SACD vs. CD - Unfair competition?

[WmAx]

The true benefit of using a higher sampling rate comes from more in band sampling of the voltage of analog waveform. The more times you sample the waveform, the more precise the imaging, the better the tonal quality, and the higher the resolution of the audible signal.

All undeniably true if you remove the word audible from the end of this statement. You state this as if it's proven fact, and I am not aware of any research coming to this conclusion as far as audibility is concerned.

The higher up in sampling frequency you go, the more these things improve up until a point Read;

http://www.digitalproducer.com/artic...le.jsp?id=7408

Certainly it improves accuracy. Audibly with normal music playback? I don't see subtantial evidence of this.

Keep in mind, tests concerning the audiblity of high frequency information above human hearing are still inconclusive

That is the issue So far, the respected controlled tests/references on this subject have not been able to achieve a positive result.

So the perceived effects of higher frequencies on the listening experience have not been determined, and therefore CANNOT be ruled out

I did not state the contrary. I stated exactly this sediment, but I also stated that it is not logical to attribute the things 'credited' to hi-rez playback since their is no strong evidence that suggests that this should be the case. Until a peer-reviewed, scrutinized, valid audiblity test has been performed that achieves positive statistical signficance, then it can not be accepted as fact.

-Chris

[Woochifer]

Transmission of a bandwidth not proven to be audible in a controlled test should not be ruled out as a casual effect for the improved sound quality that you heard? Perhaps, I don't like to 'absolutely' rule anything 'out'. However, I don't see it as logical to presume that a larger bandwidth, in itself, is enhancing the audible data appreciably. Failure to achieve positive results in controlled tests does not lend support to the idea.

-Chris

How's it not logical? I'm basically pointing out a variable that's not accounted for. I don't have the mechanisms or access to source material to prove that any other factor is a more valid causal link than another, and neither do you. I know that the bandwidth is one of the variables, so therefore it has to remain on the table as potential causal factor until it is demonstrated to me that some other variable is more responsible for what I observed.

You cite the need for controlled tests. Fine. Bring over the original master tapes and we can set the blind controlled listenings anywhere you want. If the CD, 96/24 disc, and original master are all transparent to one another under those conditions, then we have the answer. Otherwise, you are making a conclusion in the absence of proof as well by trying to rule out the bandwidth as a causal factor without knowing anything else about the source material.

[Monstrous Mike]

The higher up in sampling frequency you go, the more these things improve up until a point Read;

http://www.digitalproducer.com/artic...le.jsp?id=7408

Ok, I read that article and I have extracted the following quote:

"It’s been determined that time delay differences of 15 microseconds between left and right ears are easily discernible by nearly anyone. That’s less than the time difference between two samples at 48kHz (about 20 microseconds). Using a single pulse, one microsecond in length as a source, some listeners can perceive time delay differences of as little as five microseconds between left and right. It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. At 96kHz (a popularly preferred sample rate) there is a 10.417-microsecond space between samples. At 192kHz sample rate there is a 5.208-microsecond space between samples. This reasoning suggests that a sample rate of 192kHz is probably a good choice. As processors increase in speed and efficiency and as storage capacity expands high sample rates, long word length will become an insignificant concern and we’ll be able to focus on the next audio catastrophe.

I'm having some trouble understanding the above. The author is talking about time delay differences between the left and right ear and this might refer to sound coming from the left and right speakers which left the speaker a different times or sound coming from one speaker but the listener's ears are not the same distance (i.e. head turned) from that speaker. I believe this allows the listener to determine the direction the sound is coming from.

However, for the life of me, I cannot see how the has anything to with the sampling frequency of the digital audio signal. Do you have any ideas?

[WmAx]

[QUOTE]

Ok, I read that article and I have extracted the following quote:

"It’s been determined that time delay differences of 15 microseconds between left and right ears are easily discernible by nearly anyone. That’s less than the time difference between two samples at 48kHz (about 20 microseconds). Using a single pulse, one microsecond in length as a source, some listeners can perceive time delay differences of as little as five microseconds between left and right. It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. At 96kHz (a popularly preferred sample rate) there is a 10.417-microsecond space between samples. At 192kHz sample rate there is a 5.208-microsecond space between samples.

I'm having some trouble understanding the above. The author is talking about time delay differences between the left and right ear and this might refer to sound coming from the left and right speakers which left the speaker a different times or sound coming from one speaker but the listener's ears are not the same distance (i.e. head turned) from that speaker. I believe this allows the listener to determine the direction the sound is coming from.

