Apology for claiming not mentioning inaudibility
Rereading my posts I noticed I had actually written something indicating that it is audiable. Sorry for that, I should have been more correct and careful. I did write that it would be audable with that sentence saying it would be unnatural to listener.
I don't have proof for this, but it bugs me.
Like I wrote above, my recent acquisition of audio equipment sounds "stranger" than the previous ones I used to own. My "gut" feeling is more and more digital processing is going on, with the mind set that it is harmless.
Early on not knowing the details of theorems like sampling and other digital processing, I was also a believer that since theorem says so, CD and the whole A/D/A chain is wonderful.
Lately I happened have chance to look into more what is involved, and what are the details that are usually not talked about, and I start to get the impression that because digital is so easy now, and easy to do different things that wasn't possible before, it is being done without much care.
This is my personal feeling....
For instance now higher sampling rates are easier to achieve and the required bandwith, processing power and storage, than the initial times of the CD. And yet you are writing that, the modern DAC's use brickwall filters instead of using an analog filter that is casual that won't introduce anything like preringing and won't cause errors in the human audible band. Why, why not use high sampling with analog filters, which at least in theory for the humand audable band are at least as good as the brickwall ones and better in terms of not having these unnatural preringing?
Rereading my posts I noticed I had actually written something indicating that it is audiable. Sorry for that, I should have been more correct and careful. I did write that it would be audable with that sentence saying it would be unnatural to listener.
I don't have proof for this, but it bugs me.
Like I wrote above, my recent acquisition of audio equipment sounds "stranger" than the previous ones I used to own. My "gut" feeling is more and more digital processing is going on, with the mind set that it is harmless.
Early on not knowing the details of theorems like sampling and other digital processing, I was also a believer that since theorem says so, CD and the whole A/D/A chain is wonderful.
Lately I happened have chance to look into more what is involved, and what are the details that are usually not talked about, and I start to get the impression that because digital is so easy now, and easy to do different things that wasn't possible before, it is being done without much care.
This is my personal feeling....
For instance now higher sampling rates are easier to achieve and the required bandwith, processing power and storage, than the initial times of the CD. And yet you are writing that, the modern DAC's use brickwall filters instead of using an analog filter that is casual that won't introduce anything like preringing and won't cause errors in the human audible band. Why, why not use high sampling with analog filters, which at least in theory for the humand audable band are at least as good as the brickwall ones and better in terms of not having these unnatural preringing?
Sinc() is a brickwall, but not all brickwalls are Sinc(). And then we are not even discussing the possible variations in the transition band width.
You seem to imply that when using an analogue AA filter significant aliasing will occur unless the sampling rate is increased significantly above the original 2xFs/2. How then was this done in the 70s and 80s, with 44.1kHz?
And in real life this acausality is not a real problem. Really.
What is publicized and quoted over and over again that because of Nyquist (and Shannon and Whittikar and who else??) you record and reproduce exactly without any problems, which is not that simple.
You forgot Kotelnikov and possibly one or two others, and hundreds of mathemagicians who were intuitively aware of all this, but didn't feel it was worth their time. But let's give prime credit to Shannon, OK. After all he was the bloke who sat down to construct that nice proof (possibly while taking a break from inventing binary logic as we know it).
At the risk of being pedantic, it can occur in nature, although it is rare, tricky, and not very relevant.
The recording-side AA filter is not completely unknown. Prior to 1990 or so most recording and/or mastering was done with the Sony PCM16*0 series, Mitsubishis and 3Ms. Especially the PCM16*0 quickly became a golden standard for the delivery of stereo masters to the glass mastering facilities (and also for the distribution of LP masters to cutting houses worldwide, but that is another story).
The properties of the (analogue) AA filters in these recorders are well-known (if not widely available on the internet). These were generally 9th-11th order elliptic filters, with quite a large suppression at above 22kHz.
