None of it matters anymore. None of it.

[RexrothPigeon]

Any more than I object to synths, creative use of auto tune, effects boxes drum machines - or any other music technology.

DON'T GET ME STARTED ON THAT DIABOLICAL PIANOFORTE!

 

[bungle]

What are we trying to accurately reproduce exactly?

You need to hear it, ”as AI intended”. For that in a near future you will need AI designed electronics, AI designed speakers, and finally AI driven precision analysis.

 

[Frank Dernie]

creative use of auto tune

This is perhaps something to give thanks for!
A lot of famouus vocalists can't sing in tune and their recordings are much better, in this respect at least, than a live show.

 

[bungle]

You need to hear it, ”as AI intended”. For that in a near future you will need AI designed electronics, AI designed speakers, and finally AI driven precision analysis.

It may turn out though that for AI listeners and people with AI chips planted to their heads, the boomy bass below 80 kHz may turn out to be boring, and we are going to see in general a high-pass filter set to around 100 kHz. There is just so much more to it. People who cannot afford it, they need to equalize the signals back to 20Hz and 20kHz range and compress the dynamic range heavily. Your brain can do it. It is just that mics on the left and right of your head are poorly engineered.

 

[JeremyFife]

You need to hear it, ”as AI intended”. For that in a near future you will need AI designed electronics, AI designed speakers, and finally AI driven precision analysis.

"As AI intended" ... 0Hz up into GHz, 300dB dynamic range, earth moving SPL

 

[mhardy6647]

This is relevant to the recording process, not the hifi reproduction process

Yes, exactly so.
This thread, unlike most at these fora, is about creation not reproduction.
Note the title and the OP.

... as such, it raises an interesting point: high fidelity to what? (Which, as an aside, was my father's complaint about the term high fidelity)
@amirm offered this rubric

Quiet background. Hearing every note as it decays into nothing.

Does a wholly artificial recording offer a realistic/relevant 'decays into nothing'? It may... but there are certainly wholly un-artificial recordings that will have different things going on in them. At the reproduction end of the chain, there may well be no meaningful difference... but the feedstream may an utterly different simulacrum.

 

[RexrothPigeon]

high fidelity to what?

To the source recording

 

[nerdstrike]

To the source recording

Pithy, but just so. Whether it's a 1920s telephone call, or a 2025 AI-generated Johnny Cash, you should at least hear the amplified signal without distorting it too much.

I get the apparent futility of fixating to the point of thousands of dollarpounds on playback of an artform when the content is potentially all "fake". Starting from the click track through autotune and beyond, you start to lose the human narrative that you can empathise with. We're so keen on cults of personality that I can't see AI-generated music ever going beyond random background stuff, unless we can successfully fake personality or have a true AI doing it. Either way I sincerely hope for a big pushback against AI everything in favour of live music full of variability and motivated by human experience.

Heck, AI-induced angst will probably inspire hundreds of musicians in new directions.

 

[Adis]

But... fidelity to what?
If a guitarist plays a guitar, and their fingers squeak on the strings... that is, for better or worse, part of the performance. The human part.

Squeak isn't music.

 

[respice finem]

Quiet background. Hearing every note as it decays into nothing.

But then, this "decay into nothing" is limited by background noise in the listening room, isn't it?
If I listen at, say, 96 dB peak (which is subjectively quite loud), and background noise is at 20-30 dB, then there are "only" 66 to 76 dB left, or am I "thinking it wrong"?

 

[bungle]

But then, this "decay into nothing" is limited by background noise in the listening room, isn't it?
If I listen at, say, 96 dB peak (which is subjectively quite loud), and background noise is at 20-30 dB, then there are "only" 66 to 76 dB left, or am I "thinking it wrong"?

You can still hear lower than noise floor, say 10 db, but your thinking is right.

 

[Leporello]

AI is to be welcomed with open arms. It liberates composers from composing and enables them to do something useful and productive with their lives. Think about guys like Bach and Beethoven, Instead of becoming successful influencers and content managers they had to endlessly toil away with their tiny notes, scribbling them laboriously on paper in candlelight. Tragic indeed.

 

[mhardy6647]

But then, this "decay into nothing" is limited by background noise in the listening room, isn't it?
If I listen at, say, 96 dB peak (which is subjectively quite loud), and background noise is at 20-30 dB, then there are "only" 66 to 76 dB left, or am I "thinking it wrong"?

