The 48KHz vs 96/192/etc Khz debate

I'm not the biggest fan of science, ok? One must trust themselves. Either you hear it or you don't. No harm, no foul either way. I told you my proof. I had no idea he was going to make a change. I was terribly upset for a week or more. It was so obvious. And for whom would I be doing a rigorous blindfold test for? You?

Music is an art that depends on aesthetic choices, decisions and taste. I inveigh mightily against putting "science" ahead of my ability to hear and make those decisions on my own

Listen, I'm sorry but I've been in these debates FOR YEARS. Please, I'm not interested. I'm just here to state my opinion and experience. The difference between 96 and 48 isn't even close to me. If it is to you, great. I think there possibly is more than meets he eye in human perception than mere numbers and null tests.
 
Just to clarify my last post. Science is great. But I think we're going down a weird alley where science is the new god. It is replacing ones ability to simply know. I believe in my ability to know. I have great ears. I don't need anyone to tell me I don't. There isn't any test that will invalidate my ability, unless I let it. Art is about achieving a certain knowledge, ability and confidence in oneself. It's absolutely ridiculous that I heard no difference. I had NO IDEA that any change had taken lace or who'd take place. And I've done many, many recordings, mixes, masters at higher and lower rates.
 
BTW...


Science.. or, for the intent of what I am presenting in this post, scientific method, provides for whittling away at those "things" that can disprove something. Scientific method does not prove anything; it is a method of disproving. There is a difference between something that can become a theorem, and something that will (in all likelihood) remain a theory for a long, long time to come. I would provide an example of the latter, but it would likely ignite a "debate war" that is inappropriate for this forum. However, once something is a theorem, the likelihood of it ever being disproved are beyond astronomical. I would state that the likelihood is an absolute zero chance, but people continue to debate that statement (look for Fermat for one), so I will leave it as.. a theory ;)

Point is, "sciences" are evolving entities. As time marches on and we progress technologically, and build and refine our knowledge, "theories" will evolve with that. A simple example would be somebody stating that they have a theory that "2+2=4." Someone could very easily provide a proof" to elevate that to theorem status, but it is inherently, obviously so.. thus, no theorem/effort is required. Basically speaking. Usually, a theorem will be mathematically based. It is describing something.

In this case, it is not a matter of Nyquist/Shannon being in doubt, or worthy of any great investigation; anyone is welcome to do so, of course. Rather, the area of debate should be residing where it properly belongs: application. In order to apply the theorem as a method of audio capture and playback, for instance, one must also therefore consider the intended beneficiary of such application.. humans. Which is why we utilize the theorem as a means of determining application. In our case, we feel that limiting our method to our theorized(!!!) hearing experience range of 20-20kHz (I say theorized only because it has not been proven that we all are only capable of "experiencing" aural waveform data within that range.. more or less ;) ) is well founded. However, experience is telling us otherwise. Or, is it? Is our mechanical application (ADC, DAC.. filtering, etc) where a failing can be found? Where is this failing? Obviously, a significant portion of the population is capable of "true double blind" identification (as viable as that sort of thing is ;) ).

Once again.. it is not the theorem which should receive attention, but rather the assumptions regarding human aural experience, and the equipment being utilized. The math (theorem) stands, while the science (an evolving set of entities!!!) is about discovering, improving, etc. Scientific method has allowed us to attain what we have, and allows us to continually improve. Without that, we are nothing more than stacks of charred mice.
 
Non techie layman's terms.
Will someone with an iPhone and earbuds be able to tell the difference?
no.. I think, if you can afford to record at higher rates with hardware not hassling you, why not.
But what you use to downsample to whatever format might be the culprit of lost magic.
 
OK, you more or less say this above, but I'll clarify: Nyquist is not a scientific theorem, it is a proven mathematical fact. It cannot be challenged, any more than "1 + 1 = 2" or "Pi is irrational" can be challenged. They are fully self-contained, complete and closed off.

The technical implementation of the Nyquist theorem into usable audio technology however remains a continuing area of scientific endeavor, and will probably remain so for quite a few years yet.

Now Nikki, the following is said with all the respect I can muster, you're one of the smartest folks around here and I hold you in VERY high regard. :)

But your statement "Obviously, a significant portion of the population is capable of "true double blind" identification (as viable as that sort of thing is ). " - um, OK, please name one example. Every study I have ever examined flat-out contradicts this statement.

As a genuine skeptic, I'll guarantee that as soon as any evidence is actually gathered that proves that (under ABX double-blind conditions) people can actually tell the difference, I will gladly retract EVERYTHING I have ever said on this topic. But although I've seen people make reference to to "all the studies that prove people can do tell the difference", no-one has ever been able to direct me to one of these studies.

