Steven Wilson CD vs. High-Res: If SW can't tell the difference...

[humprof]

Saw this posted on a discussion thread I sometimes eavesdrop on. Clickbait, sort of, but I sympathize with SW's condition: 15 years ago, I took a test that NPR or somebody put online, with several sound samples available in different resolutions; then, in a blind test, I could tell the difference between 44.1/16 and 96/24 80% of the time. Now, I'm pretty sure I wouldn't hear it.

https://www.headphonesty.com/2024/09/atmos-producer-admits-difference-cds-high-res/

 

[LTW]

No one can tell. It's placebo or the 'concentration effect' (whatever it's called) where you expect a difference so you listen harder and hear 'new details'. There I said it. Keep it civil

I like how Steven is honest about this, especially when he's part of an industry that profits off snake-oil and rampant misinformation.

 

[Thonolan]

No one can tell. It's placebo or the 'concentration effect' (whatever it's called) where you expect a difference so you listen harder and hear 'new details'.

I agree wholeheartedly.

 

[ar surround]

Hearing, eyesight, etc, all goes to heck when we get old. So I’m not surprised by SW’s observations. Plus, he plays rock venues which are rough on the ears by their nature.

 

[minimumtumbleweed]

Makes me sad though; the day will come when we no longer have SW mixes to look forward to. Reviews of Alan Parsons' recent offerings were discussing how his mixing choices might be coloured by hearing loss. It happens to us all I guess.

 

[jimfisheye]

Too many variables!

The right dynamic recording might sound cleaner in 24 bit vs reduced to 16 if all other things are equal. An older DA converter with less than great anti aliasing filters might sound cleaner in HD than SD.

Other variables usually take over by a magnitude. Like bold compressor and eq moves! Most of these formats - even older tape and vinyl - are pretty robust. I can find an example of a vinyl picture disc with the cleanest sound and what sounds like the least generational loss vs a black vinyl copy, CD copy, and bluray 24/96 copy. Is vinyl better and picture disc vinyl even better than that? Nope! Worst of show. So other variables usually have greater effect.

The point is this isn't a shootout between two formats with a big divide live black and white vs color TV. But it's also not always zero and pure perception bias! Edge cases with generational loss from different formats adding together can come up sometimes too.

I think calling out CD as not being the lossy variable you thought it was is correct more than not too. But I also think 24 bit is pretty proven to eliminate even the weirdest edge cases and everything should just continue to move forward. Some of this is an attempt to give the benefit of doubt to the industry and blame a mastering mistake on a format instead of a human being with feelings too! But the unfortunate truth is there are botched masters released and it was never any shortcoming with any format. Someone just good old fashioned screwed up!

The overall sound of Wilson's mixes in general makes me think his hearing is still very on point, FWIW! He calls out the treble boosted stuff from the guys with apparent hearing loss.

 

[minimumtumbleweed]

The overall sound of Wilson's mixes in general makes me think his hearing is still very on point, FWIW! He calls out the treble boosted stuff from the guys with apparent hearing loss.

Something tells me that despite being a "rock star" he's probably very careful with hearing protection.

 

[Mr. Afternoon]

To me the benefit of hi-res is not necessarily listening but mixing/production. There are a lot of processing effects that behave very differently at higher sample rates, especially ones that distort, destroy, or transform the audio in some way.

 

[jimfisheye]

The perception bias range.

When you try to listen to a variation (eg. a tiny level change or tiny eq move) that's so subtle it falls into perception bias range. Where your perception can be tricked one way or the other.

Mute the lead vocal track in a mix.
We're all hearing an instrumental mix with no voice now, right? No one wants to argue that they aren't sure, right? Pretty obvious?
OK, good.

Now just make a 0.5db eq change in the voice at 273Hz. Now call that out in an A/B! Be quick! You think that's funny, right? Now what if I tell you I had the eq bypassed the whole time! We've all done this, right? Some of you are lying but the rest of us have all done this.

 

[jimfisheye]

To me the benefit of hi-res is not necessarily listening but mixing/production. There are a lot of processing effects that behave very differently at higher sample rates, especially ones that distort, destroy, or transform the audio in some way.

You have to really watch your levels recording with a 16 bit device! Sometimes erring on the side of a few clips or you end up with a mostly 8 bit recording. And you sure can hear quality loss in an 8 bit recording!

Meanwhile, record to 24 bit and you can screw up and have your peak level on an input at -48db and still have a 16 bit recording to show for it. That same level faux pas leaves you with an 8 bit recording with a 16 bit device.

So when we compress something to the moon and back with those kind of gain moves during the mix - as we sometimes do - it can similarly make a difference.

 

[jaap74]

No one can tell.

Ok debate closed..............

 

[JediJoker]

It's placebo or the 'concentration effect' (whatever it's called) where you expect a difference so you listen harder and hear 'new details'.

The term you're looking for is expectation bias. The audiophile industry is built upon it, to a great extent.

 

[barfle]

The study I’m familiar with was done by Dr. Mark Waldrep, “Dr. AIX,” who produced a lot of 24/96k DVD-As.

He took a handful of his hi-res files, reduced their resolution to CD quality (16/44.1k), then padded the files to return them to their original size.

He had a lot of volunteers download those files and tell him which pnes were high-res and which ones were CD quality, and the overall results were no better than random guessing.

