K2hd? Xrcd? Hdcd? Mqa?

[miguelito]

All interesting techniques intended to maximize the audible quality in a 16/44 encoding package.

 

HDCD and MQA require decoding to realize the improvement, but clearly there's so much info that can be packed in 16/44, so there's still some form of "euphonic choice" of what to keep and what to ditch.

 

What's people's take on these?

[miguelito]

All interesting techniques intended to maximize the audible quality in a 16/44 encoding package.

 

HDCD and MQA require decoding to realize the improvement, but clearly there's so much info that can be packed in 16/44, so there's still some form of "euphonic choice" of what to keep and what to ditch.

 

What's people's take on these?

And I forgot SHM-CD!

[toddrhodes]

In my experience, at least over the last 2 days, if you want to maximize Redbook (in PCM), these are the acronyms to know:

 

R2R

JPlay US (Ultrastream)

 

And if you do happen to have XRCD Redbooks, all the better

[Hiro]

Now that we have access to PCM and DSD studio masters, the band-aid solutions such as smh-cd, hdcd, etc are no longer needed (unless, perhaps, someone wants to save space on their hard drive, but these days storage is dirt cheap).

[Hiro]

And if you do happen to have XRCD Redbooks, all the better

 

XRCDs are mastered in 24bit/96kHz, so the XRCD redbooks are merely downsampled/resampled versions of the masters, with decreased bit resolution to fit on CD.

[JW Audio]

From what I have read about MQA, if they ever get that going, it may very well be a game changer.

As far as the other choices, I've heard some great recordings in XRCD, but K2 would be my choice.

[CatManDo]

And I forgot SHM-CD!

 

I thought you meant CDs with a new mastering or encoding technology. SHM-CD is about manufacturing only, like other brands (Blu-spec CD).

 

What all those (except HDCD) have in common, is that the concept comes from japan, where CD buyers seem to be more receptive to "high tech CDs". In the US and Europe, studios also upgraded their mastering equipment over the time, and manufacturing probably also improved, but they didn't give their CDs new brand names for that reason. "24bit remastering" etc is about the only marketing hype they used, whereas the list of japanese CD "formats" is endless.

[Hiro]

HDCD and MQA require decoding to realize the improvement, but clearly there's so much info that can be packed in 16/44, so there's still some form of "euphonic choice" of what to keep and what to ditch.

 

Is MQA-encoded file after unpacking a bit-perfect copy of the original file (as is the case with FLAC)? From what I heard about the technology, that's not necessarily the case.

[miguelito]

Now that we have access to PCM and DSD studio masters, the band-aid solutions such as smh-cd, hdcd, etc are no longer needed (unless, perhaps, someone wants to save space on their hard drive, but these days storage is dirt cheap).

My question is more out of curiosity than usage.

 

Yes, ideally we have higher resolution format music releases, but the reality is the vast majority of music is not available in true high res formats.

[miguelito]

Is MQA-encoded file after unpacking a bit-perfect copy of the original file (as is the case with FLAC)? From what I heard about the technology, that's not necessarily the case.

It is not - and frankly there's no way it could be since the amount of data is greatly reduced. Meridian's people insist vaguely that it is, but it simply can't.

 

I went to a Meridian MQA demo in NYC - all the demo material sounded noticeably better in their MQA versions. In particular, they played a 192/24 file followed by the MQA'd version off of the same exact file. The MQA'd version sounded remarkably better and different - nothing of the sort of jitter reduction or details like that. That convinced me that there's definitely some euphonic filtering done in the processing of the file, it is NOT a lossless process. Not that I have a problem with euphonic filters necessarily.

[miguelito]

What all those (except HDCD) have in common, is that the concept comes from japan, where CD buyers seem to be more receptive to "high tech CDs".

But I think there's a side benefit here: all those CDs are very carefully mastered, and the few I have sound amazing. To me this means that "crappy CD sound" is more about "crappy mastering" than it is about the limitations of the medium.

 

In the US and Europe, studios also upgraded their mastering equipment over the time, and manufacturing probably also improved, but they didn't give their CDs new brand names for that reason. "24bit remastering" etc is about the only marketing hype they used, whereas the list of japanese CD "formats" is endless.

