Yes Close To The Edge HDTracks

[Jud]

If we are mearuring thing by who is advocating what (not a very good measure of anything, in my opinion), I guess you have to add at least xiphmont (Christopher “Monty” Montgomery of xiph.org fame) on the side of Dan and me.

 

I'm not absolutely sure of that. I certainly understand his thoughts about the resolution of recordings. But when I mentioned internal 8x oversampling by DACs in a brief but pleasant correspondence with him, he said no one had raised that issue with him before. Remember, when 8x oversampling was developed, it was to listen to 44.1k material. So I'm talking about the filtering used in the process of conversion to analog, rather than the resolution of the digital recording.

[bdiament]

Hi Julf,

 

Barry,

 

Just because people don't agree with the point you are trying to make doesn't mean that they don't understand it.

 

So I understand your point. I just don't agree with it, and I have explained why...

 

I understand we both have different perspectives on this.

To be clear, the reason I asked is because you asked "What does it prove?"

I answered your question.

 

...I think we can move back to the original topic.

 

Fair enough.

Thanks for the dialog.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Julf]

Barry,

 

I am truly sorry to have to continue the off-topic dialog, but I feel your posting does need to be addressed.

 

I'm sorry but you are describing a theoretical concept and not a real one (nor one that any engineer I know would subscribe to --- so much for "the rest of the world", though I'm sure there are folks who share your view).

 

I would love to hear your definition for dynamic range. The one I have quoted is the one used by engineers in every field I know.

 

There is *potential* dynamic range and there is what a listener in the real world might experience. These are *very* different.

 

But we don't measure dynamic range by measuring what a listener experiences. We measure it by comparing the loudest possible signal to the quietest possible signal.

 

A signal recorded at -40 is not using anything above that level. If you record a tone at -40 and another tone at -96 (or -95), the difference between them is 56 dB (or 55 dB). That is something you can measure. The range between them will never be larger, regardless of the "potential" of the medium on which they are recorded. Take that same pair of tones to a 24-bit medium and the range between them still doesn't change.

 

So if what you are saying is that if you choose not to use the full dynamic range, you don't get the full dynamic range? In that case I have to agree with you...

 

Does the super compressed heavy metal CD with everything except the spaces between songs up near 0 have the same dynamic range as a CD of a Keith Johnson recording? As both are on CD, both have that theoretical *potential* dynamic range of 96 dB. If you believe both of these have a 96 dB dynamic range (or anything close to that), I hope we can once again agree to disagree.

 

Let's not confuse the dynamic range of the source material with the dynamic range of the media or format. According to your logic, the compressed heavy metal should sound better, because it (having a much smaller dynamic range) has much more bits available to represent the quieter parts (because they aren't very quiet). So your Keith Johnson recording, with passages that are down to -40 dBFS, only has 16 out of 24 bits available to represent it, while the heavy metal has 22 out of 24 bits available to represent it, assuming it is compressed down to 6 dB of dynamic range.

[Julf]

So I'm talking about the filtering used in the process of conversion to analog, rather than the resolution of the digital recording.

 

Thank you for that clarification - yes, in a DAC, high multiples of oversampling makes sense, but that is a different situation than when talking about the sample rate of the recorded material. Doing 8x (or more) oversampling helps minimize the filtering issues when doing the conversion to analog, but the source material doesn't need to start out at sample rates that are higher than 96k.

[swanlee]

OK so lets take the MFSL rip out of the equation,

 

Here are examples of Roundabout from the Hdtracks release of fragile and Starship trooper from the hdtracks release of The Yes Album

 

 

 

And again here is the AF SACD from CTTE

 

 

 

Something is really wrong with the supposed master tapes of CTTE, two albums recorded BEFORE CTTE by Yes show full high res information up to 46K while CTTE shows a serious buzzcutt at 22K and some type of signal at 30K.

