[JaZZ]

Quote:

Originally Posted by WmAx The graph is misleading. The waveforms demonstrated are not represenative of what is output from an actual standard A-D-A process used for audio recording that sampled the frequencies specified. The graph is a theoretical one that does not account or specify a realistic situation.

Yes, in reality, the signal shape is that of stair steps. But that doesn't change anything. The frequency modulation is real with time-optimized filters such as from Wadia or with filterless designs such as with the Ack dAck and causes the well-known drop-off. Also the filter resonance is real and can be seen on every square wave, unless it's from a time-optimized filtering.

 

[Glassman]

Quote:

Originally Posted by JaZZ Also the filter resonance is real and can be seen on every square wave, unless it's from a time-optimized filtering.

in which case it looks more like a triangle than a step square response..

 

[JaZZ]

Quote:

Originally Posted by Glassman in which case it looks more like a triangle than a step square response..

At least a 1-kHz square wave filtered through a time-optimized 22-kHz low-pass filter looks quite reasonable -- and better than with frequency-optimized filtering. Higher frequencies do change the waveform towards triangles or rather sine shape of course.

 

[WmAx]

Quote:

Originally Posted by JaZZ Yes, in reality, the signal shape is that of stair steps. But that doesn't change anything. The frequency modulation is real with time-optimized filters such as from Wadia or with filterless designs such as with the Ack dAck and causes the well-known drop-off. Also the filter resonance is real and can be seen on every square wave, unless it's from a time-optimized filtering.

True, but with conventional anti-aliasing techniques on the DAC(not considerin oddbadd designs which are questionable at best) the output is a clean sine wave. As for transient errors of standard aliasing filters; no one has shown this to be discretely audible for high frequencies in controlled tests of which I am aware. If you have references to controlled listening tests examining the audibilty of transient response of a typical 44.1 PCM anti-alias filter, please provide.

-Chris

 

[JaZZ]

Quote:

Originally Posted by WmAx True, but with conventional anti-aliasing techniques on the DAC(not considerin oddbadd designs which are questionable at best) the output is a clean sine wave.

And that's actually the problem: Do I want a clean sine wave when the recording sais there had to be transient components instead? Music doesn't consist of sine waves -- they just serve as measuring signals.

Quote:

As for transient errors of standard aliasing filters; no one has shown this to be discretely audible for high frequencies in controlled tests of which I am aware. If you have references to controlled listening tests examining the audibilty of transient response of a typical 44.1 PCM anti-alias filter, please provide.

For the next time: don't ask me for «controlled listening tests» -- I'm not the one to care for such things.

But of course you already know...

The existence of the amplitude modulation doesn't seem to be commonly known -- you seem to be surprised yourself --, so I wouldn't wonder if it's not been addressed so far by in-depth analyzes and listening tests. I even understand that it's not been advertized by the advocates and manufacturers of time-optimized filters and filterless DACs. But is there any reason to suppose that it's not audible? After all it's in the audible range, and the effect is pronounced enough. However, one side effect of the phenomenon is clearly audible in any case: a high-frequency drop-off.

But let's return to the «ideal» case of a compensated frequency response, with the side effect of the filter resonance. Although actually it's the other way around: the filter resonance with the side effect of frequency-response linearization. As stated, it implies the rather unfavorable property of «restoring» clean sine waves out of transients. Again: what is it that prevents one from hearing this phenomenon? It's the same as above, just with opposite signs. Now it's important to know that in the range above 10 kHz music contains a lot of transient events, usually much more than continuous tones. So it's logical to assume that the filter's smearing function will have audible effects.

I just can report what I hear from the CD layer compared to the SACD layer. With various players I've experienced a consistent sonic signature: overtones appear as slightly smeared and partially shrill from the CD layer, whereas the SACD layer sounds more accurate, with sharper edges and without the CD layer's tendency towards shrillness, and overall it shows the finer resolution. Still CDs can sound quite good on good equipment, but the SACD format has the edge to my ears.

What's definitely true is that there's no reason for upholding the myth of the redbook CD as THE perfect format.

