AI generated music has taken a huge leap

wesleyamltd said:
So AI music is not music because it’s not written by a musician…
Click to expand...
And in what dictionary is music defined as being exclusive to musicians?

For the record, AI music is generated based on training data that consists of music created by musicians, and it learns the patterns and structures based on original artist works much the way people learn patterns and structures from music created by other artists. In that regard, AI generated output indirectly traces back to the musicians whose work formed the foundation of its training rather than originating exclusively from an AI or end-user.
 
Jason Scott said:
Context in what sense? When I heard More Than A Feeling for the first time, I didn't need context to know I liked the song. The same is true for countless other songs in my case. Whether a song remains relevant was independent of my point.


In my experience, there's most assuredly no dearth of spontaneity and/or surprise when generating AI music within the framework of my preferences, but that's me. The internet and social media already provide plenty of opportunities to be exposed to a wide array of music, and I don't expect that will change any time soon.
Click to expand...

IMO, a song isn't just a mixture of chords, riffs and melody, and technique. Everything you happen to know about the artist(s), the era, other songs by the artist, etc. etc. adds to the context. Without all that, a song is just a short happy meal without soul. AI-written material has no context.

Plus, a sad, happy or angry song, coming from a genuine emotional state of a musician/writer, is something completely different than the result of instructing AI to write a sad, happy or angry song.

But to be honest, at this point I don’t even know anymore what the point is of this entire discussion. No one is going to convince the other side. And that’s fine.
 
yek said:
IMO, a song isn't just a mixture of chords, riffs and melody, and technique. Everything you happen to know about the artist(s), the era, other songs by the artist, etc. etc. adds to the context. Without all that, a song is just a short happy meal without soul. AI-written material has no context.
Click to expand...
Sometimes context can enhance my enjoyment of a band. For example, I've always loved KISS (in their prime, that is), though without their iconic image, I would've appreciated them less, but still a lot. For me, context can enhance my enjoyment, but it's not a prerequisite.

As I said, I didn't need context to know I loved Boston's More Than A Feeling, Dio's Rainbow In The Dark, or Christopher Cross's Sailing the first time I heard them. Maybe there's some hard rock fans out there that never liked Back In Black, but to me that song stands on its own, independent of who wrote it. It's just a damn good song.

yek said:
Plus, a sad, happy or angry song, coming from a genuine emotional state of a musician/writer, is something completely different than the result of instructing AI to write a sad, happy or angry song.
Click to expand...
If someone writes a sad, happy, or angry song that comes from a genuine emotional state, that authenticity gets embedded in the training data, and the model iterates from the learned patterns that were originally created by their emotional intent. In short, a model's ability to generate emotionally resonant music depends on the emotional depth and creativity of the person whose work comprises the training data.

yek said:
But to be honest, at this point I don’t even know anymore what the point is of this entire discussion. No one is going to convince the other side. And that’s fine.
Click to expand...
For me, the point isn't necessarily to convince anyone, but rather to give people a chance to discuss and share perspectives on AI generated music. In my opinion, it's perfectly fine to walk away in respectful disagreement.
 
Last edited:
  • Like
Reactions: yek
Jason Scott said:
And in what dictionary is music defined as being exclusive to musicians?

For the record, AI music is generated based on training data that consists of music created by musicians, and it learns the patterns and structures based on original artist works much the way people learn patterns and structures from music created by other artists. In that regard, AI generated output indirectly traces back to the musicians whose work formed the foundation of its training rather than originating exclusively from an AI or end-user.
Click to expand...
I think that you’re being reductive and conflating neuroanatomy with a computer algorithm. A software program is not sentient and therefore cannot think for itself. Very different from the human brain.

Sure the AI music is a derivative of real music created by musicians but making music using AI is not music, it’s a facsimile of music that was made by musicians. Therefore, AI generated music is not music and it will never be until AI becomes sentient and can think for itself.
 
wesleyamltd said:
I think that you’re being reductive and conflating neuroanatomy with a computer algorithm. A software program is not sentient and therefore cannot think for itself. Very different from the human brain.
Click to expand...
You seem to be under the impression that sentience is required to make music. Here's Webster's definition:

Music 1a: vocal, instrumental, or mechanical sounds having rhythm, melody, or harmony

By that definition, even rhythmic (non-human) mechanical sounds (e.g. water dripping on a pipe) could potentially be considered music. Ever heard of an Aeolian harp? It creates music composed by the wind:

 
Last edited:
Jason Scott said:
You seem to be under the impression that sentience is required to make music. Here's Webster's definition:

Music 1a: vocal, instrumental, or mechanical sounds having rhythm, melody, or harmony

By that definition, even rhythmic (non-human) mechanical sounds (e.g. water dripping on a pipe) could potentially be considered music. Ever heard of an Aeolian harp? It creates music composed by the wind:

Click to expand...