 

I honestly don't see how one can deny that either the master tapes were not used in the recent releases of CTTE or someone really screwed up at some point in the process of mixing\mastering of CTTE which did not happen in the previous 2 Yes albums

[Jud]

Thank you for that clarification - Doing 8x (or more) oversampling helps minimize the filtering issues when doing the conversion to analog, but the source material doesn't need to start out at sample rates that are higher than 96k.

 

Unless you want to minimize the need for sample rate conversion (a topic for another thread, I think).

[Julf]

Unless you want to minimize the need for sample rate conversion (a topic for another thread, I think).

 

Indeed a topic for another thread

[Julf]

I honestly don't see how one can deny that either the master tapes were not used in the recent releases of CTTE or someone really screwed up at some point in the process of mixing\mastering of CTTE which did not happen in the previous 2 Yes albums

 

Is "Never ascribe to malice that which is adequately explained by incompetence" maybe appropriate here?

[swanlee]

Is "Never ascribe to malice that which is adequately explained by incompetence" maybe appropriate here?

 

 

Maybe but their still is some deceit going on as I still don't think these recent CTTE releases came from the real original analog master tapes even though they were claimed to be from them.

 

Even throwing out the freq response of the MFSL version which MFSL released in 1982 it sounds much much better then two recent high res releases. Their are details and a better tonal balance in the MFSL rip that are completely lost and missing on both the HDtracks and AF SACD, It should not even be close.

[bdiament]

Hi Julf,

 

...I would love to hear your definition for dynamic range. The one I have quoted is the one used by engineers in every field I know.

 

The one you quoted describes the *theoretical* dynamic range. In the case of something like 16-bit digital audio, it completely ignores the very steep increase in distortion at low signal levels. This means it is completely accurate for describing the *potential* of the format, as long as one is looking at test signals and ignoring the integrity of what is captured. As soon as you switch to music and the components that many listeners will consider important (such as harmonic structure of sounds, low level detail, spatial cues in recordings that contain them, etc.), what happens is that what gets captured at low levels is missing a great deal of the information. As a result, I would not consider the lower part of 16-bits theoretical dynamic range to reach even remotely close to its theoretical noise floor (at least as far as music or any sound where fidelity is the goal). So in the real world, for high fidelity audio recordings, I think CD has a useful, i.e. *real* dynamic range of somewhat less than 40 dB.

 

 

 

...But we don't measure dynamic range by measuring what a listener experiences. We measure it by comparing the loudest possible signal to the quietest possible signal.

 

Again, that is *theoretical* dynamic range, not what is achieved in the real world, unless one chooses to ignore distortion in the resulting signal.

 

 

 

...So if what you are saying is that if you choose not to use the full dynamic range, you don't get the full dynamic range? In that case I have to agree with you...

 

What I'm saying is that if you make that recording at -40, you won't magically get the "40 + 40 = 80 dB" range you suggested but instead, no more than 56 dB.

 

 

 

...Let's not confuse the dynamic range of the source material with the dynamic range of the media or format. According to your logic, the compressed heavy metal should sound better, because it (having a much smaller dynamic range) has much more bits available to represent the quieter parts (because they aren't very quiet). So your Keith Johnson recording, with passages that are down to -40 dBFS, only has 16 out of 24 bits available to represent it, while the heavy metal has 22 out of 24 bits available to represent it, assuming it is compressed down to 6 dB of dynamic range.

 

No, what I'm saying is that the only way to use the full *theoretical* dynamic range of a 16-bit medium is to compress the audio so the level never goes below -6.02 dB. Obviously (I hope), this has nothing at all to do with sounding "better" (I'd bet quite the opposite). The *parts* of the Keith Johnson recording at -40 would be encoded at 10-bits on a 16-bit format and at 18-bits with a 24-bit format. Again, still more *resolution* (clean capture) than 16-bits at full level.

 

Now, you say you have compared recordings at different levels at both 16-bit and 24-bit and found them to sound identical. My experience has differed in that I find what I deem very significant differences as the word length is increased (or put another way, very serious losses engendered by the shorter word lengths).

