Redbook 44.1 kHz standard: theoretically sufficient timbral resolution?

[Groucho]

Been discussed plenty of times though here and at Computer Audiophile by others very experienced in implementing both.

I am sure there are people who have great experience in implementing an incorrect system, but does it give them any special knowledge?

 

[Don Hills]

Please could you be more specific with that explanation on the 1x DAC filter, because it is not really a reconstruction filter as implemented with interpolating/oversampling DACs.

It's done with an analogue low-pass filter, typically 5th to 8th order. Any one of several different alignments may be used, depending on how steep you want the cutoff. Even an oversampled DAC has (should have) a reconstruction filter at the output. The advantage of oversampling is, of course, that the final analogue filter can be lower order, which relaxes the complexity and component tolerances.

Actually a reconstruction filter is to designed to remove the images and has an influence on phase-timing-frequency, the DSD low pass filter is to remove noise without the same issues (has different one though IMO).
So it is not actually a reconstruction filter - yeah I am being an ass I know but it was due to you saying if a DAC does not have a reconstruction filter it is broken
Been discussed plenty of times though here and at Computer Audiophile by others very experienced in implementing both so happy to change the subject.

Personally, I consider DSD (single bit implementations) to be broken as designed. As such, I only considered PCM DACs when I made my remark about reconstruction filters.

Regards...

 

[Tony Lauck]

Thank you, Tony, for your reply to my question.



Of course there is acoustic energy from instruments above 20 kHz, just like there is a UV component in sunlight. The question is how relevant, if at all, it is for our auditory perception. No systematic evidence indicates that we can hear beyond 20 kHz, just like it is established that we cannot see UV light (unlike bees, who can). We do get sunburns though from UV light, but obviously this has nothing to do with our eyes. I avoid using "authorities" whenever possible. In the case of

From the link you posted, in italics:

Oohashi and his colleagues recorded gamelan to a bandwidth of 60 kHz, and played back the recording to listeners through a speaker system with an extra tweeter for the range above 26 kHz. This tweeter was driven by its own amplifier, and the 26 kHz electronic crossover before the amplifier used steep filters. The experimenters found that the listeners' EEGs and their subjective ratings of the sound quality were affected by whether this "ultra-tweeter" was on or off, even though the listeners explicitly denied that the reproduced sound was affected by the ultra-tweeter, and also denied, when presented with the ultrasonics alone, that any sound at all was being played.

Objection: Here it is not clear if the signal through the extra tweeter did not at all alter the signal in the audible range, either through electronic feedback into the chain or through acoustic vibrations that could affect a) the electronics or b) the speaker system that transmitted below 26 kHz. That listeners explicitly denied that it had an effect, does not necessarily make it so. A lack of interference with the audio signal below 26 kHz would have to be carefully established before scientifically valid conclusions can be drawn.

In a paper published in Science, Lenhardt et al. report that "bone-conducted ultrasonic hearing has been found capable of supporting frequency discrimination and speech detection in normal, older hearing-impaired, and profoundly deaf human subjects." [5] They speculate that the saccule may be involved, this being "an otolithic organ that responds to acceleration and gravity and may be responsible for transduction of sound after destruction of the cochlea," and they further point out that the saccule has neural cross-connections with the cochlea. [6]

Objection: What is not mentioned is that the ultrasound has to be directly fed to the skull, thus with much higher energy than normal, which is irrelevant for regular hearing.

I am open to more systematic results from studies, but so far I see nothing conclusive, and what we have so far is so sporadic that it amounts to little more than anecdotal evidence. Extraordinary claims require extraordinary evidence.



There is no direct evidence that this has to do with the high-frequency content. It could be other technical issues.

We all get to judge what constitutes "direct evidence" and what does not. I consider direct evidence what I can perceive with my own senses as the result of experiences and experiments that I use my own mind to evaluate. I consider indirect evidence that presented by other people, but this is diminished by my personal judgement (for better or worse) as to the competence, integrity and open mindedness of those people. I have spent dozens (if not hundreds) of hours comparing different digital formats and have found that I can often tell differences between high resolution formats and 44.1 kHz versions thereof, despite controlling for every variable that I can think of.

