[stv014]

  http://artsites.ucsc.edu/ems/music/tech_background/te-03/teces_03.html
 
"Height information is provided by the shape of our ears. If a sound of fairly high frequency arrives from the front, a small amount of energy is reflected from the back edge of the ear lobe. This reflection is out of phase for one specific frequency, so a notch is produced in the spectrum. The elongated shape of the lobe causes the notch frequency to vary with the vertical angle of incidence, and we can interpret that effect as height. Height detection is not good for sounds originating to the side or back, or lacking high frequency content."

 
Indeed, the graphs below (generated from old KEMAR HRTF files without equalization) show how the HRTF changes as the elevation is increased from -30 to +30 degrees in 10 degree steps. The azimuth (horizontal direction) is fixed at 30 degrees to the right. The differences are mostly in the notches above 7 kHz.
 
       

       
From the above, it looks like moving a narrow notch (or comb filter) between about 7-10 kHz can give the illusion of vertical movement. I did try this in the past with headphones, and the effect does work to some extent.

 

[stv014]

  Rise time of redbook  is approx 14 microseconds - or bandwitdh to 20 kHz. IF even I accept that this is "enough" - that means that this has to be maintained from input to output -from the sound waves impigning on the diaphragm of the microphone to the sound impigning on the eardrum - with every component in between fast enough to maintain that 14 uS risetime even under worst of conditions. In series as they are, that means each and every component does add so and so much delay/filtering - that means even the best equipment available does nor find maintaining that 14 uS easy - OVERALL. 
 
There is no (pre)amplifier with infinitely short risetime or infinity bandwidth, no cable, and most certainly no microphone. Each and every component is slowing the original sound down somewhat - and this can CLEARLY be heard in redbook vs "higer speed" ( be it analog or digital ) : redbook is mostly flat surface as far as depth is concerned, under best of circumstances it has some depth that is definitely limited and does not match the same as analog or hirez digital and most certainly not live feed from the microphone.

 
I thought it was already extensively discussed that Red Book does not limit phase or delay to any large discrete "steps", and that faster rise time than what is possible with Red Book mathematically requires content above 22.05 kHz. Any delay or filtering added by equipment is irrelevant as long as it is inaudible and close enough to be the same on both channels. Here you can see that even a cheap sub-$100 sound card can achieve less than 0.1 dB roll-off up to 20 kHz, less than 50 ns inter-channel delay (vs. 10 us threshold of audibility), and less than 0.1/0.02 ms group delay variation up to 20/17 kHz, respectively (vs. ~1 ms audible level in the mid-range). Even the group delay is only due to the use of a minimum(ish) phase reconstruction filter, with linear phase it would be much less.
 

Originally Posted by analogsurviver /img/forum/go_quote.gif
 
The most significantly audible jump in quality in digital occurs from redbook to 88.2 or 96 kHz sampling frequency - this IS "night and day".

 
It should be easy to prove it then with some 20/20 ABX logs. Just take some 88.2 or 96 kHz sample, convert it to CD quality with a decent resampler, and then back to the original format, and compare the files.

 

[stv014]

Originally Posted by analogsurviver /img/forum/go_quote.gif
 
Computer audio is not far enough to allow decent ABX - in native DSD. The moment we are forced to go into PCM for the ABX software(s) to be enabled, much of the difference has already been thrown away.

 
You have just said that the largest, "night and day" step is between Red Book and 96 kHz. If you have equipment that is capable of playing the latter, and you hear the improvement it makes, then you should be able to perform a 96 kHz vs. 96->44.1->96 converted sample comparison without problems. If you are worried about the conversion degrading the sound, that should just make it easier, since the 96 kHz original is not processed in any way.

 

[diamondears]

Are you sure it's not just 2 different masters/source you are comparing? Is the PCM version deived from the same DSD source?

Have you even compared a red book version between DSD128 or more of the same Master recording? The difference is so obvious I'm extremely doubting the intentions of doubters here.

The answer to your question is yes.

 

[stv014]

The answer to your question is yes.

 
Is that based on actually analyzing the samples (converting both to e.g. 176.4/24 PCM format, and subtracting the Red Book version should result in no difference other than the higher quantization noise of the CD, and the ultrasonic content of the DSD), or do you just believe it is the case ? Do you even know if the playback levels are exactly the same ?

 

[diamondears]

Is that based on actually analyzing the samples (converting both to e.g. 176.4/24 PCM format, and subtracting the Red Book version should result in no difference other than the higher quantization noise of the CD, and the ultrasonic content of the DSD), or do you just believe it is the case ? Do you even know if the playback levels are exactly the same ?

Ok. I retract. I now believe DSD ain't improving nothing as soon as I read the words "analyzing samples", "converting to etc after subtracting red book", "higher quantization noise", "ultrasonic content", "playback levels", etc. I never do that to a layman (relative to my profession), unless I have nothing to impress but use technical terms. Or confuse unknowledgeable and blind bats...

Is it possible for record labels to just sell the Master recordings themselves? In this case, everything would be the same. Oh no...quantization, recording levels, and time itself would I guess be different...

