CD/DVD-A/SACD Dynamic Comparisons

[Christine Tham]

Response to comments

Hi - i am the author of the article(s) and Gene has asked me to respond to some comments made in this forum.

here are what i gathered were the main points:

av_phile: inherent limitations in cutting LPs?

response: absolutely correct, LPs have many limitations and cutting LPs without causing mistracking is an art rather than a science. agree that article should perhaps have highlighted these. however, to my knowledge, most material should be okay, examples like telarc's cannon shots are fairly rare. i must point out that some of these "limitations" are removed using half speed mastering and a good turntable.

av_phile: "I wonder if the CD player used has the same hi-end attributes as the hi-end turntable used"

response: actually, the CD player, Sony SCD-XA777ES is a high end deck that retails at over twice the combined price of the mid-end analog rig. however, do agree that turntables are a lot more difficult to setup to sound correctly, with cds it's plug and play.

WmAx: "question the value of these comparisons of different masters/mixes of titles against 'each other' on different formats"

response: perhaps article should have made more clear, but the specific titles were chosen because it was likely that the different formats were sourced from the same master, so the comment about these being "different" masters/mixes is incorrect. indeed, i have confirmation via private email from one of the engineers who mastered one of the titles i used that CD and LP were indeed sourced from same master tape. he added that he supports my findings - he's always noted these differences but i was the first one to publish them.

docferdie: "question the validity of using the audiotrak 7.1 as a measuring device to carry out the desired comparisons in this article. CD is PCM and SACD is DSD and yet they are both being evaluated by a card that can only work on PCM."

response: if you think about it, the technology used by the soundcard doesn't matter - it is used purely as a measuring device, all that matters is that the measurements are "accurate". in this respect, the soundcard is actually far more accurate than an average CRO or test bench measuring device. the fact that it captures in PCM is immaterial - most of the limitations of PCM are in the playback, not in the capture. of course, the soundcard will not reveal any benefits of DSD from the higher sampling rate - such as perhaps higher time accuracy, and the resolution is probably too low to do accurate measurement of impulse response. all measurement devices have limitations, including heisenberg's principle. for the purposes of spectral and dynamic comparisons, which are all these articles are about, a PCM soundcard is a very valid measurement device and will highlight differences - as the articles demonstrate.

if you are not convinced, think of it this way: suppose i said i didn't use a soundcard, i used the super duper XYZ Spectral Analyser and Dynamic Comparator. would that have made the measurements more valid? well, it may surprise you, but most high end test measuring equipment these days, particularly those that analyse the measurements digitally, use PCM ADCs to sample the signals. the only difference is the quality of the PCM ADCs. and in fact, the soundcard probably has a higher resolution/accuracy than many of the older devices out there. i know - i used to work for HP many years ago (before thet spinned off agilent and bought compaq) - all the high end test and measurement devices were digital and used PCM ADCs.

 

[WmAx]

response: perhaps article should have made more clear, but the specific titles were chosen because it was likely that the different formats were sourced from the same master, so the comment about these being "different" masters/mixes is incorrect. indeed, i have confirmation via private email from one of the engineers who mastered one of the titles i used that CD and LP were indeed sourced from same master tape. he added that he supports my findings - he's always noted these differences but i was the first one to publish them.

Problem is, you just can not trust informaitn such as this(same master, no modifications) without proof. BUt, it is not any suprise that vinyl would hae to sound different, considering it's many limitatinos. But take that vinyl and sample it to 44/16 with a good ADC. I bet it will be audibly identical to the origial record. Now, try the opposite, let's see who similar a vinyl made from the 44/16 sounds to the original......

On to the CD/DVDA/SACD comparison:Capturing the data with a soundcard you can perform many simulations on this to see what is a reality. For example, the Krall CD, DSD, DVD-A waveform samples showing different amounts of compression reveals this in particulate must be a different version of the master. Not a limitatino of any format, a purposefullly applied effect, since even CD has DR athat exceeds any practical use for playback purposes. If a recording engineer said this was not so, i would be forced to doubt their true knowledge of the relevant matters.