I suspect that many NOS advocates are enjoying the high frequency roll-off provided by the zero order NOS hold. This can serve as a tone control and may improve many recordings (on a too bright system) or too bright recordings (on a neutral system). I don't believe that swapping source components is a good way to fix problems elsewhere in one's system (e.g. speaker - room interactions).

 

[Orb]

It's done with an analogue low-pass filter, typically 5th to 8th order. Any one of several different alignments may be used, depending on how steep you want the cutoff. Even an oversampled DAC has (should have) a reconstruction filter at the output. The advantage of oversampling is, of course, that the final analogue filter can be lower order, which relaxes the complexity and component tolerances.



Personally, I consider DSD (single bit implementations) to be broken as designed. As such, I only considered PCM DACs when I made my remark about reconstruction filters.

Regards...

And I know you know the issue with the analogue low-pass filter that you just described
If it was simple then many NOS DAC designers would do just that, and I know this from experience of those who spoke about their development of a good audio product around say the TDA1541.
Thorsten Loesch implemented two options with his NOS DAC and the one with the analogue filter as you mention still has issues when looking at it with modern measurements techniques, and I know of other designs where the measurements are still bad with their analogue filter implementation.

Don, like the much earlier LOL I have to do it again with your last statement, it is not directed at you but the fact I am sure such a statement will annoy quite a few (including a fair few engineers)
Nothing like prodding the ants nest
But at least I can see the statement is consistent with the previous one, and can appreciate it is your POV, like I said even I see issues or challenges with all approaches discussed and some have more cons than others.
Cheers
Orb

 

[Orb]

I suspect that many NOS advocates are enjoying the high frequency roll-off provided by the zero order NOS hold. This can serve as a tone control and may improve many recordings (on a too bright system) or too bright recordings (on a neutral system). I don't believe that swapping source components is a good way to fix problems elsewhere in one's system (e.g. speaker - room interactions).

Quite a few well designed oversampling DACs still have the high frequency roll-off, with some following the sin(x)/x.
There are good reasons to implement minimum phase/slow roll-off even if it has weaker alias rejection, albeit at the cost of accurate frequency domain 20hz-20khz (for CD)
Cheers
Orb

 

[Orb]

I am sure there are people who have great experience in implementing an incorrect system, but does it give them any special knowledge?

More than nearly everyone on this forum I would say
And no need for anyone to mention the usual 'appealing to higher authority type as fact' does not make it right.
The ones I am talking about or considering are excellent engineers.
Cheers
Orb

 

[Groucho]

More than nearly everyone on this forum I would say
And no need for anyone to mention the usual 'appealing to higher authority type as fact' does not make it right.
The ones I am talking about or considering are excellent engineers.

I hope they stick to audio and never move into automotive...

 

[Orb]

I hope they stick to audio and never move into automotive...

Groucho that is a rather harsh statement.
It comes across as a bit derogatory, which reminds me when I used to belong to HA years ago and the behaviour of some there.
To show why it is bad; we have Don Hills who says DSD is broken, but then you have others who are world class engineers who manufacturer DSD products such as Bruno Putzeys with Grimm Audio and the study at Philips research labs looking at the requirement for transparent transcoding between PCM and DSD.
So I guess we should point a derogatory comment to one of them that they never move into automotive.....
I could use more nuanced examples, but it serves its purpose I feel in why it is wrong to say what you did.
In the same way it is far from simple to create a steep high order analogue x1oversampling filter for CD frequency range, and why some looked at trying alternative solutions (which still continue to this day on diyaudio for those engineers curious with NOS chips such as the TDA1541), and measurements of NOS DACs with analogue filters has shown in the past.
But anyway the challenges and pros/cons relating to digital go well beyond NOS DACs, which are niche to say the least.

Cheers
Orb

 

[Al M.]

I have spent dozens (if not hundreds) of hours comparing different digital formats and have found that I can often tell differences between high resolution formats and 44.1 kHz versions thereof, despite controlling for every variable that I can think of.