 

The reason I posted in this thread is to say that in my experience, the differences are much greater than simple noise floor or theoretical dynamic range and in fact manifest themselves quite plainly in terms of resolution (or distortion, depending on which side of the coin we're viewing: whether we look at the strengths or the weaknesses). I don't have a noise problem with 16-bit and I have yet to make a recording (or hear one made by someone else) that even approaches 96 dB of dynamic range. Yet to my ears, 16-bit is hopeless inadequate in comparison to properly done 24-bit because of its low resolution. (I say "properly done" 24-bit because I have heard plenty of hardware and software that is not "bit clean" in the low order bits. Just as many converters with 192k in the specs don't perform up to the level of others at the 4x rates.)

 

If the two word lengths produce equivalent sonic results as far as you are concerned, I would hope we can agree to disagree.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Julf]

Hi, Barry,

 

The one you quoted describes the *theoretical* dynamic range.

 

It's not just theoretical. It is the maximum achievable range, even in practice. If you choose to leave some of that range unused, that is your choice, but it doesn't make the range that is available, should you choose to use it, to go away.

 

In the case of something like 16-bit digital audio, it completely ignores the very steep increase in distortion at low signal levels.

 

Just like with tape, and any other media, as the distortion doesn't increase in an absolute sense, it's just that the same very small distortion/noise becomes a greater proportion of the smaller signal. The distortion and noise is still at the same, low, and practically inaudible level unless you crank up the volume to unrealistic levels.

 

I think CD has a useful, i.e. *real* dynamic range of somewhat less than 40 dB.

 

You are of course entitled to your opinion, but it is your own personal (and arbitrary) definition of dynamic range. I am curious about how you measure it - what is your criteria for "loudest possible" and "quietest possible", if it isn't the industry standard "full scale" and "1 bit"?

 

What I'm saying is that if you make that recording at -40, you won't magically get the "40 + 40 = 80 dB" range you suggested but instead, no more than 56 dB.

 

Right. So if you choose to leave 40 dB of the dynamic range unused, you get 40 dB less total dynamic range. I agree.

 

If the two word lengths produce equivalent sonic results as far as you are concerned, I would hope we can agree to disagree.

 

Happy to agree to disagree. I have no disagreement with you hearing a difference between 16-bit and 24-bit - if you hear it, then great! I know I don't hear it in controlled double-blind tests. What I do disagree with is your subjective, non-standard redefinition of the standard industry and engineering term "dynamic range".

[bdiament]

Hi Julf,

 

...It's not just theoretical. It is the maximum achievable range, even in practice...

 

...What I do disagree with is your subjective, non-standard redefinition of the standard industry and engineering term "dynamic range".

 

Okay, we disagree.

What I believe you continue to miss in what I'm saying is that I don't consider it real dynamic range if the lower half of the theoretical dynamic range (as measured from noise floor to max allowable level) is distorted. You say that 16-bit and 24-bit sound identical to you and from this I take that you have not heard what I'm describing in your controlled, double-blind comparisons. As such, I can understand that you would not deem it significant in evaluating dynamic range.

 

I'm not at all talking about a larger percentage of what is heard being due to closer proximity to the noise floor.

I'm talking about missing information. (That is all I've written about in this thread: It isn't about signal-to-noise ratio and it isn't about dynamic range. It is purely about the distortion that results when important information in the input signal is not captured due to inadequate resolution of the capture format.)

 

In my experience, no analog format, even using a cheap cassette machine, suffers from this. It only occurs with digital formats of inadequate word length.

 

You can call that "subjective" if you like. I would not disagree but then, I consider belief in numeric representations no less subjective.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Julf]

Hi, Barry,

 

What I believe you continue to miss in what I'm saying is that I don't consider it real dynamic range if the lower half of the theoretical dynamic range (as measured from noise floor to max allowable level) is distorted.

 

I am not missing that. I totally get that that is what you are saying. Again, I simply disagree. What I believe you continue to miss is that that is not how the dynamic range of a media is measured and defined.

