE2 model running in WanGP. I generated the cover image with Krea 2 in Forge-Neo. I wrote the lyrics for this one a while back.

 
And now for something completely different. I used lyrics from an old metal song of mine written in the 80s. The lyrics have a bit of a horror / madness theme. I decided to redo it as a cheerful country swing style song. I generated the music and vocals with YuE2 running in Unsloth Desktop. I expected it to be completely wrong and bad, but I like the way it came out.

 
Since I'm trying out different styles, I decided to convert another song of mine, "One Last Dance" to an EDM style. I rewrote the lyrics to work better in that format and generated the new version with YuE2 in Unsloth Desktop.

 
Back
Top Bottom