Tony, I don't doubt your observations in the least.

The question, however, after 4 decades of digital audio is still the same: is

a) the application of the Nyquist theorem in the CD format, leading to the frequency cut-off of the format, somehow intrinsically flawed, or

b) does the difference in performance between hi-rez and CD lie in the current practical technical implementations?

If an audiophile in 1985 would have heard the musical resolution from current top-level CD playback they would have been hugely surprised or rather, shocked, about the actual potential of the CD format which was not at all apparent at that time, with the implementations (recording and playback) available back then. Also, it is well known and agreed upon by audiophiles that CD on top-level playback still sounds better than hi-rez on second-tier playback, which again sounds better than CD on that playback equipment. In other words, it is all about implementation, and hi-rez is apparently more easily implementable to a satisfying degree than CD is. This is not surprising given that the CD format operates really on the edge of acceptable parameters, and hi-rez simply has more room to spare when it comes to its practical implementation -- that greater ease of audibly proper application would obviously be an argument in favor of hi-rez; theoretical arguments appear more debatable.

Perhaps there is indeed a clear audible ceiling to the resolution and transparency of CD playback, compared to hi-res, that we simply cannot get beyond, but after 4 decades of digital audio this still remains an open question: apparently, ever better CD playback still reveals more and more resolution from that format that hitherto was not thought possible or heard of. A ceiling apparently has not yet been reached, even though audible ceilings to the transparency of the CD format have been declared many times in the past.

Certainly, one might argue, as Amir does, that the bit depth of CD, its dynamic range, is not sufficient, which is a completely different issue than the 44.1 kHz sampling rate. Yet the question still remains open how much this matters in practice, given that, according to some engineers who have posted numbers here, the practical dynamic range of any playback system so far does not exceed a value somewhere around 80 dB. Certainly, manipulating the signal at 16 bit depth through all steps until the final master is problematic because you lose resolution at every step, but these days nobody does that anymore. The real question remains how much actual loss of resolution occurs for all practical purposes when limiting the bit depth down to 16 bit for the final CD after the signal has been processed throughout at higher bit depth up until that point.

In practice, a signal that is, for all practical purposes, band limited to frequencies below (say) 20000 Hz can be reproduced (for all practical purposes) by sampling the signal at 44100 Hz. For example, just last week I took a 88400/24 recording (of an orchestra including trumpets, cymbals and violins) and band limited it so that there was no energy above 18 Khz above -150 dBfs. This became my reference signal. I then converted the reference signal to 44100 kHz and resampled it back to 88200. The result was a signal that agreed to the reference, at every sample, subject to a worst case error of 2 least significant bits out of 24 bits. (This was due to roundoff error in converting the high precision values to 24 bits). The average difference between the two signals was -150 dBfs. If I had wanted, I could have changed the parameters of my experiment and gotten better results. I did this experiment using the commercial software (iZotope RX4 advanced) and my preferred settings that I normally use when converting a high resolution studio master back to the 44.1 kHz sampling rate for release on CD.

I could not hear a difference between the reference file and the file that had been passed through the 44.1 kHz sampling rate. I took the error signal (difference signal) and listened to it. It was still silent after I boosted it by 60 dB. Only after I boosted it by 100 dB was there an obvious difference in the form of white noise. So yes, in practice, taking audio signals that have no energy above 18 kHz and converting them to 44.1 kHz is possible without any audible loss (at least to most people's ears).

If, for example an original 88/24 recording sounded better without bandlimiting than the same file bandlimited to 20 kHz (here you went down to 18 kHz), then there may be technical reasons for the degradation of sound during bandlimiting other than the frequency limitation. Looking to the more extended frequency response of hi-rez for an answer, when it comes to either the theoretical limits of CD playback or an explanation of inferior sound during current implementations of the CD format vs. hi-rez, is in my view questionable. There is just not sufficient, generally accepted, scientific evidence that our hearing, or any musical perception for that matter, extends beyond 20 kHz. Again, extraordinary claims require extraordinary evidence.