This sounds like reference to Nordmark research(1.5us) and another research study. These studies found that humans can detect very slight time differences between channels under special conditinos. These conditinos were (5us) impulses and (1.5us) assymetricaly jittered pulses between two channels. These have not shown to be relevant to music that I am aware. As far as general bandwidth, in testing, humans are more sensitive to test tones in special test signals as opposed to music, also. The same goes for audibility of phase, polarity, distortions,e tc. The relevance to music playback is not clear.

However, for the life of me, I cannot see how the has anything to with the sampling frequency of the digital audio signal. Do you have any ideas?

Sampling frequency(bandwidth) dictates the precison of sampling an amplitude in a given space of time. While in a simple analysis, the bandwidth of 20kHz is adequte to adress raw bandwidth sensitivy, in order to replicate the test tone circumstance times suggested(5us, etc.), a far higher bandwidth would be requrined in order to record/playback. For example, if y ou have an acoustic event that contains only audible data(<20kHz), this still does not account for the potential tiny time dealy differences between channels(ears) that will occur since each ear is a discrete sensor essentially. Apparently, the interchannel time sensitivity of human ears is far higher then the raw bandwidth detectability. For a simplistic model, imagine 2 microphones(let's pretend it has a 200khz bandwidth for this discussion) imagine a sound source, of the same spectral content that is within 20khz bandwidth, one mike is placed 2mm farther away then the other. Obviously, it woudl require a 170kHz bandwidth to accurately record and play back this difference betweeen the two sources. Your ears apparently have this type of effect. Just remember that this been demonstrated to be readily audible with special test tones, not music.

-Chris

[WmAx]

I don't have the mechanisms or access to source material to prove that any other factor is a more valid causal link than another, and neither do you. I know that the bandwidth is one of the variables, so therefore it has to remain on the table as potential causal factor until it is demonstrated to me that some other variable is more responsible for what I observed.

It has not been demonstrated to be important. Their have been careful studies to attempt to confirm, but none that stood up to scrutiny have demonstrated positive results.

Which Bandwidth Is Necesarry for Optimal Sound Transmission?
G. Plenge, H. Jakubowski, P. Schone, JAES, 1978

Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components
JAES, Preprint 5876, Convention 115, 2003
Toshiyuki Nishiguchi, Kimio Hamasaki, Masakazu Iwaki, and Akio Ando

Otherwise, you are making a conclusion in the absence of proof as well by trying to rule out the bandwidth as a causal factor without knowing anything else about the source material.

Read the above papers.

As far as conclusion without proof? No. I CAN NOT conclude that your claim has any signfigance. Data does not suggest this conclusion. You would have me assume things are true before such has been proven?

-Chris

[Woochifer]

It has not been demonstrated to be important. Their have been careful studies to attempt to confirm, but none that stood up to scrutiny have demonstrated positive results.

If the 44.1/16 bandwidth is sufficient to cover all audible phenomena, then why is it that this bandwidth has never been the standard used by professional sound engineers and mixers with original recordings and during the mixing process?

As far as conclusion without proof? No. I CAN NOT conclude that your claim has any signfigance. Data does not suggest this conclusion. You would have me assume things are true before such has been proven?

And I never suggested a conclusion. I'm pointing out that I've not ruled out any causal factors and have insufficient information to reach a conclusion. However you want to preclude variables without knowing anything about the source material is entirely your prerogative.

[kingdaddykeith]

If the 44.1/16 bandwidth is sufficient to cover all audible phenomena, then why is it that this bandwidth has never been the standard used by professional sound engineers and mixers with original recordings and during the mixing process? .

It's always been my understanding that it is more of a headroom issue, the higher sampling rates allows for more headroom so the recording level can be hotter, which lowers the noise floor.

[WmAx]

If the 44.1/16 bandwidth is sufficient to cover all audible phenomena, then why is it that this bandwidth has never been the standard used by professional sound engineers and mixers with original recordings and during the mixing process?

The optimal parameters for playback do not necsarrily translate into the optimal parameters for flexible/versatile attributes for recording/format change, etc.