You may be interested in this, half-way down the page:
The why and the how of supertweeters
Then, from 1990 on, ADC (and DAC!) architectures quickly moved towards delta-sigma: high-rate oversampled low-bit modulators with massive noise shaping and digital-domain FIR AA filtering/decimation. This allowed chip manufacturers to move their ADC/DAC products to commodity (cheap) CMOS processes, instead of the more complex, more expensive bipolar or BiCMOS processes. And virtually all on-chip ADC AA filters were of the half-band type, i.e. with -6dB point at 22.05kHz, pre/post-ringing, and quite a bit of aliasing in the transition band. All knowns.
Even sample rate conversion software (or boxes) used in those productions that started from higher sampling rates generally followed this architecture (or did something much worse, e.g. ProTools' criminally-bad SRC). Only in the past 3-4 years or so did alternative filters emerge (look e.g. at iZotope SRC).
So it wasn't too far-fetched from Meridian to posit a solution in the shape of a digital minimum-phase filter cutting in at 19-20kHz or so, thereby cutting out any 22kHz ringing imprinted by recording-side half-band AA filters.
Of course, when cascaded with a recording done on a pre-1990 ADC the phase shifts of both processes add :-( Then again, tests in the 80s concluded that such was inaudible anyway. Not entirely sure if we should believe this, but it remains that many people think our auditory system far more sensitive to phase distortion than it really is.
Maximum phase is what you get when time-reversing a minimum phase response: it has all of the ringing before the event. Max Phase is the mother of acausality. But K.Howard's test subjects couldn't reliably identify it...
Again, we do know that pre-ringing below 16kHz is audible and generally a bad thing. But above 16kHz, especially for people whose age-related wear to the auditory system limits their frequency response, there is so far no evidence. You could try it yourself: get good hi-res recordings from isolated impulsive sounds (drums, cymbals, ...) and convert them to 44.1kHz with linear phase, minimum phase, and maximum phase AA filters. Then listen.
It would be interesting if someone did such tests on properly-trained teenagers (at least those who haven't shot their ears with their portable MP3 players).
Yes, that field of auditory science also remains a bit inconclusive.
Absolutely. That is exactly why so many tests and studies are worthless.
For instance http://www.google.be/url?sa=t&sourc...sg=AFQjCNF3cdWvvNSyV7aE8l_0dbBjH_40uQ&cad=rja reports audible differences between 44.1kHz and 88.2kHz sampling rates. But the study invoked the Pyramix 6 sample rate convertor, which according to SRC Comparisons has significant aliasing that may audibly corrupt its output.
To bring this back on topic: I own many LPs, three turntables of which two top-class and even including DIY parts, and enough phonostages to build a garden wall with.
A/D to D/A chain has still has its thorns, even though it is a rose
...
And unfortunately most people don't know or see or talk about the real issue(s) but talk about wrong conclusions ...
I was also a believer that since theorem says so, CD and the whole A/D/A chain is wonderful ...
get the impression that because digital is so easy now, and easy to do different things that wasn't possible before, it is being done without much care.
Yes. Digital appears to be easy and it most definitely is cheap. Compare the cost of a 24 track recorder in 1979 and today. Analogue-done-right inevitably requires good engineering, so expensive boxes, and operator skills, coming from training and experience. Back then almost no-one could master and cut an LP an get decent results. Incompetence was punished. Today literally any moron can record and master a CD, and what is worse, doesn't even require an investment to do so.
But the truth of the matter is that digital, too, requires a certain amount of care and of knowledge. Only a different sort of care compared to analogue.
And if you want another kind of care
I'll wear a mask for you
If you mean moving to higher sampling rates to avoid compromises then I agree, but only in part. The commercial reality, today, is that there is a huge base of 44.1kHz material, and just a tiny bit of hi-res material. And the latter more often than not is:
-differing from the low-res issue by virtue of different mastering (of course it sounds different then), and/or
-a scam in the shape of upsampled low-res material. Sometimes upsampled in the most inept way (Sinatra At The Sands 176.4kHz my *ss!)
So given this reality it seems, to me, more worthwhile to study how this low-res base can be optimised.