 

[rbbert]

This last statement of Al's is just not accurate. There is plenty of evidence that people perceive frequencies above 20 kHz, just not always through the hair cells of the inner ear. How high bone conduction, skin vibration sensors and the like go is somewhat controversial, with some outlying scientists claiming perception to 100 kHz. The US armed forces have plenty of data (now probably classified at some level, although it didn't used to be) on young people with standard audio (i.e., ear) perception into the low 20 kHz's, just as many young people may have uncorrected visual acuity of 20/10 or better, or can run 100 meters in under 10 seconds (held to be physically imposible early in our lifetimes).

 

[Tony Lauck]

Tony, I don't doubt your observations in the least.

The question, however, after 4 decades of digital audio is still the same: is

a) the application of the Nyquist theorem in the CD format, leading to the frequency cut-off of the format, somehow intrinsically flawed, or

b) does the difference in performance between hi-rez and CD lie in the current practical technical implementations?

If an audiophile in 1985 would have heard the musical resolution from current top-level CD playback they would have been hugely surprised or rather, shocked, about the actual potential of the CD format which was not at all apparent at that time, with the implementations (recording and playback) available back then. Also, it is well known and agreed upon by audiophiles that CD on top-level playback still sounds better than hi-rez on second-tier playback, which again sounds better than CD on that playback equipment. In other words, it is all about implementation, and hi-rez is apparently more easily implementable to a satisfying degree than CD is. This is not surprising given that the CD format operates really on the edge of acceptable parameters, and hi-rez simply has more room to spare when it comes to its practical implementation -- that greater ease of audibly proper application would obviously be an argument in favor of hi-rez; theoretical arguments appear more debatable.

Perhaps there is indeed a clear audible ceiling to the resolution and transparency of CD playback, compared to hi-res, that we simply cannot get beyond, but after 4 decades of digital audio this still remains an open question: apparently, ever better CD playback still reveals more and more resolution from that format that hitherto was not thought possible or heard of. A ceiling apparently has not yet been reached, even though audible ceilings to the transparency of the CD format have been declared many times in the past.

Certainly, one might argue, as Amir does, that the bit depth of CD, its dynamic range, is not sufficient, which is a completely different issue than the 44.1 kHz sampling rate. Yet the question still remains open how much this matters in practice, given that, according to some engineers who have posted numbers here, the practical dynamic range of any playback system so far does not exceed a value somewhere around 80 dB. Certainly, manipulating the signal at 16 bit depth through all steps until the final master is problematic because you lose resolution at every step, but these days nobody does that anymore. The real question remains how much actual loss of resolution occurs for all practical purposes when limiting the bit depth down to 16 bit for the final CD after the signal has been processed throughout at higher bit depth up until that point.



If, for example an original 88/24 recording sounded better without bandlimiting than the same file bandlimited to 20 kHz (here you went down to 18 kHz), then there may be technical reasons for the degradation of sound during bandlimiting other than the frequency limitation. Looking to the more extended frequency response of hi-rez for an answer, when it comes to either the theoretical limits of CD playback or an explanation of inferior sound during current implementations of the CD format vs. hi-rez, is in my view questionable. There is just not sufficient, generally accepted, scientific evidence that our hearing, or any musical perception for that matter, extends beyond 20 kHz. Again, extraordinary claims require extraordinary evidence.

My comparisons used state of the art resampling software to convert between 88/24 or 176/24 down to 44/24 and then back up to 176/24. The only difference was the filtering, filtering that is mathematically required to eliminate known distortions (aliasing which is definitely audible and imaging that may be audible). This software allows control over the center frequency and slope of the filtering and the phase. In my experiments I varied all of these parameters and learned by careful listening what the tradeoffs were by comparing music in various formats that I had converted under controlled conditions. There is simply no way that 44/24 (let alone 44/16) can be audibly transparent on a good system while playing a good recording.

It is irrelevant that newer playback equipment can make older CD recordings sound "good" or "better". What would you expect? The old recordings are what they are. Nobody is suggesting throwing these recordings out. The argument is about new recordings.