JAES, July/August 1978, Volume 26, Number 7/8, Page 562

Here is the relevant quote:

In the discussions of standards relative to digital audio to date we feel that the needs of broadcasting organizations have been little mentioned, and we would like to make a few points.
In Europe a standard sampling rate of 32kHz +/- 50 parts per million, giving an audio bandwidth of 15kHz, has been agreed within the EBU for use by broadcasters. As commercial applications assume a bandwidth of about 20Khz, and hence sampling rates from 40-60 kHz, it is probable that broadcasters who will need to interface between these standards will do so by means of a digital rate-changing filter, so avoiding D/A and A/D conversion.
To make this rate-changing filter as simple as possible to instrument, it is desirable to choose certain sampling frequencies for the commercial recording application. These in order of merit are:

(1)_________48
(2)40 ______________56
(3) ___44 ________52 __60
(4) __42 _46 ___50

Each row of frequencies requires twice as many calculations in the filter as the previous one. For easy rate-changing of this kind, both the input and the output sampling rates should be locked, and so any choice of system-clock frequency should be integer related to 32kHz, as well as to the system sampling rate.

-Chris

[WmAx]

It's always been my understanding that it is more of a headroom issue, the higher sampling rates allows for more headroom so the recording level can be hotter, which lowers the noise floor.

Higher bitrate, indeed, allows or eaiser digital recording. It allows a an additional safeguard against improper level settings to the recorder causing clipping(slight improper seetings won't be a disaster later on), clipping due to extraordinary dynamic sources, as well as allowing for different dithering processes to be used when reduced to standard 16 bit depth.

-Chris

[Woochifer]

I'm certianly not going to buy into "hi-res" audio if it is inherently no better for two-channel music than CD "low-res" [??] audio. I won't buy into the new format simply because they are paying better attention to the mastering with the new format. A LOT of people join me in this sentiment. If this is what the recording companies are doing, "hi-res" is toast, IMHO.

I have a VERY rudimentary understanding of these things. As I understand it, CDs are sampled at 44.1 khz, so that the frequency response maxes out at about 22 khz, which is well in excess of the hearing of the vast majority of the human population, though dogs may be able to appreciate it.

I'm not going to be running double-blind of ABX tests between SACD and CD disks listening for audible consequences of 23 khz info in this lifetime. Life's too short, I'm not going to spend my money on such silliness if there's no support for it in theory, and I have too little expertise. If I am persuaded that CDs should have the same two-channel audio quailty as SACDs, DVD audio, etc., I'm not gonna bite for the "high-res" stuff, as a matter of principle. That' why I'm asking.

The vast majority of households, including mine, have no interest whatsoever in anything more than highly euphonic two-channel sound or in trying to hear what little information is conveyed above 22 khz.

I am quite willing to alter my views, but not based on the thin reed of purely subjective assertions.

To me, it boils down to a very simple question. Do the high res discs improve upon the listening experience over what the CD versions offer?

So far, whether it's the audible improvements I've observed with two-channel material, or with the whole new dimension of 5.1 surround mixes, my answer is a definite yes. The 96/24 discs I've bought thus far are a clear cut improvement upon their CD counterparts, and what 5.1 surround music brings to the table is a whole new way to enjoy music. If this is typical of SACD and DVD-A, then I see no drawback to it whatsoever.

There are still plenty of poorly done CDs out there, and any chance to revisit these recordings and give them an improved transfer is welcome in my view. In addition, creating a 5.1 surround mix requires going back to the original multitrack master, which means that it's possible to obtain a higher resolution mixdown than a version that was originally done using analog recorders (and potentially degraded by going through successive iterations during the mixdown process on analog equipment). This would include the two-channel mixdown as well, if the artist chooses to have an album remixed at the higher resolution.

You're more than welcome to quibble about what the true causal effect is, or choose not to get into high res based on some personal principle. I don't have the answer on what the true causal effects are (and absent access to the original source material, nobody else does either), and frankly, I don't care. In the meantime, I'll just enjoy what these new versions offer with better sound quality and listening to familiar music in a new way. In my view, results count and what I've observed so far, the new high res discs have delivered.

[Steve1000]

Fair enough! Your point of view is entirely reasonable, IMHO. Mine's just a little different.

You're more than welcome to quibble about what the true causal effect is, or choose not to get into high res based on some personal principle. I don't have the answer on what the true causal effects are (and absent access to the original source material, nobody else does either), and frankly, I don't care. In the meantime, I'll just enjoy what these new versions offer with better sound quality and listening to familiar music in a new way. In my view, results count and what I've observed so far, the new high res discs have delivered.