Actually there were pamphlets in the 90s, at the dawn of DVD-A, calling for exactly that. I wrote one myself ;-)
As an aside, Audio Note even developed a non-oversampling ADC with rudimentary analogue AA filtering. Peter Qvortrup sent me a CD with tracks recorded from LP, with his ADC and then with (IIRC) a dCS. But the track labelling was wrong, there were no pairs of matching tracks, and I couldn't discern reliably any differences between the stuff on the CD. Could well have been that all of the tracks on that CD originated from the same ADC, or that different music was recorded with different ADCs. The thing never matured or got into production ...
In memory of the LP part (and some CD reissues)
MERCURY RECORDS Living Presence - Wilma Cozart Fine and 50 Years Mercury Recordings
MERCURY RECORDS Living Presence - Wilma Cozart Fine and 50 Years Mercury Recordings
I'm hesitating to bat with such heavy hitters here, but I have understood this 'a-causality' as follows.
If you look at the output signal, you will see that there is a delay between input and output.
This delay is such that when the pre-ringing on the output starts, the input signal is actually there. It's just that the output is delayed.
So although from the output signal perspective it looks like (a-causal) pre-ringing, it is not a-causal wrt the input signal.
Have I got it right?
jan didden
If you look at the output signal, you will see that there is a delay between input and output.
This delay is such that when the pre-ringing on the output starts, the input signal is actually there. It's just that the output is delayed.
So although from the output signal perspective it looks like (a-causal) pre-ringing, it is not a-causal wrt the input signal.
Have I got it right?
jan didden
Yes. If you are prepared to pay for the necessary storage the digital domain allows one to fool around with time, so that what seems to be acausal(*) becomes possible and even trivial.
This is somewhat more cumbersome with linear analogue circuits.
None of this is important, though.
(* It is of course only acausal as long as you refuse to acknowledge the presence of a delay line in the chain.)
This is somewhat more cumbersome with linear analogue circuits.
None of this is important, though.
(* It is of course only acausal as long as you refuse to acknowledge the presence of a delay line in the chain.)
According to the very interesting discussion above it would appear that CD is demonstrably flawed in the area of ringing & hence time smearing. So, as was stated earlier, if the purpose of the signal chain is to accurately reproduce what comes from the microphone (or master tape), CD is not without it's problems.
That's not to say that LP isn't also flawed but in different ways. Is it now a matter of investigating which flaws are more benign, sonically?
That's not to say that LP isn't also flawed but in different ways. Is it now a matter of investigating which flaws are more benign, sonically?
Well, many here seem to prefer LP playback so arguably, for them, it would appear that the flaws in LP playback is more sonically benign?
If what you are repeating is the old argument that they prefer distortion, what I am positing is a possibility that perhaps, just perhaps, CD flaws are not as benign as LP? Or another possibility is that LP's flaws may be more psycho-acoustically accurate? No pleading of any kind going on here, just possibilities.
Last edited:
Well we have touched on the idea that unintended crossfeed between L&R in the LP process might help to create a more realistic stereo image in the average listening environment (duplicating 'Blumlein shuffler' EQ etc.).
Has anyone ever analysed the nature of the crossfeed one gets at different frequencies using a typical vinyl setup?
If this is the key to the marvellousness of vinyl, it should be possible to capture those 'flaws' in a CD recording of the vinyl playback, or to duplicate them 'on demand' in hardware or software without the need for Heath Robinson contraptions.
Hmmm, you seem to be trapped in the notion that CD is flawless & that "it should be possible to capture those 'flaws' in a CD recording of the vinyl playback" but what if CD itself is flawed - we have seen some of these mentioned already?