 

[WmAx]

Quote:

Originally Posted by JaZZ And that's actually the problem: Do I want a clean sine wave when the recording sais there had to be transient components instead? Music doesn't consist of sine waves -- they just serve as measuring signals.

Music is made up of a complex sum of sine waves with the rare assymetrical components here and their, of which all the important(audible components) parts are captured within the 22khz bandwidth afforded by 44.1kHz sample rate.

Quote:

The existence of the frequency modulation doesn't seem to be commonly known -- you seem to be surprised yourself --, so I wouldn't wonder if it's not been addressed so far by in-depth analyzes and listening tests.

I am certainly not an expert(far from it) in digital sampling - but their are straigtforward and well-explained papers on these subjects as well as perceptual research supporting specific parmaters of requirement. You will have to be more specific as to what you mean by modulation-I have not come across such data as being an issue. While it is clear that some reconstruction filters may cause inter-modulated type of interference(aliasing), such as if they do not contain an adequate anti-alias section(Audio Note makes a DAC such as this if my memory is correct). The graphic you showed illustrated no such modulations; the graphic essentially showed the time and amplitude coordinates of a 44.1khz sample rate of a sine wave and drew a dot-to-dot line between these coordinates--this is certainly ignoring the signal reconstruction stage, which when implemented properly, yields correct reproduction of the sine wave frequencies specified in that graphic.
Quote:

I even understand that it's not been advertized by the advocates and manufacturers of time-optimized filters and filterless DACs. But is there any reason to suppose that it's not audible?

1st, how about referencing a paper or objective article on the specific modulation effect as it supposedly pertains to a conventional DAC -- or are you only referring to the oddball DACs?

Quote:

As stated, it implies the rather unfavorable property of «restoring» clean sine waves out of transients. Again: what is it that prevents one from hearing this phenomenon? It's the same as above, just with opposite signs. Now it's important to know that in the range above 10 kHz music contains a lot of transient events, usually much more than continuous tones. So it's logical to assume that the filter's smearing function will have audible effects.

The ear is exceedingly less senstivie at higher frequencies. Additionally, if you analyse the spectrum of music that contains significant high frequency content, it is magnitudes lower in amplitude then the midrange and bass. Redbook was designed to be transparent to human perception; the perceptual research that was used to design this was considerable if you invesitigate the issue.

Quote:

I just can report what I hear from the CD layer compared to the SACD layer. With various players I've experienced a consistent sonic signature: overtones appear as slightly smeared and partially shrill from the CD layer, whereas the SACD layer sounds more accurate, with sharper edges and without the CD layer's tendency towards shrillness, and overall it shows the finer resolution. Still CDs can sound quite good on good equipment, but the SACD format has the edge to my ears.

It's not a fair comparision. 1. The levels have to set within 0.1dB between both. 2. The SACD or DVD-A versions are not identical to the RBCD releases when analysed, based on the examples so far as published by Audioholics and some other online publications. 3. An objective blind testing methodology has to be used to garner fair results.

Quote:

What's definitely true is that there's no reason for upholding the myth of the redbook CD as THE perfect format.

I don't know who has heralded RBCD as THE perfect format; it certainly was not me. First of all, it has only support for 2 channels.

-Chris

 

[eyeteeth]

Quote:

Originally Posted by Jahn IT IS ALL TRUE!

You're Right! It is true.

I did four CDRs and they all consistently displayed the same differences from the original CDs, and obviously so. Greater warmth and fullness was the big obvious, things were a little more three dimensional and a little more information was also apparent, more detail. Switching back to the original CD and they were brighter and flatter sounding. There is no mistaking the two.

The first CD I did was an awful recording of music that is wonderful to my taste, In the Nightside Eclipse by Emperor. Followed by two other metal favorites that I always wished were of a higher quality. Beyond the Veil by Tristania and DCA by Dimmu Borgir. All three were improved.

Curious about the difference if at all of a fine recording I did Yankee Foxtrot Hotel by Wilco and sure enough it was better. Increased warmth wasn't as obvious but a little extra three dimensinality and information could be heard. Bass was better with more texture.