So, satisfying all your conditions for a proper A-B testing, you mean you don't find SUBSTANTIAL improvement in DSD128 vs. red book?

 

[bfreedma]

Ok. I retract. I now believe DSD ain't improving nothing as soon as I read the words "analyzing samples", "converting to etc after subtracting red book", "higher quantization noise", "ultrasonic content", "playback levels", etc. I never do that to a layman (relative to my profession), unless I have nothing to impress but use technical terms. Or confuse unknowledgeable and blind bats...

Is it possible for record labels to just sell the Master recordings themselves? In this case, everything would be the same. Oh no...quantization, recording levels, and time itself would I guess be different...

So, satisfying all your conditions for a proper A-B testing, you mean you don't find SUBSTANTIAL improvement in DSD128 vs. red book?

 
You do realize this is the "Sound Science" section of Head-Fi?
 
Subjective opinions without proper analysis or control will always be heavily scrutinized.  You really should perform a proper ABX test - you will probably find the results surprising.

 

[analogsurviver]

   
It is a common belief that Redbook (or any digital chain) is limited in time resolution to the time represented by the time between samples. This is incorrect. 16/44.1 digital can resolve time differences into the nanosecond range, orders of magnitude less than the time between two samples. If you doubt, I can refer you to videos and papers proving it.
 
Some very early CD players had a phase shift between channels due to sharing a single DAC between channels due to cost. In one I have seen the test results for, the difference amounted to 22 degrees phase shift at 20 KHz. All players and DACs that you can buy since the 90s have one DAC per channel. Phase shift is essentially zero.
 
Finally, rise time and frequency response are one and the same thing, viewed in different domains (time versus frequency).

Please do refer me to the videos and papers proving that 16/44.1 can resolve time differences into the nanosecond range.
 
That regarding timeline of (non)availability of "digital" sharing a single DAC with consequent phase shift "since 90s" does not hold true. Some computers, notebooks, netbooks etc, as well as external DACs, have been available as recently as 2009 - if not longer. With these, reduction of phase shift is clearly audible AND visible on the oscilloscope by the use of higher than 44.1 kHz sampling frequency in any software, foobar2000 being the most commonly known and used. Audible difference lies primarily in imaging : soundstage width ( it becomes wider/broader ) and soundstage depth ( it starts to show some vestiges of depth, it is no longer glass pane flat ). Since such "digital whatevers" are usually limited to sampling frequency of 48 kHz, the shape of the output signal (square wave, frequency response etc ) will be exactly the same as if when run with the 44.1/48 kHz "sampling" in software; only phase difference between channels does get reduced by increasing the sampling frequency in the software, it can not increase the resolution which is limited by the hardware. And this most definitely IS audible - if one's laptop/netbook is not exactly young, worth trying out if it does not hide inside a single DAC ...
 
Yes, rise time and frequency response is the one and same thing viewed in different domains.
 
However, "ringing" ( the correct consequence of lacking high frequencies ) in 44.1/16 > 88.2 >etc compared to analog signal ( live microphone feed ) can be significantly reduced to (almost but still not insignificant by the use of DSD128) insignificant discrepancy from the original by high enough sampling frequency. It is perceived in the timbre - 44.1/16 sounds hard and sizzly, where ever faster sampling sounds ever softer and smoother, in the end approaching to the live microphone feed. This statement does and will continue to hold water regardless of recent improvements in filtering for the 44.1/16 - the exact same measures can be used to improve filtering of higher sample rate PCM and similarly filtering for the DSD - which by the time DSD 512 is reached, becomes practically superfluous.  It means, for all practical purposes, signal that is faster than anything possible with analog with out of audio band noise low enough not to cause trouble - in other words, something that can faithfully record and play back music.
 
It is unfortunately true that file sizes and everything that supports  hirez is mind boggling:
 
DSD 64 (SACD )         1GB                22 minutes audio
 
DSD 128                       1GB               11 minutes audio
 
DSD 256                        1GB               330 seconds audio
 
DSD 512                         1GB              165 seconds audio
 
Exact time is slightly different, but no hair splitting please, meant was to show it does look daunting  . Similar occurs with DXD that can by now go past 700 kHz sampling/24 bit. But at the rate computers are progressing, above in say a decade may look "business as usual". I certainly hope so.
 
I wish I could afford beyond DSD128 - but it will have to wait, impossible at the moment, as it means exchanging everything from recorder(s) to hard disks ( from above it should be clear SSD is out of question due to storage size/cost requirement ) and everything in between. But I do see the benefit, although it should not be as dramatic as the jump from DSD 64 to DSD 128 ( roughly equivalent in difference 44.1 vs 88.2 or 96 in PCM ).
 
Conclusion : Dear Santa....

 

[diamondears]

You do realize this is the "Sound Science" section of Head-Fi?

Subjective opinions without proper analysis or control will always be heavily scrutinized.  You really should perform a proper ABX test - you will probably find the results surprising.

Yes I do. That's why on my first post I said "I don't know about the objective measurements, but listening to the DSD128 version blablablah"...so that objective, technical mumbo jumbo is irrelevant."