Give me the evidence and I'll change my attitude immediately. But until that happens it's a little like those references certain people make to "all the studies that prove evolution is wrong" - given how important these studies are and how world-changing the consequences would be, you'd think someone would be able to show me one.

The Upshot - it is genuinely great fun having this discussion with people like Nikki and Henry (et al) and I am grateful. No harm in disagreeing, as long as no-one takes it too seriously. Meanwhile, nearby where I'm sitting I can hear people discussing a reality TV show like it actually matters. Ugh.
 
@Manning: Absolutely, and my bad, completely, for not properly qualifying my statement. With a set of Weiss converters in a pristine environment, the bulk of humans would be unable to unerringly differentiate between a source played back @ 48k vs 96k. My statement was a "mish-mash" of thoughts that collided, lol. It was the ability of any person being capable of doing a double-blind, and that most would cite a difference between 48k and 96k with (typical) con/pro -sumer DACs, and thoughts of the Weiss example I gave.

BTW- I had the privilege of experiencing Weiss converters one time only. Glorious is one of so many words to describe the experience. For me, every other ADC and DAC has been a matter of choosing which color allows for a proper reference, and/or *distorts* my audio in a pleasing manner for me. :D Oh- and external clocks? Bad. Inevitable in a multi-unit environment. But, so long as the internal clock is not severely flawed, an external clock will do nothing more than provide additional *coloring*. Whether that is good or bad is all in the ears of the beholder :D
 
Manning said:
Even Fractal land has been touched by the outrageously heated debate going on in the recording world over what frequency digital audio should be sampled at.

On one side of the debate are the audio hardware (including hifi hardware) manufacturers claiming "more is better". Their argument seems to make sense - we are sampling an analog waveform, so surely the higher the rate we sample at, the more accurate we will get, right?

On the other side are a group of scientists and engineers who assert that the above argument is complete and utter bollocks, and betrays a complete lack of understanding on how sampling actually works. Here's an excellent presentation by one of them:

24/192 Music Downloads are very silly indeed.

Make sure you seriously get your geek on before reading the above link.

For the record your Axe runs at 48KHz. Coldplay just released their most recent album after recording it at 48KHz. Lots of people (like myself) refuse to record at anything above 48K.
Click to expand...

A must read for those who are not afraid to study ;)
Thank you from me and from all those who learned a lot that article!

I have a musician's question to the author and other experienced studio folks.

Here is my situation. I'm trying to analyze a fast bass passage in a pop song. It's so fast, so processed and hard to hear in the mix that I need to slow the record down to 25% of the original speed, and also adjust the pitch an octave higher to hear better what the bassist played.
I have just that short passage sampled at 24/192K from vinyl, and as I import it to my DAW project (the project is set to 24/48K) without converting to 48K, it will be already playing at 25% of the original speed without any processing, and all I need is just to raise the pitch (which, I guess, should require 24/48K for the best processing results).
Now, the question is. Is there a way to get the same best results in a DAW if the file was recorded from vinyl at 24/48K? What is the best way?
And yet another question - if I wanted to make my own sound samples by slowing down existing samples and/or raising their pitch - would it be better to have the sourse samples recorded at 24/192? Or not really?
 
Last edited:
Hmmm...
Don't forget - just a few years ago (well decades) those same guys were telling us 16 bit, 44.1 k is as good as mere humans would ever want.
Forget maths, forget our current understanding of audio technology - go get a beautiful piano, a talented vocalist, and put them in a good room with some six thousand dollar German microphones - record them with a quality desk using well specified external preamps at 96 and 48.

Now listen through a very good hifi, headphones, or whatever as long as it's quality.

The 96 k recording will sound more present and ...well... Better! It's hard to put into words but it's real and repeatable.

Anyway - that should get some squawking :)

Pauly
 
iirc correctly, sample rate and bit depth choices really depend upon what you're doing..

if you are listening to pre-recorded audio, like popping on a CD, you shouldn't be able to hear the difference between 16-bit / 44.1k and anything higher because it's beyond human perception..
however, if you are manipulating the source audio [running it through various plug-ins <EQ, compression, various fx etc> in a DAW for example] then it makes a difference because the aliasing at higher resolutions becomes so small as to be negligible..

our beloved lil' black box is [just like a DAW] manipulating the audio stream with fx..
so it makes sense for it to run at a resolution higher than 16 / 44.1

so the reasoning is, what are you doing to the audio? manipulating it? or simply playing it back?
so when it comes to the 'high resolution audio being best' debate
for one bunch of guys, higher resolution is necessary
for the other bunch, it's snake oil
 
Some handpicked CD's could remind you of a master-tape being played on a good system. And it's still 16 / 44.1.
But the take and mix-down has probably been made in a much higher resolution.
If the Axe was processing in 16-bit like the venerable G-System I think Cliff had hit the wall before version 2.0001.
 
