None of it matters anymore. None of it.

[Grimoire]

Who said 3D spatial audio wasn't a vinyl thing?

What goes around comes around...

 

[stalepie2]

Haha, and I did read them all this time (vanity will do that).

And as you probably guessed I thought my first version read too much like tech-splaining and didn’t convey my reaction to your post so well. Copilot did rather a lot of bullshitting there before eventually allowing that ASR content could very well be in the relevant training corpus?*

It doesn’t really get what stochastic means though (if you’re on an Apple device you can use the look up, otherwise I think it’s concise Oxford rather than Miriam-Webster in my region).

View attachment 508795

Yep. In other words I don’t really mean it in the sense of ‘random-ish’. Oh and stochastic doesn’t mean probabilistic at all, that’s a bit hopeless. I expect Copilot is tweaked to lean a bit toward avuncular in preference to technical. I would find that very frustrating.

*Not the words you just typed on ASR, but wherever the training corpus ended with. Meanwhile my phone really wants to say ‘tracing’ there ffs. I’ll correct it for the third time now …

"Stochastic" sounds right then, as in the computer science section of the Wikipedia article, third paragraph: "In artificial intelligence, stochastic programs work by using probabilistic methods to solve problems, as in simulated annealing, stochastic neural networks, stochastic optimization, genetic algorithms, and genetic programming. A problem itself may be stochastic as well, as in planning under uncertainty." It's all far over my head.

The lack of a sense of "I" or self in the chatbots I think is the biggest problem that people pick up on right away, especially during an argument. I pictured Copilot more as like a help desk secretary, if I had to visualize "her." I read a little about it, just glancing at Wikipedia a while back, and seeing it relied on ChatGPT, but I did not know how much further it was tweaked by Microsoft, like comparing Microsoft Edge which runs on chromium to Google Chrome.

I'd imagine it would apply to AI music too. A lack of continuance and personality from song to song, album to album, among so many other things. No life changes to go through. Or do AI S1M0Ne and Sharon Apple products go through drug-fueled binges and spaz out and have to be handled by PR agents, just like real pop stars?

 

[Snarfie]

One of my most cherished possessions...


(a carefully curated collection of dump finds )

Guilty pleasurs are ok

 

[wadude]

Squeak isn't music.

Tell that to any electric bass player using round wound strings playing with a pick while palm muting! Extraneous finger noise and clank is a major

Squeak isn't music.

It certainly can be.

 

[Adis]

Tell that to any electric bass player using round wound strings playing with a pick while palm muting! Extraneous finger noise and clank is a major

It certainly can be.

Say squeak in Solfeggio!

 

[Robin L]

Say squeak in Solfeggio!

My cat does. But it comes out in the form of a question, more like "Squeep?" and is unusually high-pitched. I guess she's a coloratura squeaker.

 

[rdenney]

I'm just coming upon this thread. Software makes it possible for us to lay down all sorts of music directly from digital sources without ever having been heard by ears except in the recorded form. For this, the recording is the performance, and is part of the art. But even if the recording process involved acoustic transducers (microphones) to translate an aural performance into an electrical signal (subject, perhaps but unnecessarily to my definitions, digitization and manipulation either before, during, or after that digitization).

The role of playback is to convert that electrical signal back to acoustic sound in a listening space, and the philosophy of such in these parts is to do so with the least possible change to the waveforms described by that electrical signal.

But the OP's question is philosophical: If music is becoming non-human, then what's the point of listening to it? Is AI-generated music creating an existential crises for the hobby of listening to music (and perfecting the ability to do so)? This is a much bigger question than how it might affect playback, because in my view it doesn't affect playback. It's also not a problem I face: there is so much recorded music available that covers human musical development and performance for the last half millennium, ready for us to continue to enjoy, and recorded well enough such that perfecting the playback system remains worthwhile. I just reorganized my CD's into a new cabinet for my modest collection of 350 or so--all of that still brings joy and not a bit of it was AI-generated. That's perhaps 0.01% of what's generally available as recorded music performed by real people, so I don't think anyone is faced with AI intruding on their audio hobby if they don't want it to.

Most AI-generated music that I've heard uses human performance and existing three-dimensional instruments as a source of samples from which to draw to assemble something that sounds original and realistic. This is similar to LLMs using previous writing as examples for what it generates--the real performance on real instruments using real performers is what trains the AI engine. Thus, there is some tenuously tortured indexical relationship between reality and AI-generated product, at least for now. Will I listen to music once that indexicality is no longer necessary? I don't rule it out, but I also don't have to care.

One of the Youtubers I follow is a machinist in northern Wisconsin who was recently forced to close down his machine-shop business because industry is abandoning his region. He is not a musician, but (laboriously) used an AI tool to create a song that would speak his story, the title of which was, as I recall, "Too Dumb To Quit." The voice sounds realistic, the tune is as original as any country tune is likely to be, and so on. The words are his. He is open that the song was developed using an AI tool. But it crystallizes the question for me of what constitutes "real" music versus "fake" music. The answer is that what makes music is the organization of sounds in ways that follow quite a broad set of rules and that have artistic intent. Additionally, it's art if someone listening to it accepts it as art (that was C. S. Lewis's definition of art with respect to literature, and it was intended to prevent academics from aggrandizing their tastes by undefining stuff they didn't like). What is fake is the performance--it was never performed as presented in a live space using real performers, and yet it sounds just like music that was performed in a live space using real performers. In that way, it's deceptive, but that doesn't invalidate it as music or even art. Music allows, it seems to me, the distinction between what is music from what is performance of music.

But once it is somehow recorded in a way that can be played back, it's the job of the playback system to do so faithfully.

Rick "finally filtered the latest acquisitions of 30 or 40 classical CD's into the alphabetized collection" Denney

 

[egellings]

For music, mostly because it's a bit of detective work to do it.

For certain styles I think it will soon become almost impossible to tell if it's AI created or not. Especially when it comes to generic types of music/sound, for example ambient elevator music.

Elevator music was often called "Beautiful Music".

 

[Grimoire]

Elevator music was often called "Beautiful Music".

Not by me!

 

[egellings]

Tell that to any electric bass player using round wound strings playing with a pick while palm muting! Extraneous finger noise and clank is a major

It certainly can be.

Git dat mouse!