 

As such, I can understand that you would not deem it significant in evaluating dynamic range.

 

In this case it is not just me. I would love to hear how you suggest measuring the dynamic range of a media. What is an acceptable level of distortion? How do you measure the lowest (in your view) usable level?

 

Even better would be a reference or pointer to any sort of industry standard specifying those things.

 

The AES and other industry organisations seem to agree with my view, with the additional proviso that the dynamic range of a 16-bit digital system can be even larger than the 96 dB maximum signal-to-noise ratio, as dithering allows you to record signals that are below the 1-bit noise floor. Thus while I agree that signal-to-noise ratio and dynamic range are two different things, the more relevant point is that the dynamic range of a 16-bit system is *greater* than the signal-to-noise ratio, not less.

 

In my experience, no analog format, even using a cheap cassette machine, suffers from this. It only occurs with digital formats of inadequate word length.

 

If you try to record a piece of music (or even a sine tone) at -60 dB on a cheap cassette machine, and then play it back at +60 dB gain, you will see a lot of noise and distortion. Please try it for yourself.

 

You can call that "subjective" if you like. I would not disagree but then, I consider belief in numeric representations no less subjective.

 

In this specific situation, I was using "subjective" to signify a criteria that is held valid by you personally, but not necessarily accepted by the industry.

[bdiament]

Hi Julf,

 

...I am not missing that. I totally get that that is what you are saying. Again, I simply disagree. What I believe you continue to miss is that that is not how the dynamic range of a media is measured and defined.

 

I haven't missed what you said. I'm just saying doing it in such a simplistic way (yes, even though it is how the industry does indeed spec dynamic range) is missing what can actually be delivered in the real world. In other words, I'm saying what is theoretical is not at all the same, in my experience, as the practical reality.

 

 

...If you try to record a piece of music (or even a sine tone) at -60 dB on a cheap cassette machine, and then play it back at +60 dB gain, you will see a lot of noise and distortion...

 

What you won't *hear* is the stripped harmonics and obscuration of details that even the very best 16-bit systems engender. Record that same sound at -60 on your 16-bit system and you've used 6-bits. With a 24-bit format, we've got 14-bits. You've said they sound identical to you in your controlled, double-blind tests. They sound *very* different to me.

 

 

...In this specific situation, I was using "subjective" to signify a criteria that is held valid by you personally, but not necessarily accepted by the industry.

 

I meant it in both situations. ;-}

 

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Julf]

I haven't missed what you said. I'm just saying doing it in such a simplistic way (yes, even though it is how the industry does indeed spec dynamic range) is missing what can actually be delivered in the real world. In other words, I'm saying what is theoretical is not at all the same, in my experience, as the practical reality.

 

And as I have said, I of course have no issue with you describing your own subjective experience, so I think we can finally move on. Thank you for an interesting exchange!

[vinylphile]

 

You can call that "subjective" if you like. I would not disagree but then, I consider belief in numeric representations no less subjective.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

 

That's called hitting the nail on the head!

[elcorso]

Hi Julf,

 

 

 

You can call that "subjective" if you like. I would not disagree but then, I consider belief in numeric representations no less subjective.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

 

Well said, Barry!

 

Numeric representations are today 'as infallible' that a very famous European economist is blaming the enormous economic and monetary problem in Europe miscalculations made by ​​Excel sheets...

 

Kind regards,

 

Roch

[REShaman]

Well said, Barry!

 

Numeric representations are today 'as infallible' that a very famous European economist is blaming the enormous economic and monetary problem in Europe miscalculations made by ​​Excel sheets...

 

Kind regards,

 

Roch

 

Roch, that reconciles for me.

 

Barry, you not only have expertise as a producer and engineer, you are also a master of repartee which has an undeniable fidelity.

 

Julf, with respect, where is this going and for how much longer? Your turn.

 

Best,

Richard

[Julf]

Julf, with respect, where is this going and for how much longer? Your turn.