As to "extraordinary claims", all I can say is that I don't reply to "extraordinary demands". My experience has been that people using this type of language will, at best, dismisses the evidence when it is presented, and at worst, accuse the presenter of lying and cheating. The founder of this forum has been accused of this very thing.

The problems with the CD format are the same as they always have been. There were some implementation limitations associated with early equipment, but these generated different distortions. American audio engineers such as Thomas Stockham knew perfectly well that the 44.1 kHz sampling rate was inadequate from the get go and they knew why. CD was engineered from the beginning to be a cheap consumer format, not a quality format for use by audiophiles.

 

[Al M.]

This last statement of Al's is just not accurate. There is plenty of evidence that people perceive frequencies above 20 kHz, just not always through the hair cells of the inner ear. How high bone conduction, skin vibration sensors and the like go is somewhat controversial, with some outlying scientists claiming perception to 100 kHz.

As pointed out, bone conduction is with sounds applied directly to the skull, which has nothing to do with normal hearing.

 

[Al M.]

As to "extraordinary claims", all I can say is that I don't reply to "extraordinary demands". My experience has been that people using this type of language will, at best, dismisses the evidence when it is presented, and at worst, accuse the presenter of lying and cheating. The founder of this forum has been accused of this very thing.

Don't worry, I wouldn't do that, just like I stated in my earlier post that I did not doubt your auditory observations regarding CD format vs. hi-rez.

Yet at the same time I do not think it is an extraordinary demand to ask for incontrovertible evidence about auditory perception above 20 kHz, when there is, to say the least, no scientific consensus that it does exist.

 

[Kees de Visser]

we have Don Hills who says DSD is broken, but then you have others who are world class engineers who manufacturer DSD products such as Bruno Putzeys with Grimm Audio and the study at Philips research labs looking at the requirement for transparent transcoding between PCM and DSD.

FYI Bruno is no longer with Grimm Audio and he is no longer with DSD. Luckily he is still into audio though

 

[Orb]

FYI Bruno is no longer with Grimm Audio and he is no longer with DSD. Luckily he is still into audio though

Yeah I knew they separated, but he was involved at the time it was relevant to this discussion, and yeah I agree great he is still into audio
That said the other engineers at Grimm are pretty exceptional themselves.
Cheers
Orb

 

[rbbert]

...
Yet at the same time I do not think it is an extraordinary demand to ask for incontrovertible evidence about auditory perception above 20 kHz, when there is, to say the least, no scientific consensus that it does exist.

"Incontrovertible evidence" is not a very useful standard even in engineering, much less perceptual science.

 

[BlueFox]

Yet at the same time I do not think it is an extraordinary demand to ask for incontrovertible evidence about auditory perception above 20 kHz, when there is, to say the least, no scientific consensus that it does exist.

Is there any data to even suggest it is possible, or suggest that if not heard then at least felt in some manner? It would seem that if we could somehow perceive these frequencies in some manner then if they are not reproduced then the material would appear lacking in some manner.

 

[rbbert]

Is there any data to even suggest it is possible, or suggest that if not heard then at least felt in some manner? It would seem that if we could somehow perceive these frequencies in some manner then if they are not reproduced then the material would appear lacking in some manner.

Even Wikipedia has a section on ultrasonic hearing, and there is a lot more research of variable quality (like all perceptual research) to be found elsewhere.

https://en.m.wikipedia.org/wiki/Ultrasonic_hearing

 

[Phelonious Ponk]

I won't claim to understand much of this conversation, so admittedly this is a dumb question from the peanut gallery:
Are there any engineers, world class or otherwise, from outside of the audiophile world, who are making product performance claims, or forwarding theories for such, based on human perception of ultrasonic sounds?

Tim

 

[Al M.]

Even Wikipedia has a section on ultrasonic hearing, and there is a lot more research of variable quality (like all perceptual research) to be found elsewhere.

https://en.m.wikipedia.org/wiki/Ultrasonic_hearing

The wiki page? Sigh...for the third or so time: it's through bone conduction, which has nothing to do with normal hearing. It's application of sounds directly to the skull.