Here is a quick example of a DVD-A, captured at 96/24 with a soundcard that was measured before capture to confirm proper operation. The carefully FFT prefiltered and downconverted version has no added compression, nor does it need such. FYI, the statistical data performed in teh same fashion as you used(same window time, same square wave analysis) is nearly idnetical except for a 3-4dB difference in the noisefloor statistic(44/16-89, 24/96-93), as expected. Here is a graphic example of the waveform, no visible compression difference exists, unilke the extreme examples in the Krall CD:

http://www.linaeum.com/images/comparewaveforms.gif

I have performed several ABX tests as of this date, experimenting with different filters and conversions on hi-res audio,,etc. Reason being that it appears I am being forced to buy DVD-A or SACD versions simply to capture the superior version of the mix instead of putting up with the inferior, purposely degraded CD versions for use on my portable audio player. When I am careful to use a high quality FFT prefilter before resamplilng to 44/16, then converting back to 24/96 so that the soundcard does not have some odd result with different formats, I can not identify any differences by listening in DBT testing(ABX protocol). I used a Sony MDR-CD3000 reference headset for these comparisions to maximize sensitivity to differences. If their are any audible differences on special problem samples, they are so small that they are not important in any way, and can not be identified short of a tedious direct ABX session. I find that the significant audible differences(with no need for DBT, due to the terrible problems) can occur when carelessly trusting automated filters in some applications. Then, their is the issue of transducer non linearity. It is well known that when audible band and ultrasonic informaotin is played over a single transducer, the driver will create IM products in the audible band as a result of it's non linear behaviour. This is most likely when the amplitude of the ultrasonics approaches the same amplitude as the audible band components. However, most ultrasonic informaotin is tens of dBs below the main part of the audible band in recordings.

I can not conclude that RBCD format has any shortcomings for playback purposes as far as the sample rate/bit depth when combined with proper noise dithering and resampling methods. An attempted comparison of commercial variants/formats can only conclude the differences taht exist in the end, not accouting for the variables in between.

-Chris

 

[Christine Tham]

Reply to WmAx

On to the CD/DVDA/SACD comparison:Capturing the data with a soundcard you can perform many simulations on this to see what is a reality. For example, the Krall CD, DSD, DVD-A waveform samples showing different amounts of compression reveals this in particulate must be a different version of the master. Not a limitatino of any format, a purposefullly applied effect, since even CD has DR athat exceeds any practical use for playback purposes. If a recording engineer said this was not so, i would be forced to doubt their true knowledge of the relevant matters.

Forgive me for saying this, but this sounds quite badly flawed circular reasoning. You seem to be arguing: the different formats shouldn't exhibit these differences, so if any differences are observed, it must be due to a different master. And if the engineer who was actually responsible for the recording says otherwise, well, he's wrong.

I'm sorry, but i prefer to trust the word of someone who was involved in the recording than an unsubstantiated assertion of someone who was not.

Here is a quick example of a DVD-A, captured at 96/24 with a soundcard that was measured before capture to confirm proper operation. The carefully FFT prefiltered and downconverted version has no added compression, nor does it need such.

I'm not sure what point you are trying to make here. You seem to be saying if you take a recording at 96/24 (the source of the recording doesn't really seem to be matter in your example), downconvert it to 44.1/16, there are no change to the dynamics.

Well, d'oh, why would there be? in fact, digital sampling theory assures us there should be no change. But my articles are not about proving that the different formats somehow magically alter dynamics - they simply illustrate that differences exist even when they are allegedly taken from the same master(s). We can speculate about the cause of such differences, but they remain speculation. However, I don't think casting doubt on engineers who assert they come from the same master with comments like "doubt their true knowledge of the relevant matters" is necessarily helpful - that's like a scientist who discards any samples that don't fit the theory by saying "well, the samples must be wrong!"

if you really want to prove your theory that there are no inherent differences in dynamics between different formats or sampling rates, you need to simultaneously record the same analog source using multiple ADCs operating at different "native" sampling rates and compare the dynamics captured by each. that would be a very useful experiment, and i would love to see the results of such an experiment.

I can not conclude that RBCD format has any shortcomings for playback purposes as far as the sample rate/bit depth when combined with proper noise dithering and resampling methods. An attempted comparison of commercial variants/formats can only conclude the differences taht exist in the end, not accouting for the variables in between.

And that is all that these articles are attempting to say. Measurements were taken, differences were observed. The articles don't provide a definitive reason for the differences, nor do they make any unsubstantiated "conclusions" about possible "shortcomings" in any format that you seem to have such exception to.

 

[WmAx]

Forgive me for saying this, but this sounds quite badly flawed circular reasoning. You seem to be arguing: the different formats shouldn't exhibit these differences, so if any differences are observed, it must be due to a different master. And if the engineer who was actually responsible for the recording says otherwise, well, he's wrong.

I'm sorry, but i prefer to trust the word of someone who was involved in the recording than an unsubstantiated assertion of someone who was not.

My reasoning is based on lack of data showing any inherant superiority of SACD or DVD-A over properly produced RBCD for 2 channel playback. As I stated at the end of teh last reply, and as you have acknowledged, the 'why' of the differences you found are not known. But I only mean to comment on the actual limitations of the formats and changes due to this.

However, I don't think casting doubt on engineers who assert they come from the same master with comments like "doubt their true knowledge of the relevant matters" is necessarily helpful - that's like a scientist who discards any samples that don't fit the theory by saying "well, the samples must be wrong!"

I should have clarified. I mean this to apply broadly - such as if an engineer attempted to say that 44/16 must audibly compromise what a 96/24 can reproduce, etc. I did get a little carried away - plus I was tired when I responded - so I may have jumbled some of my statements. I just meant that I can not assume whatever someone says is true, regardless of their credentials, unless it's backed by hard data.

if you really want to prove your theory that there are no inherent differences in dynamics between different formats or sampling rates, you need to simultaneously record the same analog source using multiple ADCs operating at different "native" sampling rates and compare the dynamics captured by each. that would be a very useful experiment, and i would love to see the results of such an experiment.

Seems pointless. The RBCD format already allows a SNR that exceeds all practical playback use. For recording, their is no doubt that higher bitrate then 16 is desirable. That is not an issue. With higher bitrate, you have more toelrance for level set errors and peak dynamics that were not expected during the recording session(s).

And that is all that these articles are attempting to say. Measurements were taken, differences were observed. The articles don't provide a definitive reason for the differences, nor do they make any unsubstantiated "conclusions" about possible "shortcomings" in any format that you seem to have such exception to.

I understand. I apologize if I seem like I'm trying to provoke. I just did not see the point in analysing from this uncontrolled perspective. Perhaps part of my reason for replying was the titles of the articles, that at first glance seem like they may try to provide definitive information on the issue when in fact it was analysing the end result(s) of different formats of which the steps leading to the final results are unknown.

-Chris

 

[mtrycrafts]

From reading your three or 4 articles, I gather then that you measured the different formats and you measured and saw differences. There is no real conclusion why they are different. Am I correct in this?

 

[Christine Tham]

First of all, thanks for taking the effort to reply, and your clarification. It's good to have a level head when discussing these things, which often turn into an ugly technical debate!

My reasoning is based on lack of data showing any inherant superiority of SACD or DVD-A over properly produced RBCD for 2 channel playback.

I'm not sure I would agree with you on this, but I accept your right to hold that view. I would say the data is inconclusive.

From a theoretical perspective, it could be argued that there are no inherent advantage of higher sampling rates and depth over 44.1/16. Also, the advantage of DSD over PCM are somewhat open to question and debate.

However, from a practical perspective, there is simply too many anecdotal evidence of people who notice differences between formats for me to dismiss them. I myself "imagine" that I can detect some of these differences, which is my main motivation to do the measurements in the first place, to detect whether my ears are telling the truth or not. I was relieved to find out that whenever my ears tell me something is different, it is substantiated by actual measurable differences.

Seems pointless. The RBCD format already allows a SNR that exceeds all practical playback use.

Again, maybe pointless from a theoretical perspective, but not in practice. There are too many reputable engineers who I trust who assure me that they have actually done the experiment that I suggested and found differences.

I understand. I apologize if I seem like I'm trying to provoke. I just did not see the point in analysing from this uncontrolled perspective. Perhaps part of my reason for replying was the titles of the articles, that at first glance seem like they may try to provide definitive information on the issue when in fact it was analysing the end result(s) of different formats of which the steps leading to the final results are unknown.

I accept your apology, and no hard feelings on my side. You have to remember that I am not an audio engineer - my original intent in doing these measurements was to try and quantify and confirm differences I seem to be hearing.

I don't care about the reasons for the differences, as a consumer I am only interested whether there *are* differences or not, since that may influence my purchasing decisions in the future. Someone else can worry about whether these differences are inherent in the format, due to different masters being used, or other variables in the transfer process.

 

[WmAx]

However, from a practical perspective, there is simply too many anecdotal evidence of people who notice differences between formats for me to dismiss them.

THis is my quibble with the actual new formats for playback use. Only anecdotal evidence exists. Regardless of number of people that support the anecdotal evidence/accounts; this does not make it more true then one that has little to no supporters. But as you found in your measurements, differences are present that are not explained by the format specifications such as the clipped waveforms and different levels of compression applied across formats. That's why I require hard data to back up claims regardless of credentials/experience(and why I can not openly trust a recording engineer): just look at all of the very highly experienced people with excellent credentials actively promoting and claiming existance as fact of 'mysterious' cable sonics, psychic phenemona, etc. which have no substantiation.

If you wish, I can refer you to(or email to you) carefully produced JAES papers of perceptual research for bandwidth and SNR.

-Chris

 

[gene]

WmAX;

just look at all of the very highly experienced people with excellent credentials actively promoting and claiming existance as fact of 'mysterious' cable sonics, psychic phenemona, etc. which have no substantiation.

I agree with you 100% and feel it is truly sad. We will be posting an interview article on cables with a high end speaker manufacturer that will surprise everyone. Realize that many of the high end manufacturers try to support their dealers who usually promote very high $$$ cables with questionable marketing claims. Its hard to get everyone to see the light, or the cable for what it really is, wire

 

[mtrycrafts]

However, from a practical perspective, there is simply too many anecdotal evidence of people who notice differences between formats for me to dismiss them. I myself "imagine" that I can detect some of these differences, which is my main motivation to do the measurements in the first place, to detect whether my ears are telling the truth or not. I was relieved to find out that whenever my ears tell me something is different, it is substantiated by actual measurable differences.

But, if I read your papers right, you only measured the products after it was on their format, not throughout its chain of production what compression may have been applied to the CD, or how the other two were treated differently.

An article on SACD by Dr David Rich

http://www.stereophile.com//features/374/index.html

While it is not peer reviewed, Dr Stanley Lip****z has publisd a JASE paper on this if I remember correctly. Sony has stopped using 1 bit processors a long time ago in their authoring and mastering





I don't care about the reasons for the differences, as a consumer I am only interested whether there *are* differences or not, since that may influence my purchasing decisions in the future.

So, then the CD producers could factually cause their CD to sound awful to force you to choose the hi rez system that they carefullly engineered and you would not care?


Someone else can worry about whether these differences are inherent in the format, due to different masters being used, or other variables in the transfer process.

Yes, but then your work is for not and is half the picture.

 

[Rob Babcock]

Yes, but then your work is for not and is half the picture.

I think you mean, for "naught."

Anyway, I like the feature. I've not seen anything that really slam dunk saying RBCD is as transparent as DSD. As a matter of fact, I've seen no compro's at all. At the moment, any mention of it is met with innuendo. Even if Ms. Tham's work isn't definitive, it's a start. As there's virtually nothing else out there I find it interesting.

I still think it would be cool if the AH staff could make some of their own recordings at varying bit depths and utilize their considerable expertise in devising some tests. Unfortunately, that probably won't be possible with DSD. Last I checked there aren't many direct to DSD recorders, and they're very expensive.

 

[WmAx]

I think you mean, for "naught."

Anyway, I like the feature. I've not seen anything that really slam dunk saying RBCD is as transparent as DSD.

That's backwards logic. Your argument assumes RBCD is not transparent when this is unfounded.

I still think it would be cool if the AH staff could make some of their own recordings at varying bit depths and utilize their considerable expertise in devising some tests.

Unfortunately, that probably won't be possible with DSD. Last I checked there aren't many direct to DSD recorders, and they're very expensive.

Such a test has many complications! It's alot easier by starting with hi res then manipulating it into the various lower resolition versions. I did such a thing for some recent ABX tests I put together to find a transparent method of downconverting hi-res to 44/16 for use on a portable player, that I mentioned earlier. It's not at all difficult to convert the hi resolution informaiton to transparent(or extraordinaly difficult to discern) 44/16 format with the proper steps and software. If you want research into actual bit depth(SNR), refer to the reference I gave earlier in this thread. CD exceeds the required optimal SNR for playback purposes by a large margin.

Now, their may be some potential for 'different' sound from the hi res formats due to intermodular products created from the extended bandwidth in a single transducer covering band if in fact the audible band and ultrasonic amplitudes appraoched unity for a sutible duration. This would be a distortion, though, not accuracy.

-Chris

 

[Rob Babcock]

Yeah, certainly S/N-wise Redbook is sufficient. I'd really be interested to do a compro with Redbook vs DSD. On paper, there are things that look squirrelly with DSD, and I've convinced myself it doesn't sound as good as PCM (no, you don't have to point out that it's irrational & unscientific- I already know! ). Certainly it sounds very good, though.

I'll certainly agree with you when you say hard to discern. But in this instance, I'd welcome your customary blizzard of links about one subject: DSD. I've never seen one single bit of empiracle evidence to support your assertion that Redbook is as transparent as DSD. Nor have I seen anything to refute? Backwards logic? You misunderstand the situation- it's just a lack of any evidence at all. However, you are well versed about such things, and such evidence may very well exist. I merely have yet to see any.

I do have a few questions: is there any good documentation of Redbook vs, say, 2" analog tape? Are you assertion 16 bits is perfectly transparent to any source we could name? As you're aware, early CDs came with a notice printed in their jackets saying "Due to the high resolution of the CD medium, limitations of the analog source tape may be revealed." Newer remasters show this was absurd- many early CDs sounded awful compared to the original analog tape. I blame the master of the CD, not the Redbook spec itself, but I put it to you bluntly: is Redbook utterly transparent?

And is there any good scientific DBT comparing DSD to either a resampled PCM recording or an original analog tape?

I realize we might be straying a bit from the original topic, but hopefully the author will take it as an exploration, not a hijacking!

BTW, Chris, you quote me so much I'm starting to feel like a rock star! Thanks!

 

[WmAx]

Yeah, certainly S/N-wise Redbook is sufficient. I'd really be interested to do a compro with Redbook vs DSD.

It would be intersting to see a scientifcally valid test, but no one has motivation to do this. Sony knows the results of the peer reviewed perceptual studies on these matters.....

I'll certainly agree with you when you say hard to discern. But in this instance, I'd welcome your customary blizzard of links about one subject: DSD. I've never seen one single bit of empiracle evidence to support your assertion that Redbook is as transparent as DSD.

Kind of backwards, again. ALl of the paramters of RBCD have been proven transparent -- too meat the criteria for human detectability thresholds. How d o you get more transparent than transparent?

Unfortuntely, I don't have links. The papers are primarily AES papers that are not available except by way of aes.org. Or you can email me. The last reference is available online somewhere, you can search for it with google. I don't have the link at the moment. Or email for that one, too.

Which Bandwidth Is Necessary for Optimal SoundTransmission?
*. PLENGE, H. JAKUBOWSKI, AND P. SCHONE
JAES, 1980 March, Vol. 28, No. 3

Signal-To-Noise Ratio Requirement for Digital Transmission Systems
Spikofski, Gerhard
AES preprint 2196, Convention 77(1985)

Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components
Toshiyuki Nishiguchi, Kimio Hamasaki, Masakazu Iwaki, and Akio Ando
NHK Labs

You misunderstand the situation- it's just a lack of any evidence at all. However, you are well versed about such things, and such evidence may very well exist. I merely have yet to see any.

I see. I thought I had provided references for you in other threads.

I do have a few questions: is there any good documentation of Redbook vs, say, 2" analog tape?

Redbook has specificaitons that suggest transparency, period, for playback. You realize, that depending on the noise reduction used, multritrack or stereo, etc., that analogue tape can introduce audible distortions. So it would be ideal to A--D--A the analoge source to compare the original to the sampled version. Such a test was done with 1/2" 15 IPS a few years ago in a studio. The results of this test are found on the official ABX website. But this was not a scientifcally valid test.

Are you assertion 16 bits is perfectly transparent to any source we could name? As you're aware, early CDs came with a notice printed in their jackets saying "Due to the high resolution of the CD medium, limitations of the analog source tape may be revealed." Newer remasters show this was absurd- many early CDs sounded awful compared to the original analog tape. I blame the master of the CD, not the Redbook spec itself, but I put it to you bluntly: is Redbook utterly transparent?

Earliest CD technology had problems: anti-alias filters, quantization noise, etc.. I prefer to focus on current possibilities, not ones from the past that are no longer an issue.

And is there any good scientific DBT comparing DSD to either a resampled PCM recording or an original analog tape?

I know of exactly zero scientificly valid perceptual research projects analyisng DSD audibility.

-Chris

 

[Rob Babcock]

You bring up a point that I've been thinking about for a long while, but have never mentioned in print. I find a lot of the older studies not very compelling due to just that: older DACs really weren't nearly as good. That's one reason I really don't pay much attention to studies from the early days of digital. Such a high percentage of your links on digital are dated from 85 or before that I don't think they reflect the reality of current digital. So far I haven't seen a lot of compelling evidence that shows Redbook is absolutely transparent. What I've really love to see would be DBTs comparing a microphone feed to PCM playback. That would determine if any PCM recording can truly be transparent.

At any rate, tape certainly introduces distortions, but certainly the majority of our CDs (unless the purchaser is very young) began as analog tapes. And naturally, the A/D conversion process needed to turn music into PCM introduces it's own set of distortions. Therefore, the issue of CD fidelity becomes fidelity to the original analog master in those cases.

As for the DSD studies, I think perhaps Sony might not be completely impartial. And I feared you might just say that- that no one has yet done a good test of DSD. As its existence is based on the need to create a revenue stream more than to replace a flawed technology in PCM, they probably have no incentive to subject it to rigorous DBTs.

I feel you may be implying super powers to RBCD that can't possibly exist, without really meaning to. Do you truly mean Redbook is "perfect", ie that given perfection in the rest of the chain, it would be indistinguishable from reality? Or that 16/44.1 is so utterly faithful as to be totally indistinguishable from a live mic feed? That seems to strain credulity- please confirm this is what you meant or clarify this if I've put words in your mouth. (Your actual words-"ALl of the paramters of RBCD have been proven transparent -- too meat the criteria for human detectability thresholds. How d o you get more transparent than transparent?").

 

[WmAx]

You bring up a point that I've been thinking about for a long while, but have never mentioned in print. I find a lot of the older studies not very compelling due to just that: older DACs really weren't nearly as good. That's one reason I really don't pay much attention to studies from the early days of digital.

Perhaps you should actually read the studies. FOr example, the bandwidth study does not even use a DAC. It's examines the nescearry filter rates and relevant bandwidths among a group of audio experts as listening subjects, comparing a harmonic rich test tone to find the thresholds.

What I've really love to see would be DBTs comparing a microphone feed to PCM playback. That would determine if any PCM recording can truly be transparent.

As long as the A-D-A is measured/confirmed to be operating correctly and that the A-D is not clipped.

At any rate, tape certainly introduces distortions, but certainly the majority of our CDs (unless the purchaser is very young) began as analog tapes.

True. But my point meant to encapsulate the issue of the distortions on their own. That is, if you recorded the same source onto tape and onto PCM at the smae time then cmopared them, you would likely hear differences due to non linearity of the tape.

And naturally, the A/D conversion process needed to turn music into PCM introduces it's own set of distortions.

But the distortions are at extraordinary low levels, under human detectability in music playback.

Do you truly mean Redbook is "perfect", ie that given perfection in the rest of the chain, it would be indistinguishable from reality? Or that 16/44.1 is so utterly faithful as to be totally indistinguishable from a live mic feed?

That seems very dramatic! But as far in as audibility, evidence suggests that RBCD format has parameters that are optimal(transparent) for music playback under realistic conditinos/environments.

-Chris

 

[Rob Babcock]

Just two observations: I wasn't referring to the bandwidth studies. A cursory examination of my actual post would reveal that. Second- what does drama have to do with it? It's transparent to the mic feed or it's not: what could possibly be easier to test or more telling than that? Either that document exists or it doesn't. I'd imagine someone has tested that.

 

[WmAx]

Just two observations: I wasn't referring to the bandwidth studies.

The two main issues I refer to are the main two that are touted as being superior with the he-res formats compared to CD: bandwidth(sample rate) and SNR(bit depth).

It's transparent to the mic feed or it's not: what could possibly be easier to test or more telling than that? Either that document exists or it doesn't.

Perhaps it exists; but how is a live feed superior to a recorded one, as long as both contain the bandwidth and signal dynamics/noise levels that are under examination? I don't know of any live mic feed tests. But I'm sure you can find some anecdotal accounts, if that is of interest. I think either Monstrous Mike or Resident Loser used to talk about some tests on live feeds they used to cmopare with 44/16 converted simultaneously. I think they ahve joined this forum recently(?). Perhaps they will chime in. I can not do such a test, as I do not have a method to perform a high quality A-D-A at 44.1/16 in real-time.

-Chris

 

[Rob Babcock]

Thanks, Chris. Come to think of it, I do know a guy who does a lot of live 24 bit recording (he's actually recorded some very cool concerts for me- legit stuff, not pirated). I might see what the other guys think. There's reams of anecdotal evidence, but I don't put a lot more stock in that than you do. Okay, probably a bit more than you do, but not a huge amount.

If you're discussion fidelity, how could a test being one extra generation removed from the source be better than testing a live feed? I don't know in what way the recorded version would differ, but I'd be stunned it it didn't. To suggest otherwise would be to imply some method existed to make a flawless recording. Much is made of the quality of live music, and a mic feed is only one layer away from the actual event. Probably better than comparing a recording to live, because in the latter case you'd be comparing "live" to the entire chain (ie mic, mixer, cables, amp, pre, speakers, room, etc).

So as not to further confuse you, I simply don't care in this context about the two metrics you're talking about. I'm discussing only the veracity of the sound recording and how it would stand up under DBT. "Is it live, or is it Memorex" kind of stuff!

Hmmm...I've really steered this off course. My apologies. The information I'm looking for must exist somewhere. I'll have to see what I can dig up.

 

[WmAx]

Thanks, Chris. Come to think of it, I do know a guy who does a lot of live 24 bit recording (he's actually recorded some very cool concerts for me- legit stuff, not pirated).

Now remember, 24 bit is an advantage for recording purposes. It is good to have the extra bits available for unexpected transients. You can have your nominal signal quite a ways from the top of the range, to make sure you have no unexpected problems.

If you're discussion fidelity, how could a test being one extra generation removed from the source be better than testing a live feed?

For comparison purposes, their just is not anything unique about the live feed. No matter what the source, if it contains the desired SNR, distortion and spectral content, then the purpose is to identify 'difference' from the source whether it is live or recorded. But recorded material has a distinct advantage for comparing. In ABX tesing, short samples can be used so tht one can rapidly compare the same passage over and over to attempt to find a pattern problem and identify. If you DBT switch between a continous stream of audio from one to the other, you never actually get to compare the time same passage on both sources, therfor it's not as easy to pick out differences.

I don't know in what way the recorded version would differ, but I'd be stunned it it didn't. To suggest otherwise would be to imply some method existed to make a flawless recording.

'Flawlessness' or 'perfection' is not required to be transparent to human ears.

-Chris

 

[mtrycrafts]

I think you mean, for "naught."

Well, good thing we are not on a spelling board

Even if Ms. Tham's work isn't definitive, it's a start. As there's virtually nothing else out there I find it interesting.

In that respect, it is indeed a start. I really appreciate that. However, some may take that to mean that one format is indeed superior? Or, that is audibly better? Not knowing if it is the fault of the format or the recording/masterig transfers it can mislead those who are not critical thinkers?


I still think it would be cool if the AH staff could make some of their own recordings at varying bit depths and utilize their considerable expertise in devising some tests. Unfortunately, that probably won't be possible with DSD. b]

You should check out a CD called "The Digital Audio CD" by Markus Erne. Available at Amazon. I think that is where I bought mine

It has tracks to show influence of sampling rates and number of bits used. Different tracks, of course It has a few more interesting tidbits on it.
If your remote can jump from track to track without being audible inbetweem, and the player is not visible, somone can jump the tracks for you, randomly, to see if you can differentiate in the degraded track and at what level

Much more difficult than one would think and relatively easy to test.