Right. I have plenty of "20-bit SBM" CDs from Sony Classical and the like, and many sound pretty good.

[Jud]

The MQA'd version sounded remarkably better and different - nothing of the sort of jitter reduction or details like that. That convinced me that there's definitely some euphonic filtering done in the processing of the file, it is NOT a lossless process. Not that I have a problem with euphonic filters necessarily.

 

I think we've got to be careful that even correct terms don't bias our thinking (not saying yours is). There's a lot of manipulation/filtering of digital signals that certainly isn't "lossless" in the proper sense (there is no mathematical operation for exact retrieval of the original signal) where the resulting signal has *more* information than the input, and sounds better (literally more euphonic) as a consequence. Ordinarily we'd think of a lossy process with a euphonic result as achieving euphony through *in*accuracy, whereas it's conceivable that digital manipulation/filtering can achieve euphony through *greater* accuracy, in the sense of greater fidelity of the analog output to the original analog input.

 

That said, I haven't done any reading on how MQA works and have never heard a demo, so the foregoing may be a useless aside in this particular context. Carry on.

[firedog]

But I think there's a side benefit here: all those CDs are very carefully mastered, and the few I have sound amazing. To me this means that "crappy CD sound" is more about "crappy mastering" than it is about the limitations of the medium.

 

 

Right. I have plenty of "20-bit SBM" CDs from Sony Classical and the like, and many sound pretty good.

 

yes. I have several that sound very good, but I assume it mostly is just better mastering - and maybe better manufacturing.

[CatManDo]

But I think there's a side benefit here: all those CDs are very carefully mastered, and the few I have sound amazing. To me this means that "crappy CD sound" is more about "crappy mastering" than it is about the limitations of the medium.

 

Well, there are plenty of examples where the mastering of these "newer technology" CDs from Japan is bad, with high dynamic compression or weird EQ. Most SHM-CDs and BluSpec-CDs use an existing mastering, and not always the best one. It's hit and miss as with regular CD reissues.

[Tony Lauck]

It is not - and frankly there's no way it could be since the amount of data is greatly reduced. Meridian's people insist vaguely that it is, but it simply can't.

 

I went to a Meridian MQA demo in NYC - all the demo material sounded noticeably better in their MQA versions. In particular, they played a 192/24 file followed by the MQA'd version off of the same exact file. The MQA'd version sounded remarkably better and different - nothing of the sort of jitter reduction or details like that. That convinced me that there's definitely some euphonic filtering done in the processing of the file, it is NOT a lossless process. Not that I have a problem with euphonic filters necessarily.

 

Sounds like you were the victim of a shell game. There are many ways to rig an audio demo.

[miguelito]

I think we've got to be careful that even correct terms don't bias our thinking (not saying yours is). There's a lot of manipulation/filtering of digital signals that certainly isn't "lossless" in the proper sense (there is no mathematical operation for exact retrieval of the original signal) where the resulting signal has *more* information than the input, and sounds better (literally more euphonic) as a consequence.

Understood. Upsampling for example.

 

Ordinarily we'd think of a lossy process with a euphonic result as achieving euphony through *in*accuracy, whereas it's conceivable that digital manipulation/filtering can achieve euphony through *greater* accuracy, in the sense of greater fidelity of the analog output to the original analog input.

Well, I don't know what you mean by "greater accuracy". Given a digital source, you have a certain amount of information. Adding information (such as in upsampling) might be done in an informed way but you're still making a guess, one that is euphonic, but a guess nonetheless.

 

That said, I haven't done any reading on how MQA works and have never heard a demo, so the foregoing may be a useless aside in this particular context. Carry on.

My understanding is MQA "folds" the spectrum, storing the higher freq components as a low level lower frequency component. I imagine it's like adding some very low level signal to a high level signal that can then be reinterpreted as a higher frequency signal. But you must be destroying information compared to the original higher resolution signal, albeit possibly information that is barely audible.

 

But I tell you: the difference with the 192/24 and the MQA'd version of that same source was nothing short of stunning, so at least in that case there was massive euphonization of the original material. Euphonization I say!

