192kHz considered harmful

[darrenyeats]

I could clearly pick out individual voices from the 30 or more singers, whether massed, lead or background singers. What did the CD lose that a high-rez would have revealed? Clearly and accurately presenting the voices of 30 people singing together would seem a good test for the ability of a recording to maintain clarity and not lose articulation.

It's hardly fair to ask people not to trust their ears and then provide arguments like this. Shame on you (just kidding!)

Wikipedia is written by nerds. Therefore: As long as something that can safely be ignored in practice, cannot be ignored in principle, Wikipedia will not ignore it.

As an example, see the article on Evolution; it actually even mentions Intelligent Design (in the last sentence, under "Social and cultural responses").

But would you say that about Wikipedia if Wikipedia had agreed with your own views?

I am not against your conclusions, necessarily.
Darren


Sent from my HTC Sensation Z710e using Tapatalk

[darrenyeats]

I have always wondered in general if not many "exotic" designs used by many audiophiles have unforeseen technical limitations that no normal engineer expects that can explain why some things are audible that should not be ?

I find that a very interesting point.
Darren


Sent from my HTC Sensation Z710e using Tapatalk

[mlsstl]

It's hardly fair to ask people not to trust their ears and then provide arguments like this. Shame on you (just kidding!)

My apologies to the group. I am thoroughly chastened. Is purchasing a album from HDTracks that I already have on CD sufficient penance?

[Mnyb]

My apologies to the group. I am thoroughly chastened. Is purchasing a album from HDTracks that I already have on CD sufficient penance?

It must be a 40 year old rock classic ripped from SACD and get the 192k version bonus point if they used thier cryogenic treated cables in the process ( that was not a joke they do that, amazing ?)

[Phil Leigh]

Well sorry to be playing devils advocate here.
But right now I read the argument as follows

a) there is a theorem which proves that under ideal (in the realworld not achievable) conditions a sampling frequency of x will be good enough
b) we use some totally different process to do the reconstruction but still refer to the theorem and claim that it still applies.

For me this sounds like comparing Apples with Bananas.

Or to put it differently:

Mr. Dan Lavry begins his document with the statement:

(sorry I didn't have time yet to read the paper to the end, will do it later).

And quite opposite to this statement a quote from Wikipedia about Nyquist:



So as the preconditions of the theorem cannot be met (according to Wikipedia, sorry I don't have better source nor knowledge) I think it is a valid question if we can overcome the (possible) deficiencies with higher sample rates (or other means).

So for me this part of the argument has its flaws (while of course I'm very happy to believe all the other proofs which include double blind tests)

It is trivial to show that the error inherent in the process is below the range of detection.
A computer-generated sine wave dataset (I.e. not recorded via an ADC) can be passed through a DAC and null-compared. I've done this and the result should be no surprise to anyone.
PS don't try this with a nos DAC!

The human ear/brain also operates on a sampling basis (there's nothing "analogue" about the ear by the way) and uses a reconstruction filter just like a DAC to integrate the discrete samples into something we can understand.

What some people are misunderstanding here is that it is the reconstruction filter that recovers the analogue signal, not the DAC. The DAC simply presents the filter with a set of voltages over time. It is within the filter that the sinc function becomes manifest and this is indeed an infinite series - the mathematical definition of a filter is a continuous function over time. A filter is not a step function!

To be clear on this, what comes out of the filter IS a mathematically perfect sine wave... All the way up to the Nyquist frequency. This is both predicted by the theorem and demonstrable in practice. There is no known way to differentiate between a 1khz sine wave sampled at 44.1 or 192. In every conceivable way of " measuring" or analysing that sine wave it they will be indistinguishable.

What is true for one sine wave is also of course true for any combination of sine wAves (or as we usually refer to it... Music).

Or do we need to have a conversation about Fourier transforms and the practical implications of mathematical infinite series as well?


Now there IS a way to break this model. Try a perfect square wave!. This has to be done using a mathetmatically generated waveform/data set because it is impossible to generate or record a perfect square wave with infinite rise/fall times in the analogue domain. The filter will introduce non-linear ringing.

This is all predict by the theorem because an infinitely fast slope requires an infinite number of samples... I'm sure you get the idea.

Bottom line is this; for real world sound distribution 44.1 is fine ... Which is why there are many many fine sounding red book CD's... And why there is no published evidence that stands scrutiny to support the idea that anyone can tell the difference between the same master distributed and played back at 192 or down sampled to 44.1.

[maggior]

Now there IS a way to break this model. Try a perfect square wave!. This has to be done using a mathetmatically generated waveform/data set because it is impossible to generate or record a perfect square wave with infinite rise/fall times in the analogue domain. The filter will introduce non-linear ringing.

This is all predict by the theorem because an infinitely fast slope requires an infinite number of samples... I'm sure you get the idea.

Just trying to educate myself here...

Does this mean that there could be audible distortion introduced due to ringing in a recording that has clipped waveforms? Taken to an extreme, clipping could start to approximate a square wave.

BTW, I'm finding this discussion to be very interesting...I'm learning a lot.

[darrenyeats]

My apologies to the group. I am thoroughly chastened. Is purchasing a album from HDTracks that I already have on CD sufficient penance?

That more than covers it!

Sent from my HTC Sensation Z710e using Tapatalk

[Mnyb]

Hmm thinking about it there can not really be any invalid combinations of samples ?

Whatever gets coded inside the 16/44.1 code space or whatever must be recoverable you might not want to listen to it and the square wave is approximate as you can't have infinite slope and it do ring .
But afaik the ringing are above >20kHz .

[DaveWr]

We do have to distinguish properly between CD at 44.1k sampling with 16bits and 44.1k - 24 bits. The 24 bits do have an effect. They allow a significantly greater dynamic range, assuming the mastering engineer does't get into loudness wars compression.

It may be for some people that they compare 96/24 with CD and it is the 24bit resolution not the sampling frequency differences that are being heard.

Also if its old music, my late 60s and 70s youth, any new releases will be from analogue tape remastered. This remastering may (and I know several cases of did) produce entirely different mix from early CD versions. IMHO not always better.

Phil Leigh would have more input on these recording and mastering issues.

Today with the freedom to use anti-aliasing and reconstruction filters that are low order and relatively benign, due to very high frequency oversampling techniques, I believe we have an almost untainted record / replay chain.

Dave

[Soulkeeper]

The article linked in the OP covers bit depth too.

The incorrect '96dB' figure ignores the spectral power density of a signal. 16 bit audio can go considerably deeper than 96dB, and deeper yet with proper use of dither. Handled correctly, the dynamic range of 16 bit audio reaches 120dB in practice [10], more than twenty times deeper than the 96dB claim.

That's greater than the difference between a mosquito somewhere in the same room and a jackhammer a foot away.... or the difference between a deserted 'soundproof' room and a sound loud enough to cause hearing damage in seconds.

16 bits is enough to store all we can hear, and will always be enough.