In article , Eric Stevens
wrote:

Nothing new there. It is well established that music can be created by
AI. What is lacking is the human nuance.

I doubt that AI will ever produce a Bach, Beethoven or Monteverdi.

there's no requirement that it should.

there are many more composers in this world than just those three.

Well, maybe it will, but it will be a long time.

where 'a long time' is 'now':

https://www.technologyreview.com/s/6...chine-listens-
to-bach-then-writes-its-own-music-in-the-same-style/
These guys have developed a neural network that has learned to
produce choral cantatas in the style of Bach. They call their machine
DeepBach (see also 3AI Songsmith Cranks Out Surprisingly Catchy
Tunes2).

You miss my point. I wasn't reffing to mere copying.

i didn't miss a thing and what they did was *not* copying, which means
*you* missed the point (as usual).

Take an example. You say "These guys have developed a neural network
that has learned to produce choral cantatas in the style of Bach." Now
neural networks learn. In this case it would have learned the style of
Bach. Let it loose and it will use it's learning to create more music
in the style of Bach. But how long would it take for it to create from
scratch a different and equally powerful style of music from scratch?
From what you have said it never would.

not what i said at all.

clearly you don't understand what ai can do and more importantly, what
it will be able to do (and sooner than you think).

teach it a different style and it will create something in that style.

one day, it won't need to be taught anything because it will have
already scanned and analyzed all existing music at that point in time
and will be able to create its own style, mimic an existing style, or
some combination.

All of those
composers broke new ground and have left a lasting impression. Thomas
Tallis is another.

ai composers will break new ground.

Some do it with an excavator. Others do it with a tea spoon. Bach,
Beethoven, Monteverdi and Thomas Tallis all made musical breakthroughs
the effects of which are still being felt today. There was no copy or
neural learning by these composers: they broke almost com[letely new
ground.

so what?

one day, ai will also break completely new ground.

your mistake is assuming ai will always be incapable.
that's very shortsighted.

you're also oblivious to what that could potentially become.

some people will even prefer what it creates over bach, beethoven, etc.

What's that got to do with it?

it has everything to do with it, because music is meant to be enjoyed.

your entire premise is that if it's not bach or beethoven, it's not
worthy.

that's pretentious garbage.

if a computer composes music that people enjoy listening to, then it's
a success.

Don't bother asking. It's only another
of your red herrings.

nope, and you meant don't bother *answering*, not asking, further
demonstrating your confusion.



And just consider what modern pop music has done with Bach.

what about it?

modern pop music is a totally different genre than classical.

That shows how much you know about these things.

which is *much* more than you.

you're stuck in outdated myths.