Yup.
"Starts makin' vinyl and tape look more attractive, doesn't it?"


In seriousness, this is, I'd opine, a more interesting topic than it might seem at first blush.
@Adis absolutely true; indeed, my impression is that good guitarists don't squeak, but I suspect the truth is more subtle and complex.
Again in seriousness -- John Cage's 4'33" focuses on the non-musical part of performance, doesn't it?
Indeed, I wouldn't be at all surprised if there are AI versions of 4'33" floating around... I am loath to search, though.

 

[mhardy6647]

AI is to be welcomed with open arms. It liberates composers from composing and enables them to do something useful and productive with their lives. Think about guys like Bach and Beethoven, Instead of becoming successful influencers and content managers they had to endlessly toil away with their tiny notes, scribbling them laboriously on paper in candlelight. Tragic indeed.

J.S. Bach did do his part to maximize the dispersion of (his) genetic material...

Bach: The husband, the father, and the family man - Oxford Bach Soloists

Tomorrow is Father’s Day, so in this week’s newsletter we take a closer look at Bach the husband, the father, and the family man.

www.oxfordbachsoloists.com

 

[AndyMorris]

When we buy, acquire, or find our records, CD's, tapes, 78s or even downloads we are, in some small way putting ourselves out, spending time and possibly money, to acquire an artifact. We may treasure that artifact, or we may think "why the heck...", but it ours. It was made by someone, or a group of people, who put something of themselves into the making of that artifact. We have a connection to the people involved in the making of the artifact

When we stream, we own nothing, we have no commitment. We are merely Flanneurs, observing, momentarily the passage of sounds presented by an algorithm.

I want my playback kit to play those artifacts as faithfully as possible to what the originators listened to and said 'That will do'. But I do spend time and effort with the old records I buy, cleaning the sound up, getting the closest to what the originators heard. So cheap distorting kit is a distraction, good enough kit is invisible to my ageing ears and 'That will do'.

That said I might get a 1210 because its history resonates. It looks like it's a pleasure to use. I'll never scratch on it, but I'll use the 78 +16% to clean records, the detachable shell makes it easy to inspect and clean the needle, before and after I play all a bunch of old records.

Buying records is a commitment. Streaming is casual, AI is a robot.

 

[respice finem]

IMHO it's "a generation thing". The "cemetery blond" generation (me included) has a general problem to be paying without actually buying/owning. The younger generations less so, they only acknowledge it when the material they paid for is not available any more.

 

[Philbo King]

So much music today is generated by AI, does it really matter if your DAC and amplifier have an SNR of >120dB? On a whim, I went to Suno and typed in the keywords that describe my favorite music genre and what it created was surprisingly good and only took a few seconds. I'd dare say it was downright excellent compared to what I could do with real instruments and my gruff voice. The music was clear, detailed, articulate and believe-it-or-not enjoyable through my HD-650 headphones and DX5ii. If modern music is simply generated from hundreds of samples leached off YouTube, Spotify, Apple Music, and Pandora, I argue this whole high-fi thing no longer matters any more. None of it. What are we trying to accurately reproduce exactly? Heck, with modern DSP mediocre speakers cand sound pretty darn good so what exactly are we striving to achieve?

A little known secret: None of it ever did matter

 

[Timcognito]

And there is Floe where 2 flutes, 4 saxophones, and 2 french horns keep in time with 1 synthesizer in 1981 proving we can all get along.

 

[egellings]

Why lay the blame on AI?
Go back a bit more and perhaps lay the blame on an IC, or a transistor or a tube.
Maybe going even further back; you can lay the blame on the phonograph.

If the 'music' claims your soul or makes your foot tap, or you get an ear-worm; does it really matter whether the source is live, or from a studio, or using an nVidia chip?

Short of stopping to enjoy the 'music'; you can always stick to the 'music' you indicate you've enjoyed!

The devices (tubes, turntables, etc.) simply reproduce music; AI generates it. Whatever, if I like the tune, that's all that counts.

 

[mhardy6647]

One of my favorite albums, and I'll never be able to listen to it again without remembering that image. Thanks a lot! Not!

Yeah, sorry about that! I felt the same way... especially since it was a Klipsch Heresy!
I mean, if she were a Quad ESL-57... that'd have been OK.


PS Margo will always be at least an ESL-57 to me.