simply asserting CD is audibly flawed is begging the question too
I was pointing out that it is still debatable that CD audio limitations are at all audible - kind of a logical precondition to ranking preference
to posit that CD errors are "worse" than clearly audible vinyl playback errors when it is not firmly established that anyone can DBT a CD resolution digital link simply is not logical - it is special pleading
I think part of the perception "digital is bad" problem is the fact that most vinyl distortion theory and measurements are offline in printed journals and requires it requires considerable math and mixed domain modeling skill to follow the electromechanical reproduction chain and translate to signal properties
while digital audio technology is well described in fairly simple, widely taught at undergrad level math - and thoroughly examined to orders of magnitude greater detail than is even p0ossible with phonograph recording, manufacturing, playback
the insertion of a good quality digital ADC/DAC link in series with a "great" vinyl setup’s RIAA preamp output and doing the bypass test really seems the best test
I wouldn't totally be surprised if a 3-4 sigma tail of trained listeners could distinguish 16/44 but would be surprised that more than a few % of listeners could
at higher rates my knowledge of the multiple BW restrictions all along the record/reproduce chain make it really hard to believe that higher sample rate digital audio could be detected compared to vinyl
I was pointing out that it is still debatable that CD audio limitations are at all audible - kind of a logical precondition to ranking preference
to posit that CD errors are "worse" than clearly audible vinyl playback errors when it is not firmly established that anyone can DBT a CD resolution digital link simply is not logical - it is special pleading
I think part of the perception "digital is bad" problem is the fact that most vinyl distortion theory and measurements are offline in printed journals and requires it requires considerable math and mixed domain modeling skill to follow the electromechanical reproduction chain and translate to signal properties
while digital audio technology is well described in fairly simple, widely taught at undergrad level math - and thoroughly examined to orders of magnitude greater detail than is even p0ossible with phonograph recording, manufacturing, playback
the insertion of a good quality digital ADC/DAC link in series with a "great" vinyl setup’s RIAA preamp output and doing the bypass test really seems the best test
I wouldn't totally be surprised if a 3-4 sigma tail of trained listeners could distinguish 16/44 but would be surprised that more than a few % of listeners could
at higher rates my knowledge of the multiple BW restrictions all along the record/reproduce chain make it really hard to believe that higher sample rate digital audio could be detected compared to vinyl
Last edited:
Yes, debatable, so your plea to logical precondition makes no sense!
Again your use of "is not firmly established" leaves open all possibilities - so no special pleading
So what you're saying is that digital is at a disadvantage because it is easier to analyse mathematically & shows some flaws which can't be as easily done with vinyl? Is vinyl therefore more flawed or are you using it as special pleading? You are using mathematics as the criteria for defining which is pyscho-acoustically more accurate? I would love to see your mathematical model for the era & brain
Can't really make out what you are saying here but I sense high levels of expectation bias so maybe you wouldn't be the best person to perform the test
?
Last edited:
We have found that the CD format can, in fact, have manufacturing limitations which seem to effect the sound. This was also mentioned earlier.
However, this won't prevent those who come in and read only the last 3 or 4 posts of the entire thread and then start going over the same ground all over again.
Best Regards,
TerryO
I didn't see where the proposed vinyl source, digital link bypass test has been done?
a few peer reviewd journal articles that come to similar conclusions would be really nice
I certainly don't report my listening impressions as "evidence" and I have mentioned positive and negaitve controls for good subjective test design before
a few peer reviewd journal articles that come to similar conclusions would be really nice
I certainly don't report my listening impressions as "evidence" and I have mentioned positive and negaitve controls for good subjective test design before
Last edited:
I'm also reluctant to go over old ground again, but we did discuss how some vinyl pressings are actually from digital masters, or that the analogue master is apparently passed through a decidedly non-audiophile digital delay attached to the cutting equipment.
So earlier, did we have any suggestions as to how the glaring flaws of digital are purified and cleansed by the vinyl process? Is it a strictly one way process, so that you can't get the same effect by digitally recording analogue?
If the 'Blumlein Crossfeed' idea is correct, then it would allow for the process to work either way, and would explain the apparent mystery of how vinyl sanctifies a sinful digital source.
I was fascinated to find that Decca classical recordings from the 70s onwards were digital, often using their own homebrew systems and yet, as far as I know, no one has ever said that Decca's reputation of producing many of the finest classical recordings ever - on vinyl and CD - was sullied by a sudden fall-off of 'musicality', in the 70s onwards. It seems that Decca, with a priceless reputation to consider, decided that digital was 'perfect'. The poor deluded fools.
- Status
- Not open for further replies.