[kingdaddykeith]

I'm starting to believe that a hi-Rez recording is better when down converted at the end of the chain. The DAD-A's and especially the SACD’s that I've sampled are, well I hate to use the word grunge, but that what the higher frequencies sound like for the lack of a better word. Maybe it’s some kind of artificial noise or something but it's not right to my ears. However when I buy a 24/96 DTS or DD 5.1 mix it sounds much more real and less fatiguing on the top and much fuller on the bottom then any of my SACD’s.

I don’t know if this is true but I've read, and was told by a Parasound engineer that the higher the sampling rate the more noise, and the reason most all (if not all) 24/96 input compatible processors down convert to 16/48 (even my Halo C2) is because of the lack of technology to either filter or negate this noise. It makes sense at the recording end to have the most resolution and headroom as possible, but the playback of this full resolution has yet to impress me.

[Monstrous Mike]

If the 44.1/16 bandwidth is sufficient to cover all audible phenomena, then why is it that this bandwidth has never been the standard used by professional sound engineers and mixers with original recordings and during the mixing process?

When working with digital audio it is better to record and process the audio at higher sampling rates and bit lengths and then produce the master CD at 44.1/16. If an audio engineer kept his signal at 44.1/16 from microphone to master, the possibility of some losses and distortions being introduced are more likely than when handling the digital audio at the higher levels during processing, mixing, filtering, normalization, equalization, etc.

[Monstrous Mike]

[QUOTE=WmAx]

Sampling frequency(bandwidth) dictates the precison of sampling an amplitude in a given space of time. While in a simple analysis, the bandwidth of 20kHz is adequte to adress raw bandwidth sensitivy, in order to replicate the test tone circumstance times suggested(5us, etc.), a far higher bandwidth would be requrined in order to record/playback. For example, if y ou have an acoustic event that contains only audible data(<20kHz), this still does not account for the potential tiny time dealy differences between channels(ears) that will occur since each ear is a discrete sensor essentially. Apparently, the interchannel time sensitivity of human ears is far higher then the raw bandwidth detectability. For a simplistic model, imagine 2 microphones(let's pretend it has a 200khz bandwidth for this discussion) imagine a sound source, of the same spectral content that is within 20khz bandwidth, one mike is placed 2mm farther away then the other. Obviously, it woudl require a 170kHz bandwidth to accurately record and play back this difference betweeen the two sources. Your ears apparently have this type of effect. Just remember that this been demonstrated to be readily audible with special test tones, not music.

-Chris

I really don't have any idea what you are explaining here.

The author of my previous quote is very clearly implying that a 96 kHz sampling rate is audibly better and a 192 kHz sampling rate is even better than that. And the implication is that the time distance between the samples can affect the imaging vis-a-vis a time delay.

I agree that a phenomenon called "interaural time delay" can be detected by humans and is used in conjunction with intensity differences to determine a direction for sound. But like I said I do not see how time delay is related in any way to bit sampling rate.

[WmAx]

The author of my previous quote is very clearly implying that a 96 kHz sampling rate is audibly better and a 192 kHz sampling rate is even better than that. And the implication is that the time distance between the samples can affect the imaging vis-a-vis a time delay.

I was not addressing the author, I was addressing the specific question/issue you presented about the interchannel time delays that he mentioned.

I agree that a phenomenon called "interaural time delay" can be detected by humans and is used in conjunction with intensity differences to determine a direction for sound. But like I said I do not see how time delay is related in any way to bit sampling rate.

I don't know what it has to do with bit sampling rate, either. BUt as for frequency sampling rate, how else would you propose to record/playback at a 5us or 1.5us accuracy with 44.1Khz? While the audible spectral informatin will be recorded with redbook format, the specific coordinates in time will be shifted into what can be stored in a 44.1kHz rate.

Here is an over-simplified illustration of my understanding of this phenomena and how it related to sampling frequency:

H=2mm(approx. 6us)(0dB)
UUUUUUUU=17mm(50us)(1 cycle 20khz sine wave)

Potential difference example:

170Khz bandwidth limited
L:
HHHHHUUUUUUUUHHHUUUUUUUUHHHHHHHHHUUUUUUUUHH
R:
HHHHUUUUUUUUHHHHUUUUUUUUHHHHHHHHHHHUUUUUUUU

20kHz bandwidth limited
L:
HHHHHHHHUUUUUUUU UUUUUUUUHHHHHHHHUUUUUUUU
R:
HHHHHHHHUUUUUUUU UUUUUUUUHHHHHHHHUUUUUUUU

The higher bandiwdth can allow for the interchannel time difference to exist at finer resolution, as illustrated in the crude graphic above. While a 20kHz cycle is 50us in duration, the actual time at where this amplitude can actually originate is not limited. The 20kHz wavelength can begin at 200us or 204us or 201.3486 us, etc. Lower sampling rated reduces this possible difference relative the sampling rate limits.

