Is this what will literally completely devalue song-writing, engineering, mixing, mastering?

[RichardT]

Playback from scoring software has come on leaps and bounds...Strings are there already, IMO - at least, I don’t think most people could tell the difference.

Sounds interesting...I think you might have mentioned something along these lines before. What kind of software are we talking about...? Unless it's a trade secret.

Noteperformer. It runs in scoring software such as Dorico. It interprets the score and controls the sample libraries.

https://www.noteperformer.com/

[alexis]

Here's another negative aspect of "AI Music" I just thought of (negative to creators only, no one else cares ...). I'm probably late to the party on this but here goes:

You come up with a nice song, and more specifically a nice little musical arrangement of instruments within that song.

You play it for someone, and they say, "Oh that's really good, you probably used AI, right? My nephew does that sometimes, he doesn't have a musical bone in his body, but some of the stuff he makes isn't bad at all!".

You worked so hard on what you did, but there you go, the burden of proof is on you that you created that.

And the worst part (?) is no one really cares, and you can't blame them - to them, AI has made "creating music" to be as creative as choosing what fake background to display on your Zoom call.

(I'm writing this this morning because I was watching an AI YouTube where some dude was "AI-ing" a drum track, and got one of "OK" quality in a matter of seconds. It made me think of a Caribbean sounding drum intro I wrote that took literally weeks to get to the point where it could be rendered, from literally sketching it out on napkins in the car at stop lights, seeing how it actually sounded rhythmically with a dummy drum track, revising and repeating, auditioning tons of Caribbean drum instruments to match the sound I was looking for ... It's hard for me to not ask myself, "Why? What's the point? You could have just asked AI to spit out a Caribbean beat, and it would have worked just fine as far as anyone else was concerned ...")

[Drew Stephenson]

I think being able to play your music live is going to become something of a differentiator. I've already seen various 'hot takes' along those lines about how you're not a real musician if you can't play it live etc. etc.

Obviously this is a pretty stupid intellectual position to hold once you give it a moment's thought but you can easily see why people will cling to it as a way of identifying 'real' music.
Which is great if that's your thing, not so good for others though.

But it's disruptive technology, it will disrupt. For a while. Then things will settle down and we'll get on with things in whatever new paradigm we're operating in.
It's what we do.

[alexis]

I think being able to play your music live is going to become something of a differentiator ...
Which is great if that's your thing, not so good for others though.

Most of what we hear can't be performed live of course, but Taylor Swift shows the audience will fill seats even so.

Will AI ever write songs good enough for her to perform? Will song writers become redundant? I don't know the answer to that question. As you say, it's hard to predict what things will look like after a disruption.

I was writing with the focus of AI's effect on the very local hobbyist, like me. It's a selfish point of view, given the much more massive disruptions to others, I know. I guess I hadn't realized AI's effect would reach all the way down to me.

[RichardT]

Here's another negative aspect of "AI Music" I just thought of (negative to creators only, no one else cares ...). I'm probably late to the party on this but here goes:

You come up with a nice song, and more specifically a nice little musical arrangement of instruments within that song.

You play it for someone, and they say, "Oh that's really good, you probably used AI, right? My nephew does that sometimes, he doesn't have a musical bone in his body, but some of the stuff he makes isn't bad at all!".

You worked so hard on what you did, but there you go, the burden of proof is on you that you created that.

And the worst part (?) is no one really cares, and you can't blame them - to them, AI has made "creating music" to be as creative as choosing what fake background to display on your Zoom call.

(I'm writing this this morning because I was watching an AI YouTube where some dude was "AI-ing" a drum track, and got one of "OK" quality in a matter of seconds. It made me think of a Caribbean sounding drum intro I wrote that took literally weeks to get to the point where it could be rendered, from literally sketching it out on napkins in the car at stop lights, seeing how it actually sounded rhythmically with a dummy drum track, revising and repeating, auditioning tons of Caribbean drum instruments to match the sound I was looking for ... It's hard for me to not ask myself, "Why? What's the point? You could have just asked AI to spit out a Caribbean beat, and it would have worked just fine as far as anyone else was concerned ...")

A lot of painters in the past had studios in which other artists did the leg work on their pictures. Damien Hurst has done that on a huge scale recently.