I used Exact Audio Copy to rip but it didn't have support for my older LG burner, so I used Burn4free which has been onboard for some time. Black Memorex CDRs, X4 burn speed and a minimization of all other program activities.

It's certainly not the miraculous improvement or anywhere near what the dude in my original post described but it is unmistakable improvement. And damn cheap too!

Edit: for reference
http://www.genesisloudspeakers.com/w...lack_CDsII.pdf

http://www.6moons.com/industryfeatures/eac/eac.html

 

[JaZZ]

What may have caused confusion is the term «frequency modulation». It actually should mean «amplitude modulation». Sorry! I've meanwhile replaced it in my previous posts.

Quote:

Originally Posted by WmAx Music is made up of a complex sum of sine waves with the rare assymetrical components here and their, of which all the important(audible components) parts are captured within the 22khz bandwidth afforded by 44.1kHz sample rate.

So what? We're all made of atoms. So are we atoms? The ability of a system to create sine waves doesn't automatically imply the ability to create transients and complex waveforms, which is what music virtually exclusively consists of.

Quote:

You will have to be more specific as to what you mean by modulation - I have not come across such data as being an issue.

The «modulation» we were talking about before is the easily recognizable amplitude modulation (now corrected term). The reason why it's commonly not seen as an issue is that it's eliminated with common «reconstruction filters».

Quote:

While it is clear that some reconstruction filters may cause inter-modulated type of interference (aliasing), such as if they do not contain an adequate anti-alias section (Audio Note makes a DAC such as this if my memory is correct).

No, it's neither aliasing nor it's «caused» by any «inadequate» filter. It's a format-inherent issue caused by the interference between sampling rate and signal frequency, as you could easily reconstruct yourself by looking at the graph. It's just that those «inadequate» filters designed to leave transients intact («time-optimized») as well as filterless DACs logically also leave the amplitude modulation intact.

Quote:

The graphic you showed illustrated no such modulations; the graphic essentially showed the time and amplitude coordinates of a 44.1khz sample rate of a sine wave and drew a dot-to-dot line between these coordinates --this is certainly ignoring the signal reconstruction stage, which when implemented properly, yields correct reproduction of the sine wave frequencies specified in that graphic.

Of course the reconstruction filter is deliberately ignored. The curve shows the signal after D/A conversion before low-pass filtering -- as described from the beginning.

Quote:

1st, how about referencing a paper or objective article on the specific modulation effect as it supposedly pertains to a conventional DAC -- or are you only referring to the oddball DACs?

Yes, as already described I'm referring to «oddball DACs» -- such with time-optimized filters. (Why do I have to repeat all this?)

Quote:

The ear is exceedingly less sensitive at higher frequencies. Additionally, if you analyse the spectrum of music that contains significant high frequency content, it is magnitudes lower in amplitude then the midrange and bass. Redbook was designed to be transparent to human perception; the perceptual research that was used to design this was considerable if you invesitigate the issue.

I'm sure redbook was designed to provide a good enough sound quality for the majority of the potential customers on the basis of the then technology. It's still satisfying for a broad majority (myself included), but it's not perfect. I don't think I'm buying your «doesn't matter because high frequencies aren't so important» argument.

Quote:

It's not a fair comparision. 1. The levels have to set within 0.1dB between both. 2. The SACD or DVD-A versions are not identical to the RBCD releases when analysed, based on the examples so far as published by Audioholics and some other online publications. 3. An objective blind testing methodology has to be used to garner fair results.

It doesn't really matter that much. The result has been absolutely consistent with various hybrid disks and various players so far. I wasn't even describing the sound as more or less pleasant, just some characteristic sonic properties.

Quote:

I don't know who has heralded RBCD as THE perfect format; it certainly was not me.

Good to know.

But I for one am not referring to the multichannel option, as you may have noticed.

 

[WmAx]

Quote:

Originally Posted by JaZZ What may have caused confusion is the term «frequency modulation». It actually should mean «amplitude modulation». Sorry! I've meanwhile replaced it in my previous posts.