You seem to glean over the context of my sentence. The implication is the music is made by musicians. The definition of music is the end result. Music is our brain making meaning from sound, without the human link there is no music.
 
I remember when MIDI quantization was in it's infancy but was so effective that the race was on to 'humanize' the results.

There's a reason for that.
 
Considering if something is music is up to the listener.
After all, you can hear music in 2 cement mixers next to each other. Either way just a semantic discussion.

I don't believe most people can tell the difference between AI created music and actually recorded music anymore. Especially since al lot of recorded tunes are overproduced and compressed into nothingness. But that is a creative choice in the end.
And don't forget that AI will get better exponentially in a short time.

Needing context to enjoy a piece of music seems illogical to me. Maybe it's clutching at straws in order to dismiss AI, but music essentially is just a play on rhythm, harmony and melodie. I read somewhere that Trent Reznor was surprised people didn't needed that background, but I believe some artist are a but full of themselves and over romanticize things (which he admitted to LOL).
Myself, I like music better if I don't know anything about the personal life and struggles of the artist and like to judge music on it's own merit.

Anyway, just my 2 cents while being bored at work, which AI will take over anyway in the not so distant future :p
 
wesleyamltd said:
You seem to glean over the context of my sentence. The implication is the music is made by musicians. The definition of music is the end result. Music is our brain making meaning from sound, without the human link there is no music.
Click to expand...
I believe I did understand the context. You seem to be claiming that our brains interpret certain sounds as music, and without us, there would just be sound, because music is an interpretive process. However, even if you accept that music is interpretive, that's independent of whether a human is required to have created the sound in order for us to interpret it as music. For instance, whales and the Aeolian harp (above) are two examples of sounds that people interpret as music that are not created by humans. It's a bit like claiming that the beauty of a sunset is interpretive, and as such, requires that a human is responsible in some fashion for its existence.
 
Last edited:
wesleyamltd said:
You’ve drank the Kool-aide
Click to expand...
What Kool-aide? Did you not read my post? It doesn't logically follow that if humans interpret sound as music that humans therefore must have created said sound, anymore than it follows that humans must have created sunsets simply because we interpret them as beautiful.
 
Last edited:
The best music to my ear has an aspect of "humanness" (unpredictability, phrasing ...) in it. Imo, AI doesn't replace that any time soon tho it might succeed faking some of the bland oversimplified templated garbage often passing for music these days - and since there seems to be less and less room, as time goes forward, for music that has a depth of human vibe to it, maybe its best just to let AI take over the job. While its at it, AI could generate concerts with 3D holographic players, to allow young people to actually see at least some sort of remnance of a rock show without having to take out a mortgage on the tickets. So glad to have grown up in the 70s, with fond memories of many headlining live shows attended with few odd-jobs $, and now having a few $ / mth to stream all those irreplaceable, unautogenable classics.

(ya ya - old man yelling at clouds).
 
Last edited:
Jason Scott said:
What Kool-aide? Did you not read my post? It doesn't logically follow that if humans interpret sound as music that humans therefore must have created said sound anymore than it follows that humans must have created sunsets simply because we interpret them as beautiful.
Click to expand...
You’re are overlooking the fundamental truth, perception is reality. Also, based on your definition of musician and music, my logic stands.
My Venn diagram:
A musician is someone who writes, compose, preforms music.

Music is vocal or instrumental sounds (or both) combined in such a way as to produce beauty of form, harmony, and expression of emotion.

Both cannot exist without the other.

If all AI generated music is based on the music created by humans then AI cannot create its own music much less interpret an emotion or opinion of music generated without the human component.

AI generated music is prescribed meaning based on direct human interaction.

Final point, you are prescribing meaning to AI generated music throughout this entire thread and in so doing you are quantifying what it means to you based on feelings and emotions. Human feelings and human emotions.

Lastly, your arguments are conflating widely different ideas that in themselves seem related but lack the nuance to be a compelling argument. Like the idea of humans taking responsibility for the sunset. This makes no sense.

“It doesn't logically follow that if humans interpret sound as music that humans therefore must have created said sound anymore than it follows that humans must have created sunsets simply because we interpret them as beautiful.”

I said, humans give sound meaning, otherwise it exists in a vacuum.

Who/what creature has the ability to reason?

The world will exist without people to ask the question, “I think therefore I am” and sunsets exits with or without human interaction.
But the sunset will not be beautiful without a human to appreciate it.
 
I had this old, short track I wrote as a teenager. Very bad mix -- using some usb mic, I think. I used AI and came up with this edition. It is quite faithful to the original, melodies and riffs, but the mix sounds way better. I left solos out. So I think there's such uses for AI as well. It is really quite amazing what you can do.

 
You must log in or register to reply here.
Back
Top Bottom