If my understanding is wrong, please explain.

All of this and how it relates to audibility are a different issue.

-Chris

[Monstrous Mike]

I don't know what it has to do with bit sampling rate, either. BUt as for frequency sampling rate, how else would you propose to record/playback at a 5us or 1.5us accuracy with 44.1Khz?

-Chris

I do not know what you mean by a 5 microsecond accuracy.

Let's look at this from another angle. Let's assume we have two signals which are identical but signal B is delayed by 5 microseconds. I presume this is what we are talking about. Now let's assume these signals are a 1kH sine wave with an amplitude of +/- 1 volt.

And I presume the premise is that a 44.1 kHz sampling rate will not be able to accurate capture this time delay. Using 48 kHz (which is close to 44.1) the time between each sample is 20 microseconds. Again the premise is that this long period between samples is not sufficient to capture the 5 microsecond delay. I guess that would seem obvious at first glance.

However we have our two identical 1 kHz sine waves where the second has started 5 microseconds behind the first. This represents a 1.8 degree phase shift. Therefore, sampling at 48 kHz (i.e. 20 microsecond intervals) yields this:

Digital Sample Number One

Time = 0 seconds
Signal A = sin (0 degrees) = 0
Signal B = sin (-1.8 degrees) = - 0.0314

Digital Sample Number Two

Time = 20 microseconds
Signal A = sin (7.2 degrees) = 0.126
Signal B = sin (5.4 degrees) = 0.0941

Digital Sample Number Three

Time = 40 microseconds
Signal A = sin (14.4 degrees) = 0.249
Signal B = sin (12.6 degrees) = 0.218
Digital Sample Number Four

Time = 60 microseconds
Signal A = sin (21.6 degrees) = 0.377
Signal B = sin (19.8 degrees) = 0.339
.
.

Digital Sample Number Ten
Time = 200 microseconds
Signal A = sin (72.0 degrees) = 0.951
Signal B = sin (70.2 degrees) = 0.941
.
.

Digital Sample Number Thirty
Time = 600 microseconds
Signal A = sin (216.0 degrees) = - 0.588
Signal B = sin (214.2 degrees) = - 0.562
.
. etc.

I think that clearly shows that a 5 microsecond delay is captured quite well with a 20 microsecond sampling interval. As a matter of fact, I think the number of bits representing those analog amplitude values have more of an affect on capturing the delay than the sampling rate does.

Further, using Nyquist's Theorum, the above will hold true for frequencies up to 24 kHz.

[WmAx]

However we have our two identical 1 kHz sine waves where the second has started 5 microseconds behind the first. This represents a 1.8 degree phase shift. Therefore, sampling at 48 kHz (i.e. 20 microsecond intervals) yields this:

I think that clearly shows that a 5 microsecond delay is captured quite well with a 20 microsecond sampling interval.

Thank you for the correction. I simulated this after your explanation, and you are correct, that the time difference is indeed more accurate then the raw sampling frequency lead me to believe. Appears I was shortsighted - thinking of only the sample frequency. My error.

-Chris

[kingdaddykeith]

I though someone would find this interesting, it’s a recording engineers opinion of the compression problem with CD's. Didn’t read it all so forgive the poor description.

http://georgegraham.com/compress.html

[Feanor]

...Let's look at this from another angle. Let's assume we have two signals which are identical but signal B is delayed by 5 microseconds. I presume this is what we are talking about. Now let's assume these signals are a 1kH sine wave with an amplitude of +/- 1 volt.

...
I think that clearly shows that a 5 microsecond delay is captured quite well with a 20 microsecond sampling interval. As a matter of fact, I think the number of bits representing those analog amplitude values have more of an affect on capturing the delay than the sampling rate does.

Further, using Nyquist's Theorum, the above will hold true for frequencies up to 24 kHz.