Hans Zimmer uses other musicians to help realise and orchestrate his ideas, in fact lots of composers for film do that. And many jazz musicians just write a lead sheet, and their band members realise the details for each instrumental part collaboratively.

The famous painters had the vision and artistic ideas, provided guidance, and sometimes painted critical parts themselves. I think music with AI may become like that.

The key is finding a role for yourself in this world. It will involve less detailed work but more concepts and ideas.

[amanise]

....I was writing with the focus of AI's effect on the very local hobbyist, like me. It's a selfish point of view, given the much more massive disruptions to others, I know. I guess I hadn't realized AI's effect would reach all the way down to me.

The hobbyists are in the best position, being able to decide how much to let AI in to their processes - and how much to keep it out. The ones who'll be hit hardest and first will probably be the pros who depend on income from library or jingle type music, as we're already seeing here on this forum. For me it's the long hand processes I like. I won't be changing anything of what I do.

[RichardT]

....I was writing with the focus of AI's effect on the very local hobbyist, like me. It's a selfish point of view, given the much more massive disruptions to others, I know. I guess I hadn't realized AI's effect would reach all the way down to me.

The hobbyists are in the best position, being able to decide how much to let AI in to their processes - and how much to keep it out. The ones who'll be hit hardest and first will probably be the pros who depend on income from library or jingle type music, as we're already seeing here on this forum. For me it's the long hand processes I like. I won't be changing anything of what I do.

Yes, those professionals are in trouble - anyone, really, who creates functional music.

I will make some changes, I think. For example, an AI that could write a proper piano reduction of an orchestral score would be a total godsend. Without it, it’s a pure chore.

[amanise]

....I was writing with the focus of AI's effect on the very local hobbyist, like me. It's a selfish point of view, given the much more massive disruptions to others, I know. I guess I hadn't realized AI's effect would reach all the way down to me.

The hobbyists are in the best position, being able to decide how much to let AI in to their processes - and how much to keep it out. The ones who'll be hit hardest and first will probably be the pros who depend on income from library or jingle type music, as we're already seeing here on this forum. For me it's the long hand processes I like. I won't be changing anything of what I do.

Yes, those professionals are in trouble - anyone, really, who creates functional music.

I will make some changes, I think. For example, an AI that could write a proper piano reduction of an orchestral score would be a total godsend. Without it, it’s a pure chore.

I suppose I'd use an AI agent/promoter bot that crawled the Internet pitching my tunes at labels, curators, radio or whatever. Just set it going with instructions to not come home until its scored

[James Perrett]

I would guess that run of the mill pop music will embrace AI but anyone who wants to create something different will go out of their way to create something that AI simply couldn't come up with.

Feel is already important in live music - subtle timing changes that turn a mechanical interpretation into something with emotion. It isn't something that people necessarily recognise but it makes the difference between a functional piece of music and a great piece of music. Can AI re-create that feel?

[Drew Stephenson]

Feel is already important in live music - subtle timing changes that turn a mechanical interpretation into something with emotion. It isn't something that people necessarily recognise but it makes the difference between a functional piece of music and a great piece of music. Can AI re-create that feel?

Agree entirely. And for those musicians who don't play live, moving away from the grid will become one of the ways that they will be able to differentiate themselves.
I reckon.

[RichardT]

Feel is already important in live music - subtle timing changes that turn a mechanical interpretation into something with emotion. It isn't something that people necessarily recognise but it makes the difference between a functional piece of music and a great piece of music. Can AI re-create that feel?

Agree entirely. And for those musicians who don't play live, moving away from the grid will become one of the ways that they will be able to differentiate themselves.
I reckon.

I think that won’t last, and AI will be able to generate feel before long.

[BWC]

...moving away from the grid will become one of the ways that they will be able to differentiate themselves.
I reckon.

I really hope so!! I quickly get bored with most hard-quantized music (timing and pitch). It's starting to seem to me that this "AI" thing might just lead to a good (maybe surprising) result. The quick & easy, "anyone can slap that together", stuff will be devalued, but this could then increase (or remind everyone of) the value of real human creative expression.

[BWC]

...and AI will be able to generate feel before long.

I'm less sure about that. Again, not really autonomous or self-aware, so not really "AI". When it is, that might be a concern. I'm always hearing about "the next five to ten years!", but I saw an article the other day saying that we'd be learning a lot more about how the brain works over the course of the next... century!