[miguelito]

Sounds like you were the victim of a shell game. There are many ways to rig an audio demo.

I don't think so. Official Meridian people were running the show. They might be opaque as to what the process is but I do not think for a minute it was a trick.

[miguelito]

Well, there are plenty of examples where the mastering of these "newer technology" CDs from Japan is bad, with high dynamic compression or weird EQ. Most SHM-CDs and BluSpec-CDs use an existing mastering, and not always the best one. It's hit and miss as with regular CD reissues.

I really only have experience with a few XRCDs that I have, eg: "Everybody Digs Bill Evans."

[Tony Lauck]

I don't think so. Official Meridian people were running the show. They might be opaque as to what the process is but I do not think for a minute it was a trick.

 

Caveat emptor!

 

You are the emptor. Meridian is the vendor.

 

I think it's obvious you've never run any kind of product demonstration, especially at a trade show, whether it was audio or any other product. If you did, you would know that any demonstration can be rigged and that this is the normal practice of people who are selling something. I know. I've done this myself (with computer communications equipment and when running computer system benchmarks). Is it cheating to select those demonstrations that put your product in the best light and avoid any pitfalls?

[miguelito]

I think it's obvious you've never run any kind of product demonstration, especially at a trade show, whether it was audio or any other product. If you did, you would know that any demonstration can be rigged and that this is the normal practice of people who are selling something. I know. I've done this myself (with computer communications equipment and when running computer system benchmarks). Is it cheating to select those demonstrations that put your product in the best light and avoid any pitfalls?

I wouldn't call it "rigging" to chose specific examples that highlight the virtues of what you're trying to showcase - that's all I'm saying.

[Tony Lauck]

I wouldn't call it "rigging" to chose specific examples that highlight the virtues of what you're trying to showcase - that's all I'm saying.

 

I used to "rig demos". (These were for computer hardware and software and this happened about 45 years ago.) I didn't actually cheat, I just made sure that nothing but the most favorable cases got shown. In one case, we were supposed to demo a suite of new hardware, programming language support and a new operating system release at the manufacturing plant, before the customer would sign off on the product shipment. After working all night, I got all but the very last item on the demonstration list to work properly, but the last item failed because of a bug in the (then brand new) file system and the developer responsible happened to be out sick. When the customers came with our salesmen, I took the salesmen aside and told him the situation. We arranged to space out the demos so that we didn't make it to the last item prior to lunch time. We also arranged for the customer to be highly lubricated with intoxicating spirits at lunch. The customer signed off, the machine was crated for shipping and during the time it took to arrive on the customer site we fixed the software bug. That's what I call "ethical rigging". Another example was a competitive FORTRAN benchmark where the competitive machine was 15% faster. I analyzed the performance bottleneck as being in the library routine that computes square roots. I rewrote the library routine and made it 25% faster, so that we won the benchmark. My new code was in the released software when the customer took delivery.

 

There are many games one can play with audio. The most obvious way to rig a demo is to pick the recordings one demos. This would not cross the line, IMO, but it could be highly deceptive. It would cross the line if an actual "studio master" file were degraded and presented as the unprocessed file to be "fixed" by the MQA encode-decode process. I don't know the people involved personally, so I can have no opinion on whether or not they are likely to have pulled such a stunt. Another common game is to mismatch the levels and boost the level of the version that is supposed to sound better by perhaps 0.5 dB. This can happen unintentionally. Level errors are a known problem that mastering engineers must deal with when comparing processing of a recording to see if they have actually improved its sound. Level mismatches can also be used to trick a customer into believing that there was an actual improvement when there was none.

[esldude]

 

Euphonization I say!

 

Is that like, you-phony-sensation??

[sdolezalek]

Caveat emptor!

 

You are the emptor. Meridian is the vendor.

 

I think it's obvious you've never run any kind of product demonstration, especially at a trade show, whether it was audio or any other product. If you did, you would know that any demonstration can be rigged and that this is the normal practice of people who are selling something. I know. I've done this myself (with computer communications equipment and when running computer system benchmarks). Is it cheating to select those demonstrations that put your product in the best light and avoid any pitfalls?