Your example proves that 44.1KHz can distinguish two sine waves 5 us apart. But we don't listen to sine waves. Suppose there is a pair of complex wave forms where there are instantaneous spikes 5 us apart?

[WmAx]

Your example proves that 44.1KHz can distinguish two sine waves 5 us apart. But we don't listen to sine waves. Suppose there is a pair of complex wave forms where there are instantaneous spikes 5 us apart?

Spikes? You mean 'impulses'? That's not something you find in music or nature. However, with nearly any natural sound, you can express the signals as a sum of sine waves. This extends to any symmterical waveform(as opposed to assymetrical which is only common in a synthetic environment(usually test signals)): square wave, triangle wave, etc. They are a result of many sine waves.

-Chris

[Feanor]

Spikes? You mean 'impulses'? That's not something you find in music or nature. ...

... if not in music. Isn't true that supersonic events, (at least), cause impulses? These everts aren't all that uncommon, e.g. base ball hit by a bat, crack of a bull wip, gun shots, some explosions. No wonder these things never sound real except heard live!

[Monstrous Mike]

Your example proves that 44.1KHz can distinguish two sine waves 5 us apart. But we don't listen to sine waves. Suppose there is a pair of complex wave forms where there are instantaneous spikes 5 us apart?

A spike of 5 us would certain be missed by a sampling interval of 20 us. However, that 5 us pulse would have frequency components in the 384 kHz range or greater and thus are not audible and cannot be reproduced by standard amplifiers or speakers. A sampling interval of 20 us represents a sampling rate of 48 kHz. According to Nyquist Theorem, that means it can only capture frequency components of 24 kHz or less. Spikes and other spurious noise over 24 kHz will not be captured.

So a 48 kHz sampling rate can capture frequencies with a period of 20 us or more and can also capture two signals which are offset by 5 us (assuming a large enough bit word length).

You are confusing time shifting with spectral content..

[WmAx]

... if not in music. Isn't true that supersonic events, (at least), cause impulses? These everts aren't all that uncommon, e.g. base ball hit by a bat, crack of a bull wip, gun shots, some explosions. No wonder these things never sound real except heard live!

The events you describe are composed of primarily symmetrical waveforms. Perhaps not perfect, but I don't know. An impulse is assymetrical. However, this is immaterial. It is possible for some speakers to reproduce assymetrical waveforms with great accuracy.

An assymetrical waveform is one that does not have inversely matching values in it's two 180 degree halves. These can be seperated and looked at as negative and positive sections of the waveform.

Here are two simplified illustrations, represent a waveform with symetry and the same waveform without.

Symmetrical Wave form

+ Pos
.......H..............
.....H...H...........
...H.......H......H.. 0 zero
.............H....H..
................H.....
- Neg


Assymetrical(extreme - for illustration)

+ Pos
......H...............
....H..H.............
..H......HHH....... 0 zero
........................
........................
-Neg

-Chris

[Sir Terrence the Terrible]

All undeniably true if you remove the word audible from the end of this statement. You state this as if it's proven fact, and I am not aware of any research coming to this conclusion as far as audibility is concerned.

Chris, it is not financially feasible for any engineer to sit around a wait for science to tell them what they already hear. It is well documented that engineers get better imaging from the use of higher sampling rates. It is well documented that engineers hear their mixes more clearly at higher sampling rates, so I don't think any intelligent engineer is going to sit around waiting for research on the issue.

Certainly it improves accuracy. Audibly with normal music playback? I don't see subtantial evidence of this.

What would constitute substantial to you? I mean considering that just about every studio in Los Angeles, New York, Memphis, and every other major city that has a large music community has migrated from 16/44.1khz to 24/96khz, I would call that VERY substantial. Someone had to have heard an audible improvement, or there would be nothing to justify the cost of the upgrade, which can run into hundreds of thousands of dollars. So if you are looking for science to prove what many already know, then by all means do so, but that doesn't make good business sense to me.

That is the issue So far, the respected controlled tests/references on this subject have not been able to achieve a positive result.

They have not been able to acheive a negative result either. So it would be short sighted to discount it altogether.

did not state the contrary. I stated exactly this sediment, but I also stated that it is not logical to attribute the things 'credited' to hi-rez playback since their is no strong evidence that suggests that this should be the case. Until a peer-reviewed, scrutinized, valid audiblity test has been performed that achieves positive statistical signficance, then it can not be accepted as fact.