I don’t think that quite equates to the double-blind testing that drugs go through, but it’s pretty telling of how so many of us really can’t hear any difference beyond CD quality. I know I can’t, and I admit it. Of course, I can hear the difference between multi-channel and stereo, which validates the use of higher bit-rate media than the stereo-limited CD (yes, I know the original redbook allowed for quad, and that there are a handful of matrix CDs out there).

I’ll buy 24/96k recordings, though. I think they probably have a more careful production than CDs, but there’s no guarantee. A higher bit-rate also means that any post-production will have less effect on the resolution of the finished product (just a minor bump in level pretty much wipes out the two least-significant bits, and there are a lot more things that can be done than volume tweaks).

 

[jimfisheye]

I can tell you about perception bias. I think this is what is trying to be described. We filter our sensory input through perception. Observing raw visual or auditory input is actually difficult. That sounds absurd on the surface... Think about stereo perception and depth perception in both sight and hearing. The brain stitches the two pickups together and works off the timing differences and all that.

Audio gets into a lot of human cues. Brighter and more dynamic transient sound means close proximity sound. The high end and pointed dynamics diminish as something moves farther away. Your hearing can adjust to background noise and "work" the perceived volume level you hear just like your eyes adjusting to light. We need meters to reveal absolute values.

Where I'm going with all this is there's a range of subtle where it's hard to quality what you are hearing vs your perception skewing it. I can hear if the vocal track is muted and the song is instruments only no matter what I'm doing. Can I tell if there's a 0.5db .2 octave eq cut at 300Hz vs not at a glance? Maybe... But I can also trick myself into hearing a small change like that. Turning my head to the side might even sound similar. What to trust?

Have you ever worked with audio and were auditioning a subtle eq setting? A/B between a small cut or boost and flat. Spend a good 10 minutes and come to what you feel is a solid decision. Then you look and find you had the thing bypassed the whole time and it was flat the whole time!

Maybe you hear the mix you made the next day and some subtle work you did adds up to the whole sounding 'right'. Maybe if you had skipped doing whatever, the mix would be too bright or too muddy overall. Or maybe it was fussing over some move that can be proven to be meaningless at the end of the day. This conversation is pure subjective and could spiral out of control quickly! The only point is perception bias - or whatever you want to call it - is a very real human condition. And why we make tools to help tell absolute values of things.

 

[seremetcc]

Sample rate (e.g., 44.1k, 96k) and bit depth (e.g., 16, 24), not bit rate.

 

[jimfisheye]

The real answer:

So many novelty volume war masterings were put to CD just because of the time frame and what was in vogue that many people associate that sound of brick wall limiting and bright eq boost with the CD format itself.

 

[stoopid]

I'll attach the older study that effectively concluded that the benefit/detail from higher resolution audio is inaudible to most people.

It's been a long standing seesaw discussion on audio forums, pretty much since the first DVD-A were released and higher resolution formats were attainable to the mainstream music listener. There's been no progress made by the pro-high resolution camp in all those years, simply people claiming 'they can hear the difference'. Often the exact same people spending $150 on an audio cable or $5000 on a DAC. Anyway, you can guess where I stand on the topic.

This topic sits quite literally locked arms at the heart of the whole 'audiofool' economy.

https://cdn2.imagearchive.com/quadr...bility-of-a-CD-Standard-v-High-Resolution.pdf

 

[stoopid]

To me the benefit of hi-res is not necessarily listening but mixing/production. There are a lot of processing effects that behave very differently at higher sample rates, especially ones that distort, destroy, or transform the audio in some way.

This is 100% true. 24bit at 44.1Khz (or higher) is a must for the music producer. But it's more for the mechanics of the thing being done as you say, versus having to render the final mix at a higher resolution in order for it to be palatable. It's this reality distorted through the lens of the uninformed that generates enough of the confusion that perpetuates this discussion for all these years.

 

[KG10]

No one can tell. It's placebo or the 'concentration effect' (whatever it's called) where you expect a difference so you listen harder and hear 'new details'. There I said it. Keep it civil

I like how Steven is honest about this, especially when he's part of an industry that profits off snake-oil and rampant misinformation.

You might find this press release about the foo fighters commitment to hi res from 2015 amusing

https://web.archive.org/web/2015071...ner-to-promote-hi-res-audio-across-the-globe/

 

[jimfisheye]

Before it was possible to reproduce sound with a virtual reality realism, early attempts had a "signature" to their sound caused by artifacts of the system. And caused by artistic use of the system that would try to give the vibe of unachievable realism in artistic ways.

Using green and purple makeup in a black and white movie production because it made more realism looking contrast. No one was thinking they actually wanted to see purple in that example. Early recordings had similar artistic approaches to get across touches of realism that was technically beyond the tech.

So that all gets unceremoniously dumped now that we can actually simply reproduce full sound. You can just buy clean devices and clean recordings now. We're done here. I think some people are still pining for some artistic magic that leads to hidden treasure. Draw a marker around the edge of your CD and reveal the hidden fidelity! Like you're unlocking a hidden bonus track or something. 96k is bigger than 44.1k... Oh hey, I'm gonna get to hear something special for sure!

Capturing stray frequencies beyond the range of hearing turns out to not do anything for anyone and makes an easy target for ridicule. But it's also a strawman for what someone actually heard a problem with - a poor analog crossover eq that was being used for an anti aliasing filter cutting into the audio band. A digital system will make a mighty artifact if you allow aliasing! So something had to be done. A big ol' wide margin above the audio band is still a good passive solution to that. (ie. what 96k gives you) Doing that on playback by upsampling in an AVR is just as valid but not as sexy to market apparently.