 

Tony: You are generally pretty sophisticated about these new technologies. But I can't tell from your comments here whether you have actually carefully looked at and/or listened to Meridian's MQA or are just being generally skeptical. From my own listening to MQA I have found some pieces to be extraordinarily good (those were direct recorded and MQA encoded by Meridian) and others that I thought were at best so-so (those were older recordings that in some way had MQA applied). From looking at the science though it seems that half of what Meridian is doing is very similar to what we at CA are doing -- start with 24/96 or better material, upsample to the DAC's native conversion frequency through something like HQ Player (choosing your filter settings carefully based on the original recording) and then play it through a high quality DAC. The other half is quite different, because it does involve introducing losses (to minimize data and allow the whole thing to be streamed as though it was 16/44) but choosing where to be lossy at frequencies where (according to Meridian) our ears aren't sensitive to it.

 

Curious as to your more detailed views, but assuming you have taken the time to form them based on the limited amount of specifics that Merdian has been willing to let out.

[Tony Lauck]

Tony: You are generally pretty sophisticated about these new technologies. But I can't tell from your comments here whether you have actually carefully looked at and/or listened to Meridian's MQA or are just being generally skeptical. From my own listening to MQA I have found some pieces to be extraordinarily good (those were direct recorded and MQA encoded by Meridian) and others that I thought were at best so-so (those were older recordings that in some way had MQA applied). From looking at the science though it seems that half of what Meridian is doing is very similar to what we at CA are doing -- start with 24/96 or better material, upsample to the DAC's native conversion frequency through something like HQ Player (choosing your filter settings carefully based on the original recording) and then play it through a high quality DAC. The other half is quite different, because it does involve introducing losses (to minimize data and allow the whole thing to be streamed as though it was 16/44) but choosing where to be lossy at frequencies where (according to Meridian) our ears aren't sensitive to it.

 

Curious as to your more detailed views, but assuming you have taken the time to form them based on the limited amount of specifics that Merdian has been willing to let out.

 

I rely on the mastering engineer to polish the sound I hear when I play commercial recordings. I know how to do this (within the limits of my experience, room and equipment) and know that doing a proper job can easily take a day's work for an hour's worth of music. I do this for older analog recordings that I remaster, because it is worthwhile to spend 6 hours to improve an hours worth of music if a dozen or more people enjoy it. I have done this to a few CDs, including one Brucker Symphony conducted by Ormandy, but this was a learning exercise, not a realistic method of improving my record library.

 

I do not believe that Meridian has any "magic" recipe for automating the resampling procedure. I know enough about the choice of resampling filters (downsampling and upsampling both and how they interact) to know first hand what is achievable and what is not. Filters can be a significant issue at 44 kHz, at least to my ears. As the sampling rates increase the benefits of tweaking filters diminishes and this includes all of the filter parameters, e.g. minimum phase vs. linear phase, cut off frequency offset, filter steepness, dither level, noise shaping, etc... These benefits will be minimal at best and if they are demonstrated in a way as to seem otherwise, I consider such demonstrations to be dishonest.

 

As to saving bits that have to be stored and transmitted, I agree this is definitely possible. But what is the point today? Bits are essentially free in comparison to the prices that have to be charged to compensate the composers, lyricists, performing artists, engineers, middlemen and retailers in the distribution chain. Do we need an "MP3 for high end audio"?

 

Were there to be detailed technical descriptions I might reconsider my opinion. In the absence of this information I consider this a non-solution to a non-problem, and a potentially evil one at that, since MQA appears to be a proprietary format. If MQA were open, then I would devote some effort to understanding it. An example of an open format is FLAC. For FLAC there is a specification and a reference implementation and there are no intellectual property restrictions. In the absence of this situation for MQA I will be an active opponent. We do not need any new digital formats, especially proprietary ones.

[Jud]

In the absence of this situation for MQA I will be an active opponent.

 

It would have been more forthcoming to indicate this source of potential bias when you first began warning miguelito (and by extension the rest of us) of the potential for demonstrations that were in some sense "rigged."