Thank you for the clarification.

Quote:

So what? We're all made of atoms. So are we atoms? The ability of a system to create sine waves doesn't auromatically imply the ability to create transients and complex waveforms, which is what music virtually exclusively consists of.

The waveforms are a composition of sine waves.

Quote:

The «modulation» we were talking about before is the easily recognizable amplitude modulation (now corrected term). The reason why it's commonly not seen as an issue is that it's eliminated with common «reconstruction filters».

The coordinates given, when used with the intended(correct) reconstruction filters do not have this amplitude modulation issue, as you recognize and point out. I don't see this as an issue since it does not pertain to conventional reconstruction filters. The nesecarry data is present to reconstruct the signals when used with properly engineered reconstruction filters.

Quote:

Of course the reconstruction filter is deliberately ignored. The curve shows the signal after D/A conversion before low-pass filtering -- as described from the beginning. Yes, as already described I'm referring to «oddball DACs» -- such with time-optimized filters. (Why do I have to repeat all this?)

You have to repeat 'this' because I was not sure why you were making a deal of out non-existant errors on conventional DAC units. Now it is clear you were making your point based on radical DAC designs. I wanted to confirm.

Quote:

I don't think I'm buying your «doesn't matter because high frequencies aren't so important» argument.

Perceptual research has repeatedly shown(when careful controls were established) that test subjects have degrading sensetivity to high frequencies and that data above the RBCD capability has [1][2]not been shown to make any audible difference to the humans tested. The tests that showed effect of ultrasonic data had vague information about the specific auditory listening section (most commonly cited test is the one by [3]Ooashi) of the tests -- all of the issues when tightly controlled and invesitgated were [2]NOT repeatable in peer review. BTW, it is important to note that the subjects in these tests are nearly always sound professionals and musicians that have training in order to make the tests of [4]higher value then using un-trained listeners. Their have been tests[1][5] that have investigated the potential audible effect of ringing caused by sharp filter slopes as is required for digital reconstruction fiters, with the results failing to demonstrate audibility. These listening tests have been carried out with music and impulse signals.
Quote:

[/i]It doesn't really matter that much. The result has been absolutely consistent with various hybrid disks and various players so far. I wasn't even describing the sound as more or less pleasant, just some characteristic sonic properties.

If their is such a consistant improvement why does no data exist in the form of perceptual testing that holds up to scrutiny? I find it odd that neither DVD-A or SACD's increased 'resolution' was based on credible perceptual research -- if it was, then they have not disclosed this material or I have failed to notice it --- and I routinely look out for these types of perceptual tests.

-Chris

Footnotes
[1] Which Bandwidth Is Necessary for Optimal Sound Transmission?
G. PLENGE, H. JAKUBOWSKI, AND P. SCHONE
JAES, Volume 28 Number 3 pp. 114-119; March 1980

[2] Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components
AES Preprint: 5876
Toshiyuki Nishiguchi, Kimio Hamasaki, Masakazu Iwaki, and Akio Ando

[3] Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect
Tsutomu Oohashi, Emi Nishina, Manabu Honda, Yoshiharu Yonekura, Yo****aka Fuwamoto, Norie Kawai, Tadao Maekawa, Satoshi Nakamura, Hidenao Fukuyama, and Hiroshi Shibasaki4
The Journal of Neurophysiology Vol. 83 No. 6 June 2000, pp. 3548-3558

[4] Selection and Training of Subjects for Listening Tests on Sound Recording Equipment
Soren, Bech
JAES, Vol. 40, 1992

[5] Perception of Phase Distortion in Anti-Alias Filters
Preis, D.; Bloom, P. J.
AES Preprint: 2008

 

[JaZZ]

Quote:

Originally Posted by WmAx You have to repeat 'this' because I was not sure why you were making a deal of out non-existant errors on conventional DAC units. Now it is clear you were making your point based on radical DAC designs. I wanted to confirm.