There is a huge difference in audio quality from 44 to 192, we record in 192 because we believe its the best audio quality and the best way to capture the subtle richness of different analog preamps we use, the sound quality between 44 and 48 is nothing compared to 192 or even 96. If you guys want to get all technical that is great, but we drive our choices by ear and 192 just sounds better for us.
 
I read something somewhere once that said that anything played back beyond the resolution of 16 bit / 44.1k is beyond human perception..
and if someone can actually hear a difference between one recording at 16 / 44 and 24 / 96, they're most likely hearing something else..
like the equipment itself or how the original audio was recorded / processed..
 
The key-word here is processing a signal through a vast amount of digital algorithms - in the case of the Axe a signal from a guitar.
I don't think it would sound more "natural" if the output was 64 bit /3.200 kHz
 
clarky said:
I read something somewhere once that said that anything played back beyond the resolution of 16 bit / 44.1k is beyond human perception..
and if someone can actually hear a difference between one recording at 16 / 44 and 24 / 96, they're most likely hearing something else..
like the equipment itself or how the original audio was recorded / processed..
Click to expand...

ProTools has a great feature called HEAT, when enabled it makes digital audio sound closer to analog, we did this A/B and effectible it does close the gap a lot between Digital and analog, still analog has something.... we are doing this test in top notch studio gear.
 
If you are using a DAW and your gear can handle it, recording at a higher resolution is advantageous for this reason: processing and summing. that is to say dither. The lower your resolution (sample rate), the more artifacts you will get every time yo do a process (compression, eq, effects, summing signals, changing sample rate) and the more they will happen in the audio range. Keep it high when the DAW is crunching numbers and then dither it down at the end.

I'm with Clarky. While I am happy to believe Henry heard a difference when the sample rate switched, i dont think necessarily shows that it is the sample rate itself that caused the change. Seems more likely to me that it was an artifact of the gear, or (most likely to me) caused by the process of downsampling itself.

I want to follow the links given upthread a bit later on when I have time, but in the mean time here is a favourite vid of mine on the subject.
 
nicolasrivera said:
ProTools has a great feature called HEAT, when enabled it makes digital audio sound closer to analog, we did this A/B and effectible it does close the gap a lot between Digital and analog, still analog has something.... we are doing this test in top notch studio gear.
Click to expand...

I got a great piece of kit that makes a digital signal sound analogue…
you'll no doubt be familiar with it.. 19" rack mounting.. black with a green screen.. : )

seriously though.. there are lots of plug-in / AU things out there that recreate that analogue warmth..
even tape saturation..
essentially.. your digital signal has a pile of math thrown at it so that it emulates that characteristic that we perceive as analogue..
with HEAT, your signal is still in the digital domain.. nothing has changed.. apart from it being run though an effect that maybe does some EQ stuff and maybe adds / excites [or softens] for example some harmonics in clever and cool ways..

the debate here though is about bit depth and sample rate…
the question is usually why have the choice if you can't hear the difference..
and from what I understand, the answer depends if you're simply listening to the final piece of audio, or are you processing it..
if you're listening to the finished piece, you shouldn't be able to tell the difference once you get beyond 12 bit / 44.1k
if you are processing the audio, you'll start introducing artefacts at lower resolutions.. some of which may actually be audible..
but if you're processing the audio [applying EQ, compression, reverb, and other effects etc] at a higher resolution [like 24 / 48 in out lil' box.. and beyond]..
the errors introduced are so minute that to the human ear, they simply don't exist..

I have sample packs that are designed to run at 24 / 192 that I use on the movie music I work on..
when I run them down at 24 / 48 I cannot perceive any difference at all.. the same source audio through the same monitoring system.. different sample rate..
that said.. the movie guys run at these very high resolutions to make any chance of unwanted / perceivable crap being introduced pretty impossible..
I guess the thing is that in a modern theatre, you're hearing the audio pretty loud through a very high spec sound system..
I'd imagine this'd put the audio under the microscope some and so if there were any warts in there, the likelihood of them being exposed is higher..

interestingly though… although I know the studios mix and master at 24 / 192 for movies..
I'd be interested to find out what sample rate they bounce down too [if they do that at all]
 
Here's another factor. Cheap converters may sound better at higher frequencies because they use cheap ant-aliasing filters that compromise frequency response at or near the cutoff point 22.05KHz. for 44.1KHz. sample rate. 22.05 is darn close to 20KHz. and some humans can actually hear as high as 20KHz.
When you run them at higher frequencies, that compromising takes place outside (above) the audio spectrum. A quality converter is able to filter unwanted frequencies without affecting the audio frequencies even at clock speed 44.1KHz. because they have better anti-aliasing filters.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom