Two unresolved issues

[microstrip]

This is hurting my head. So can someone tell me simply, could this eliminate the commonalities between digital and analog reproductions of the same recording, leaving us with the unmeasurable stuff that analog preserves and/or adds? And once you've done that, could you then look/listen and understand what that stuff is? IE: Is it signal digital loses or distortion analog adds?

Tim

Tim,
It will always hurt someone's head. Even if carrying an ideal experiment you can get the residuals, you have to analyze them. Then , you have two possibilities - declare them as inaudible - end of story, or take an aspirin and do the long and tiresome job of trying to establish the needed correlations with the data itself and the listening perceptions.

 

[Phelonious Ponk]

What of the sound of what's left? What if you remove everything the two media, files, or components have in common and what's left is hash and grunge? Can we conclude, then, that it was not air, inner detail, reverb tails, deeper sound stage that was left behind, but noise and distortion? Or will we need to analyze that grunge using the same tools that have been consistently rejected when measuring the whole signal? And then, would the believers not just conclude that there is something musical in the hash and grunge that the instruments were inadequate to measure?

Tim

PS: Sorry about the digital/analog reference above. I'm trying to pretend that's not what we're talking about, but I keep forgetting.

 

[JackD201]

In all honesty Tim, I personally think the differences are a combination of both. I'm just in no position to prove it. I also believe that somewhere down the road it really will be figured out beyond any reasonable doubts. The beauty of discussions such as these is that it means there is interest in the subject and for as long as there is interest, there will be some intrepid soul who'll actually go through the rigors required.

 

[Phelonious Ponk]

In all honesty Tim, I personally think the differences are a combination of both. I'm just in no position to prove it. I also believe that somewhere down the road it really will be figured out beyond any reasonable doubts. The beauty of discussions such as these is that it means there is interest in the subject and for as long as there is interest, there will be some intrepid soul who'll actually go through the rigors required.

Yep. Curiosity is a good thing. I still wonder, however, if we're making this far too complicated. Remove everything they have in common. We'll assume that what's left is going to be pretty quiet, but it must be audible, must be above the noise floor, or it is irrelevant, so crank it up to 0 db and listen. Now, if it is inner detail, air (an odd word for treble extension), reverb tails etc, will it not sound something like music? Faint, perhaps. Incomplete, of course. But it shouldn't simply sound like noise, and if it does can we not logically conclude that it is noise? Let's trust our ears dudes!

Tim

PS: By the way, I acknowledge the possibility that I am completely missing something important here and awaiting the feedback of some resident experts.

 

[DonH50]

Actually, scopes are not usually good for fine differencig operations as they are simply not that linear, and of course most of the scopes available now are digital so you still have the A/D issue to contend with. It can be fun to build difference circuits but is a royal pain to do well, even at audio, if you want to look 100 dB down. (Aside: I sincerely doubt anything down that far matters, but OTOH somebody will claim to hear below the floor no matter where we set it.) Doing it in real time, even more painful, but not impractical.

I was pondering ways I did things like this in the past with some half-baked thought of building up something, but have it on good authority that Satan, and perhaps fat Elvis, lives in the null and I couldn't bear to hear that...

 

[Lee]

I would like to add some observations.

By way of background (since I am new to this forum) I do 24/176 recordings of classical music in Atlanta with a new label launching soon, possibly with an LP. I'm a management consultant and banker by background but I was involved from 1990 to 1996 with Chesky records during the Bob Katz era as a recording assistant and assistant producer. We were working with both analog tape and digital hirez recording during part of that time. The analog tapes were later used for some of the Chesky SACDs but now Chesky records in 24/192. We have been working primarily on live to 2 track recordings using Sound Devices 722 boxes and AKG microphones at live, small ensemble events with some of the better musicians from the Southeast including many ASO members.

So my point is that I have tried a variety of things the past twenty plus years and have some hopefully useful observations:

1. Hirez digital is in fact quite audible. We make both 16/44 and 24/176 recordings at our sessions. Only the 24/176 files capture the essence of the live events. We also did 24/88.2 and 24/96 recording earlier on but found the same conclusion.

2. 24/176 with the right ADC and DAC is noticeably better than 24/96.

3. I'm a huge fan of DSD and have recorded it as well for both events and high quality needledrops. I have a slight preference for DSD over even hirez PCM because it seems more natural to me. SACDs made from tapes I am familiar with seem to capture extremely well what is on the tape.

4. I do analytics for a living and by academic standards the Meyer-Moran study was a sloppy mess not fit for print. As Bruce described, too many variables unaccounted for. Poor source material, poor resolution playback.

5. In my experience, DBTs can be challenging to conduct with precision. Often one gets more of an analysis of the listening skills of the audience than a good unbiased survey on sonic differences of the gear or music.

6. Format arguments seem to take on a religious nature on music boards and often no one's mind has changed. WBF has impressed with its civility and the open-mindedness of many members and I think the mods are doing a good job here and I think the level of discussion is quite high.

7. At this point in playback, in my opinion R2R, CD, SACD, LP, and DVD-Audio/Hirez Files all can sound spectacular. You need great playback gear, a great room, a great mastering, and a great recording (mic placement, great ADC, best practices) for the best possible results and yes I do prefer hirez playback in either LP, SACD, or DVD-Audio form to CD but CD can be pretty great also.

 

[microstrip]

I would like to add some observations. (...)

Lee,

Thanks for your testimony. Could you tell us which ADC-DAC units you have used?

 

[Phelonious Ponk]

1. Hirez digital is in fact quite audible. We make both 16/44 and 24/176 recordings at our sessions. Only the 24/176 files capture the essence of the live events. We also did 24/88.2 and 24/96 recording earlier on but found the same conclusion.

I don't think anyone is questioning the audibility of recording, and the subsequent processing in hi res. The only question, from my perspective anyway, is whether or not the final distribution media cd vs sacd is audibly different.

4. I do analytics for a living and by academic standards the Meyer-Moran study was a sloppy mess not fit for print. As Bruce described, too many variables unaccounted for. Poor source material, poor resolution playback.

What Bruce said about the source material is that some of it was bogus hi-res. We still don't know how much of it and what part it played in the study, so it's pretty meaningless at this point. Also, correct me if I'm wrong, but I thought I read that the players used measure just fine, so if that is correct, once again we are deep in familiar territory in which those who don't like the results of a study are blaming the gear used without substantive evidence to support their objections. I'm personally not going to dismiss a year-long study on that basis.

5. In my experience, DBTs can be challenging to conduct with precision.

No doubt.

Often one gets more of an analysis of the listening skills of the audience than a good unbiased survey on sonic differences of the gear or music.

If you're running DBTs to understand the listening skills of the participants you shouldn't be at all surprised if you come away without a thorough understanding of the sonic differences of the gear. In fact, the gear shouldn't change at all. This statement just indicates bad methodology or a bad understanding of the objectives.

6. Format arguments seem to take on a religious nature on music boards and often no one's mind has changed. WBF has impressed with its civility and the open-mindedness of many members and I think the mods are doing a good job here and I think the level of discussion is quite high.

7. At this point in playback, in my opinion R2R, CD, SACD, LP, and DVD-Audio/Hirez Files all can sound spectacular. You need great playback gear, a great room, a great mastering, and a great recording (mic placement, great ADC, best practices) for the best possible results and yes I do prefer hirez playback in either LP, SACD, or DVD-Audio form to CD but CD can be pretty great also.

Agreed on most of the above, but every time I've heard a difference between the same material in hi-res and redbook, it ended up being the mastering, not the format. When I down converted the hi res to 16/44.1, the difference disappeared. Could be my expectation bias. We'll have to agree to disagree on the idea that LPs are high resolution. If nothing else, the noise floor prevents that from becoming a practical reality.

Tim

 

[amirm]

I don't think anyone is questioning the audibility of recording, and the subsequent processing in hi res. The only question, from my perspective anyway, is whether or not the final distribution media cd vs sacd is audibly different.

I set out to discover this for myself when SACD and DVD-A came out. I bought 2-3 players of each format and got my hands on every title that was available in multiple formats. I then tested the following:

1. DVD-A/SACD players playing CD and comparing them to SACD/CD of the same.

2. Comparing the above but this time, routing the output of the CD through my Mark Levinson DAC as to optimize the performance of 16-bit/44.1.

3. Comparison of DVD-A players with their own DAC against SACD using its own DAC.

4. Comparison of DVD-A players feeding Mark Levinson with DVD-A content against SACD with its internal DAC.

The players I used at the time cost around $1,000 to $1,500 and included the Toshiba, Panasonic and Sony. I also tested the Marantz but did not like its sound.

Testing harness was my Stax headphone which has a pre-amp with two inputs. I would sync up the players on revealing tracks and then switch back and forth quickly. I did not attempt to level match with external boxes as I believed such boxes would invalidate some of the difference.

Here is what I found:

1. The Mark Levinson improved the fidelity of all sources playing CDs. The high frequency would be slightly smoother and I could hear more ambiance in notes decaying. And the decay was smoother going nicely into noise as opposed to stopping faster with the internal DAC.

2. The fidelity improved when I would turn of video and front panel displays in transports (with or without external DAC). Later I read reviews of these players that showed their DAC linearity improved by more than 2 bits with these tweaks!

3. CD against DVD-A and SACD would lose. It would suffer the reverse of what I mentioned in #1. Adding the Mark Levinson DAC would narrow the gap substantially but it would still lose to the stand-alone players using their internal DAC by a small factor.

4. The DVD-A sources would all lose to SACD. Addition of Mark Levinson sharply closed the gap but a tiny advantage remained in favor of SACD.

I later went to an AES conference where David Chesky to my amazement said the titles that I tested where SACD sounded better than DVD-A were actually authored in the native format of DVD-A! (PCM at 24-bit/96 Khz). This convinced me that some of the preference for SACD comes due to some manipulation of signal we don't yet fully understand.

This last week, I took out the old titles and tested them on our reference music system at work (Berkeley DAC driving Mark Levinson amp which powers Revel Salon 2). I was surprised to hear the same improvement in 24-bit/96 KHz tracks against their CD equiv. that I had heard some 10 years back. The sound is smoother, with better decay into noise. Here I thought I had gone deaf in my older age . This was a sighted test though.

I think it is pretty easy for people to try the above. All it costs is buying the DVD-A/SACD version of a high fidelity title you like with nice transients into silence. For $20, you could form your own opinion .

 

[RBFC]

every time I've heard a difference between the same material in hi-res and redbook, it ended up being the mastering, not the format. When I down converted the hi res to 16/44.1, the difference disappeared.

Tim

Tim,

This statement made me pause. You took recordings from the same session, one high-res and one redbook, and heard a difference. Then, if you down-sampled the high-res version, the difference disappears. Does this not imply to you that there is an audible difference between redbook and high-res? If down-sampling erases the benefit of better mastering, would you not consider that process degradatory?

Lee

 

[Phelonious Ponk]

Tim,

This statement made me pause. You took recordings from the same session, one high-res and one redbook, and heard a difference. Then, if you down-sampled the high-res version, the difference disappears. Does this not imply to you that there is an audible difference between redbook and high-res? If down-sampling erases the benefit of better mastering, would you not consider that process degradatory?

Lee

That should have given you pause, Lee. It was very sloppy English. What I meant to say is that whenever I've heard a difference between SACD and redbook I've concluded it is the mastering not the format, because I when down convert hi-res to redbook it seems to lose nothing. And again, this could be my expectation bias. The differences other people hear are, by all reasonable reports, pretty subtle and no, I didn't expect there to be a difference. I'm not immune to expectation bias. I am pretty immune to hi-res at this point, though. There isn't nearly enough of it in my collection to worry over it, so I just let iTunes convert it to 16/44.1. Good recording and mastering is a thing of wonder. Crap is crap. Format resolution will not turn one into the other.

Tim

 

[fas42]

I later went to an AES conference where David Chesky to my amazement said the titles that I tested where SACD sounded better than DVD-A were actually authored in the native format of DVD-A! (PCM at 24-bit/96 Khz). This convinced me that some of the preference for SACD comes due to some manipulation of signal we don't yet fully understand.

A key observation. All my experience with fiddling with gear over the years has said to me that there are no problems with the recording and mastering of the material: it's all to do with the playback. The "worst" recording using the worst recording technology played back on correctly sorted out reproduction gear will always sound superior, from the point of view of being musical and natural, then the "best" recording, using the theoretically optimum techniques, heard on an average or normal sound system.

For me, it has been chalk and cheese for 25 years. The engineers doing the recording HAVE been fussy enough, it's on the other side, where the consumer is, that the problems lie ...

Frank

 

[Phelonious Ponk]

A key observation. All my experience with fiddling with gear over the years has said to me that there are no problems with the recording and mastering of the material: it's all to do with the playback. The "worst" recording using the worst recording technology played back on correctly sorted out reproduction gear will always sound superior, from the point of view of being musical and natural, then the "best" recording, using the theoretically optimum techniques, heard on an average or normal sound system.

For me, it has been chalk and cheese for 25 years. The engineers doing the recording HAVE been fussy enough, it's on the other side, where the consumer is, that the problems lie ...

Frank

I sure get that this is your observation, Frank, what I don't get is how you can possibly have come to these conclusions.

Tim

 

[Ethan Winer]

It then follows that we can identify and measure everything we hear.

Let me clarify before we go off onto a tangent. I try to remember to qualify certain statements every single time, but I forget that not everyone reads all my posts. When I say "measure everything we can hear" I refer only to those things that gear does to degrade audio fidelity. I am specifically not talking about psychoacoustics, such as how we perceive the "imaging" that's embedded in a recording.

IMO, if you believe we can ascribe a measurement to everything we hear, then we should be able to assemble a satisfying musical system based on measurements alone.

Yes, exactly, but with one caveat: This assumes that the goal is high fidelity, where the playback chain alters the sound of the recording as little as possible. That's certainly my goal, but it's not the goal of everyone. Some people prefer slight added distortion, or specific non-flat frequency response curves. Buying on specs may not work for those people.

--Ethan

 

[Bruce B]

That should have given you pause, Lee. It was very sloppy English. What I meant to say is that whenever I've heard a difference between SACD and redbook I've concluded it is the mastering not the format, because I when down convert hi-res to redbook it seems to lose nothing.Tim

How did you down convert the DSD layer in SACD? How did you get the digital DSD information off the SACD?

 

[RBFC]

How did you down convert the DSD layer in SACD? How did you get the digital DSD information off the SACD?

Without specialized equipment, I'd guess that the work was done to the redbook layer of hybrid SACDs.

Lee

 

[microstrip]

What of the sound of what's left? What if you remove everything the two media, files, or components have in common and what's left is hash and grunge? Can we conclude, then, that it was not air, inner detail, reverb tails, deeper sound stage that was left behind, but noise and distortion? Or will we need to analyze that grunge using the same tools that have been consistently rejected when measuring the whole signal? And then, would the believers not just conclude that there is something musical in the hash and grunge that the instruments were inadequate to measure?
Tim

You are making my point - that the null test is useless for this comparison purposes. You will always get a difference signal, but what can you do with it?

The null test is useful you when you are adjusting something - you adjust to get a minimum in one significant measurement or if you know how to valuate the residuals. The big advantage is that you can use a higher sensitivity instrument as the range is much smaller.

Please see post #130 by ScottB on this thread.

 

[Ethan Winer]

What do you mean by "analyzed?" A sum difference of two signals gives you *data*. It does not give you analysis.

Yes and No. A null test makes it easy to analyze the difference both by ear and / or FFT, and by residual level. Even if the difference residual is "only" 100 dB down, that's the same audibly as a total null. And you can also analyze the residual. So even if the residual is only 60 dB down, if all the frequency content is below 6 Hz it's an interesting anomaly worth knowing about, but still proves there's no audible difference.

--Ethan

 

[Ethan Winer]

You refer very often to measurements and measured data to support your arguments, Ethan.

So I'll ask you the same thing - what specific physical property of sound does digital audio not capture? And why do you think a null test is not proof?

would you mind listing the test equipment you use, please.

I use SONAR to create null tests, and Sound Forge (and sometimes the Rightmark analyzer) to view FFTs.

it is indisputable that there are measurable properties that affect the quality of the perceived imaging.

Sure. If one channel's frequency response is different than the other by an audible amount, that will affect imaging. Same for the other parameters such as noise and distortion. Adding stereo noise to a mono recording will also affect the perception of width. I'm convinced that's why some people believe that analog tape and LPs sound "wider" than digital audio. The hiss (or scratches and pops) differ left and right, giving the perception of width via the addition of a stereo artifact added to the music.

--Ethan

 

[Ethan Winer]

Guys, I just started a 5-hour video render, and it's slowed my computer enough to make posting too tedious to continue. This doesn't usually happen when rendering, but this one is 1920x1080 at Blu-ray bit-rates. More tomorrow. I promise!

--Ethan