Come on! I have made my points clear that and why I see conventional reconstruction filters as flawed because of their transient-smearing tendency, and you even gave your comments to the issue sounding like: «The ear is exceedingly less sensitive at higher frequencies. Additionally, if you analyse the spectrum of music that contains significant high frequency content, it is magnitudes lower in amplitude then the midrange and bass. Redbook was designed to be transparent to human perception; the perceptual research that was used to design this was considerable if you invesitigate the issue.»

So I have to repeat the whole story for you:

(quote)
In the case where something like the displayed amplitude modulation is the actual signal to be reproduced, common frequency-optimized filters will not allow this, instead they'll make a continuous sine wave out of it. In the case where a continuous sine wave ist wanted, the time-optimized filter will fail.

(quote)
But let's return to the «ideal» case of a compensated frequency response, with the side effect of the filter resonance. Although actually it's the other way round: the filter resonance with the side effect of frequency-response linearization. As stated, it implies the rather unfavorable property of «restoring» clean sine waves out of transients.

And please note (again!) that this happens in the audible range! Although just in the «unimportant» high-frequency range...

 

[WmAx]

Quote:

Originally Posted by JaZZ In the case where something like the displayed amplitude modulation is the actual signal to be reproduced, common frequency-optimized filters will not allow this, instead they'll make a continuous sine wave out of it. In the case where a continuous sine wave ist wanted, the time-optimized filter will fail.

Fail? If I am to take the waveforms on that graph as literal, then they contain frequencies that exceed 22kHz Nyquist limit, as is observable by the sharp/sudden angled changes in amplitude. This would be the equivalent of expecting RBCD to reproduce a high frequency triangle wave as a triangle. All data that exceeds the Nyquist limit is in effect discarded.

-Chris

 

[JaZZ]

Quote:

Originally Posted by WmAx Fail? If I am to take the waveforms on that graph as literal, then they contain frequencies that exceed 22kHz Nyquist limit, as is observable by the sharp/sudden angled changes in amplitude. This would be the equivalent of expecting RBCD to reproduce a high frequency triangle wave as a triangle. All data that exceeds the Nyquist limit is in effect discarded.

Bull****! -- We've already agreed on that in reality the «waveform», as you call it, meaning the DAC's output signal before filtering, looks like stair steps, not like these triangles. (And of course the raw analog signal before filtering contains lots of ultrasonics!) But it doesn't change anything on the sample and amplitude-value positions which form an amplitude modulation.

But from your perspective you don't have to care about the amplitude modulation anyway... it's inaudible.

Well, I don't think it is, but if you really consider the transient-smearing effect of sharp-edged «reconstruction» filters inaudible, then the amplitude modulation is as well.

«In the case where something like the displayed amplitude modulation is the actual signal to be reproduced, common frequency-optimized filters will not allow this, instead they'll make a continuous sine wave out of it. In the case where a continuous sine wave ist wanted, the time-optimized filter will fail.»

If one effect is audible, the other is as well.

BTW, I don't want to discuss you reference sources and studies -- we wouldn't come to an end. All I want is to demonstrate the possibility that the CD format implies audible flaws.

 

[WmAx]

Quote:

Originally Posted by JaZZ Bull****! -- We've already agreed on that in reality the «waveform», as you call it, meaning the DAC's output signal before filtering, looks like stair steps, not like these triangles. But it doesn't change anything on the sample and amplitude-value positions which form an amplitude modulation.

You stated: "In the case where something like the displayed amplitude modulation is the actual signal to be reproduced, common frequency-optimized filters will not allow this".

The 'displayed' waveforms in your graphic include data that exceeds 44.1khz nyquist, if taken as a literal input signal to a ADC. That was the prompting of my last reply.

If the waveforms sampled are <22kHz in frequency, they should be reproducable with correct amplitude with a conventional DAC reconstruction filter.

-Chris

 

[JaZZ]

Sorry, you're standing completely beside your shoes! Of course the raw analog signal before filtering contains lots of ultrasonics! That doesn't mean the sampled frequencies are above the Nyquist frequency. Oh my...

 

[JaZZ]

You can control the sampled frequencies yourself: If there are less than two sample points on one wave cycle, then the Nyquist frequency is exceeded.