 

And as I have said, I of course have no issue with you describing your own subjective experience, so I think we can finally move on.

 

The birds! The birds!

[elcorso]

Roch, that reconciles for me.

 

Barry, you not only have expertise as a producer and engineer, you are also a master of repartee which has an undeniable fidelity.

 

Julf, with respect, where is this going and for how much longer? Your turn.

 

Best,

Richard

 

This is my hope Richard... I guess I wouldn't listen anymore "Close to the Edge"!

 

BTW, a very symbolic name of the situation in Europe...

 

Kind regards,

 

Roch

[Don Hills]

... In other words, I'm saying what is theoretical is not at all the same, in my experience, as the practical reality.

 

What you won't *hear* is the stripped harmonics and obscuration of details that even the very best 16-bit systems engender. Record that same sound at -60 on your 16-bit system and you've used 6-bits. With a 24-bit format, we've got 14-bits. You've said they sound identical to you in your controlled, double-blind tests. They sound *very* different to me. ...

 

Barry,

I will assume at this stage that you fully understand the theory, in that the effect you hear can not, in theory, occur. The maths of sampling (with dither) is solid - there is no "information loss" as the bit depth decreases, just a worsening of the noise. The information is still there, mixed with the noise. This is no different than any other information channel.

If you do believe that digital sampling is a special case, as compared to, for example, a tape machine, please say so, and stop reading now.

 

On the other hand, if you do understand the theory, and you do hear a difference, then what you hear may be real or imagined. I won't insult you by implying it is imagined - I know from personal experience that even differences that turn out to be illusory are perfectly real to me, and I have no doubts that you are better able to discern differences than I am. I'm willing to discuss what may cause practical implementations to fall short of theoretical perfection, though I suggest we do it in a new thread.

 

How about "Reduction in detail as bit depth is reduced"?

Any better suggestions? Anyone?

 

Regards,

Don.

[bdiament]

Hi Don,

 

...there is no "information loss"...

 

Tell that to the audio.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Julf]

"A witty repartee always trumps any facts or science". Applies to much more than just audio.

[REShaman]

"A witty repartee always trumps any facts or science". Applies to much more than just audio.

 

The words! The words!

 

Dear Julf and Barry,

 

Not knowing whether or not my earlier remarks fueled an unintended offense or perspective, in any case, I'll offer my apologies to you both for interfering in something I know nothing about.

 

My earlier remarks were specifically about subjectivity (the idea of it as it is applied as a substitute for truth which in my view is subject to change and awareness in general) and not about specific facts and science nor intended to take sides. Taking sides would be presumptive of me for I claim no competence or understanding about these matters nor meant to referee what (here I go again) has since gone over the edge into OT galore.

 

Obviously I am not commenting on your right to engage in whatever this "discussion" is about. I respect you both.

 

Have a productive time sorting out what this is about. Don's prudent suggestion to start a new thread, practically even theoretically, has currency in my view.

 

TTFN,

Richard

[Julf]

Richard,

 

Not knowing whether or not my earlier remarks fueled an unintended offense or perspective, in any case, I'll offer my apologies to you both for interfering in something I know nothing about.

 

No offence taken.

 

My earlier remarks were specifically about subjectivity (the idea of it as it is applied as a substitute for truth which in my view is subject to change and awareness in general) and not about specific facts and science

 

And my comments about subjective standards had nothing to do with the apparently fanatically religious issue of "subjectivists" vs. "objectivists". My point was that if you are going to use numbers to justify your argument, you'd better use them in the precise, standard way the industry is using them, rather than some personal, arbitrary and relative (so "subjective" in that sense) standard.

 

If somebody asks you "what is the top speed of your car, and how many miles per gallon do you get?", the answer "I never drive more than 60, and I fill up when the petrol gauge shows half empty" doesn't really help...

 

Don's prudent suggestion to start a new thread, practically even theoretically, has currency in my view.

 

Agree.