Ears don't lie. There are still so many things in this world that still undiscovered in terms of measurements and science. Hearing is science too, I reckon. But I digress. Go on with it. Just thought of saying the obvious.

 

[bfreedma]

Yes I do. That's why on my first post I said "I don't know about the objective measurements, but listening to the DSD128 version blablablah"...so that objective, technical mumbo jumbo is irrelevant."

Ears don't lie. There are still so many things in this world that still undiscovered in terms of measurements and science. Hearing is science too, I reckon. But I digress. Go on with it. Just thought of saying the obvious.

 
Ears most certainly do lie - environment, conditions, and controls also are variable.  That's why objective measurements and controlled testing are required for the topic at hand.
 
Saying that "things are undiscovered" is just a cop out in Sound Science.  Particularly when claiming that the differences are "night and day" or "substantial".

 

[diamondears]

Ears most certainly do lie - environment, conditions, and controls also are variable.  That's why objective measurements and controlled testing are required for the topic at hand.

Saying that "things are undiscovered" is just a cop out in Sound Science.  Particularly when claiming that the differences are "night and day" or "substantial".

That's what they said centuries ago when someone said the earth is round.

Anyway, are you saying that in a proper ABX testing, the DSD128 or up is the same as red book?

 

[bfreedma]

Ears most certainly do lie - environment, conditions, and controls also are variable. That's why objective measurements and controlled testing are required for the topic at hand.

Saying that "things are undiscovered" is just a cop out in Sound Science. Particularly when claiming that the differences are "night and day" or "substantial".

That's what they said centuries ago when someone said the earth is round.

Anyway, are you saying that in a proper ABX testing, the DSD128 or up is the same as red book?

The old "the earth is round" canard. Also not appropriate for sound science.

Yes, I'm saying that legitimate controlled testing to date has demonstrated that humans cannot hear any differences between red book and any of the hi res formats.

Do you have any evidence to the contrary? Are you stating that you believe there are "night and day"/"substantial" audible differences in formats that can't be measured with modern instruments?

 

[RRod]

Yes I do. That's why on my first post I said "I don't know about the objective measurements, but listening to the DSD128 version blablablah"...so that objective, technical mumbo jumbo is irrelevant."

Ears don't lie. There are still so many things in this world that still undiscovered in terms of measurements and science. Hearing is science too, I reckon. But I digress. Go on with it. Just thought of saying the obvious.

 
Ears may not lie but the brain sure does. It's very good at confirming things to us that we want confirmed without objective proof.

 

[stv014]

  Please do refer me to the videos and papers proving that 16/44.1 can resolve time differences into the nanosecond range.

 
It follows directly from the Nyquist-Shannon sampling theorem, which states that (quote): "If a function x(t) contains no frequencies higher than B cps, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart". Since delaying a signal by any amount is a linear transform (it does not introduce any new frequencies that did not exist before applying the delay), if sampling a band-limited input is mathematically lossless, then so should be sampling its delayed version. Therefore, the resolvable time differences are limited only by the (possibly shaped) quantization noise, and that is no different from an analog signal with the same noise density. If you still do not believe, I can post a Red Book format file with nanosecond delays that you can detect by analyzing the file (unfortunately not by listening, as the delay would be too small to be audible).
 
Quote:

However, "ringing" ( the correct consequence of lacking high frequencies ) in 44.1/16 > 88.2 >etc compared to analog signal ( live microphone feed ) can be significantly reduced to (almost but still not insignificant by the use of DSD128) insignificant discrepancy from the original by high enough sampling frequency. It is perceived in the timbre - 44.1/16 sounds hard and sizzly, where ever faster sampling sounds ever softer and smoother, in the end approaching to the live microphone feed.

 
That should just make the test suggested in post 3303 even easier. That is, as long as short ultrasonic ringing is actually audible, let alone with the masking effect from the higher magnitude content in the audio band. Any results ?
 

Originally Posted by analogsurviver /img/forum/go_quote.gif
 
That regarding timeline of (non)availability of "digital" sharing a single DAC with consequent phase shift "since 90s" does not hold true. Some computers, notebooks, netbooks etc, as well as external DACs, have been available as recently as 2009 - if not longer. With these, reduction of phase shift is clearly audible AND visible on the oscilloscope by the use of higher than 44.1 kHz sampling frequency in any software, foobar2000 being the most commonly known and used. Audible difference lies primarily in imaging : soundstage width ( it becomes wider/broader ) and soundstage depth ( it starts to show some vestiges of depth, it is no longer glass pane flat ). Since such "digital whatevers" are usually limited to sampling frequency of 48 kHz

 
See above regarding clear audibility and proving it. By the way, a limitation to 48 kHz sample rate was typical of AC97 audio, which is not exactly new. Currently, even HD audio codecs on PC motherboards support 192 kHz playback in hardware.

 

[stv014]

Ears don't lie.

 
Subjective perception can and does lie, depending on the circumstances. Several examples have already been posted (probably more than once) in this same thread.
 

That's what they said centuries ago when someone said the earth is round.

 
It is ironical that the "flat earth" belief is actually the result of naive subjectivism (after all, it obviously looks flat from where one is standing, so it must be true, right ?), yet keeps getting brought up in its defense.