I disagree with your perspective entirely. In case you didn't know it, I (like many other engineers) sit down for many hours testing and listening to new equipment to decide whether it is worth my investment. I (like many engineers) have my own set of test that allow me to do this in a way that I can make an educated decision. It is not my job to become a scientist, conduct listening test to obtain a statistical measure just to justify my purchase. That is inefficient and unnecessary. After I am finish testing a piece of equipment, I know for a fact that my decision to purchase, or not is an educated one. I do not need DBT , and a peer review to make that decision for me. It is my feeling that most engineers feel this way.

(This is just my opinion) DBT, research and publishing for peer review is for the scientific community. That is not the job of a audio engineer. We only need one answer, does it sound better than my current equipment. According to polls taken at the Surround 2004 conference, about 86% of engineers polled believes that 24/96khz sounds better than 16/44.1khz. Is that scientific? No, but it leads me to believe that where there is smoke, there is fire.

I have taken this position and I am going to pretty much stick by it for now. I have done my own homework listening to various recordings I have done at various bit and sample rates. I have used several recorders during the same session set at various bit and sample rates so I can play them back and listen. I made my decision based on what I heard. If I heard no differences between 44.1, 48, and 96khz, I would have probably stuck with 44.1 since it required no investment. That however was not the case, and I invested in what I thought sounded the best.

Does the sample rate make a difference in sould quality? Definately. Why? I know it improves imaging, and the sound is cleaner and more distinct to the ear, but otherwise I don't know. Does bitrate matter? Only in recording and post production. I'll let the scientist figure the other crap out

[mtrycraft]

Chris, it is not financially feasible for any engineer to sit around a wait for science to tell them what they already hear.


Or what they imagine to hear?
They should at least see what science has to say about it when that data is available.


It is well documented that engineers get better imaging from the use of higher sampling rates.

What kind of documents? Not all documents are created equal.


It is well documented that engineers hear their mixes more clearly at higher sampling rates,

Same as above.

o I don't think any intelligent engineer is going to sit around waiting for research on the issue.


But what will that intelligent engineer do when the data is in? Or, cannot be demonstrated? Ignore it?


I mean considering that just about every studio in Los Angeles, New York, Memphis, and every other major city that has a large music community has migrated from 16/44.1khz to 24/96khz, I would call that VERY substantial.


Substantial only by numbers. Doesn't mean much beyond that though. After all a huge number of people on the planet believe in the supreme being.

Someone had to have heard an audible improvement, or there would be nothing to justify the cost of the upgrade,


That is absolute nonsense. One only has to look at the high end audio, and audio cable industry in specific.
This is a trend driven by numerous drivers. Besides, mastering is different from consumer audio listening and reproduction.


So if you are looking for science to prove what many already know,

Or, what they only think they know as that is certainly not out of question and is certainly a valid and real possibiolity.





In case you didn't know it, I (like many other engineers) sit down for many hours testing and listening to new equipment to decide whether it is worth my investment.


Subjectively, of course, right? So, it is prone top bias and gullibility?

It is not my job to become a scientist, conduct listening test to obtain a statistical measure just to justify my purchase.

Ah, but if you did do such lisening tests, maybe you wouldn't follow the herd blindly and not waste you money foolishly?


That is inefficient and unnecessary.

Not if it gets you to an objective answer instead of guessing or just an expensive preference issue.

After I am finish testing a piece of equipment, I know for a fact that my decision to purchase, or not is an educated one.

How can you? It is based on a very subjective test prone to bias and unreliability.

I do not need DBT ,

That is unfortunate.



It is my feeling that most engineers feel this way.

That is unfortunate also.


(This is just my opinion) DBT, research and publishing for peer review is for the scientific community.


While you have this opinion, it is unfounded.

That is not the job of a audio engineer.

Why not? I would think you wanted real answers, the truths, not maybe or whatever.


We only need one answer, does it sound better than my current equipment.


That is the whole point. You don't know, not in an objective manner. You think you do but far from being a fact.

According to polls taken at the Surround 2004 conference, about 86% of engineers polled believes that 24/96khz sounds better than 16/44.1khz. Is that scientific? No, but it leads me to believe that where there is smoke, there is fire.

Well, at least you know it is not scientific. Why not find out for sure?
A higher percent believe in the supreme being. Where there is smoke there is fire, right?
How about psychics? Homeopathic medicines? We can go on and on, audio doesn't have immunity from nonsense, myths, hype, etc.