[Drew Stephenson]

I think that won’t last, and AI will be able to generate feel before long.

Yeah, I may be being optimistic. On the one hand we have the old saw that many things that humans find easy to do are difficult for robots to do (and vice versa), but the counter argument is to look at how quickly AI is improving and to remember that we're not stuck in the old quandary of having to define 'feel' before we can programme it, AI will learn it as we do without having to worry about those restrictions.

Maybe I'll just give this all up and start cleaning bathrooms.

[Drew Stephenson]

The quick & easy, "anyone can slap that together", stuff will be devalued, but this could then increase (or remind everyone of) the value of real human creative expression.

We can but hope.

[BWC]

I should say that, while I maintain that it's not really "AI", I do think it's impressive, and valuable, and will disrupt a bit. It might be able to reproduce past "feel", or even occasionally combine such things in new, maybe interesting, ways, but it won't be creating new deviations from the grid based on relatable human emotions. That's the part that I don't think is coming along anytime soon.

...we're not stuck in the old quandary of having to define 'feel' before we can programme it, AI will learn it as we do without having to worry about those restrictions.

We have to define it to program it, but not to play it. We have the advantage of being able to feel the feels. I understand your point, but the neuroscience researchers aren't against using "AI" in their work, and are still saying that we have a century's worth (or more) of work to do to understand how these brain thingies work.

[RichardT]

Just out of interest, and because as a programmer I find it so amazing, here’s a chatGPT dialogue where the mathematician Terry Tao asks ChatGPT to generate some code.

https://chatgpt.com/share/79ab37e7-d14d ... 794a0a9581

I put the same question to ChatGPT to see how long it would take to do that - it took about 15 seconds.

This line works out if a number is the square of an integer:

return int(n ** 0.5) ** 2 == n

This is so clever. It takes the square root of the number n, takes the integer part of the square root and squares it again, and returns TRUE if that number equals n, FALSE otherwise.

[Folderol]

I'm not up to date with python, but it seems to take one unnecessary step.
In C/C++ I'd simply take the square root and test if it is an integer.

[BJG145]

For info, ChatGPT agrees that your way is better.

[BWC]

For info, ChatGPT agrees that your way is better.

Maybe because he spent more than 15s on it? So many are obsessed with speed, but speed often costs more time than it saves. Efficiency is far better for saving time. Time spent in traffic, shopping for groceries, etc. has shown me that most people don't understand (have never even considered) that.

[RichardT]

I'm not up to date with python, but it seems to take one unnecessary step.
In C/C++ I'd simply take the square root and test if it is an integer.

True. But I’m still impressed.

Look at the indices i, j and k in the routine valid_subset. I know it’s it’s a standard technique for iterating permutations of a set, but I think it’s slightly mind boggling that ChatGPT can build that without ever having been taught how to code. The inventors of AIs were not expecting them to be so good at this.

Personally, to be fair, I would modify the code a little as it’s relying on Python behaving a certain way when ranges are invalid, but still….

The day is close when a good design can be converted into code automatically.

[BWC]

...without ever having been taught...

Um, training data? It didn't "figure it out", it was taught, just more quickly than humans, generally, can be.

[RichardT]

...without ever having been taught...

Um, training data? It didn't "figure it out", it was taught, just more quickly than humans, generally, can be.

Well obviously. I mean, it deduced how to write code just by reading lots of code. It wasn’t taught explicitly.

[BWC]

Yes, it was obvious what you meant, but I think it's an important distinction when discussing this subject.

[SafeandSound Mastering]

You control the contents of your own mind ultimately. You can choose to ignore this kind of thing totally if you wish and push yourself and your creativity and art and enjoy yourself as you did before it was here.

My job was supposed to be gone, did not happen, it's had 10 years now.

In a nutshell it's nowhere near as good a I am at mastering and does a 1/2 baked job of what the true scope of engineering tasks that comprise what mastering actually is. Not to mention it seems like a mere extention of DIY self finalizing.

I would just enjoy whatever it was you were doing before and maybe use it as inspiration to push yourself into a new area, new genre, new skillset, learn an actual instrument or a new one. Use it to inspire and motivate and push you forwards.