DSD mastering and quality

Could not agree more with the point I think you are making…

My best recordings are the ones where the music is “well played”, engineered well and “well recorded” - resolution be darned! :grinning:

4 Likes

Yup - there is this reductionist tendency we have with respect to the processes and the gear that often loses touch with this ground truth.

One DSD I bought is, as far as I am aware, a straight analogue tape to DSD transfer with no editing (other than splicing) and hence no conversion needed.

How common this is, and how transparent companies are about their processes - well, who knows?

Here’s another magnificent recording, they are clear it was recorded in 24/352 (DXD) and is sold by the label for €20.

When the same album gets to nativedsd.com, they offer DSD conversions, upsampled at different rates, increasing the price by up to 50%.
Tabea Zimmermann: Solo II - NativeDSD Music

I can only think this is a scam, even though they still say it was recorded in 24/352 DXD (really PCM). What can the benefit possibly be?

I think I bought it in 16/44 PCM from Presto Classical for €10.

These are technical matters and it’s not a scam. Hence my (and at least one other’s) interest in the question.

Virtually every digital conversion or handling results in losses. Over time more studios will record in higher res formats like Octave, at least DSD256 and possibly DSD512 and DSD1024.

Losses through these processes will matter for the final product of both good and bad performances. If there are losses back from the DXD edited master, it may make no sense to return to DSD, which will influence equipment choice. dCS only offers playback up to DSD128 for example, while MSB and others offer up to DSD512, some up to DSD1024.

With respect to NativeDSD, in my experience they are good guys and will answer questions about recordings directly. Their site provide relatively good data on music providence, unlike HDTracks for example, which won’t tell you anything about their files. I will not purchase from HDTracks as a result.

1 Like

I think part of the point nowadays is that though there is inevitably some loss, most of the very high res digital gear is “near as dammit” with respect to that particular aspect. Approaching vanishingly small. There are other areas where you’re likely experiencing more “loss” or which have more of an effect on the sound than the conversion alone, such as cables and power.

Having said that, converters do sound different, and the same digital chip can sound different depending on the “A” side electronics.

Makes perfect sense…

Probably the majority of DACs these days are made with Sabre chips (especially since the AKM factory fire) - though they keep improving them, and there are lots of iterations. The main difference in the sound is the quality of the analog output side rather than the digital chip. It is the more costly aspect, so most do it with cheap op amps.

1 Like

Only just over half of the items listed on NativeDSD are actually recorded in DSD. The rest are PCM or analogue transfers.

High rate DSD presents real hardware problems for data transfer. dCS only accepts up to DSD128 because that’s the limit at which it operates, even when upsampling.

One hour of DSD1024 is something like 20gb of data. That’s about 50 CD rips. Madness.

MSB DACs cost $50,000 to $100,000. It’s the world of fantasy audio for the criminally wealthy.

None of this is lossy. To the contrary, it involves vast amounts of data. I think the highest rate for DSD data capture is DSD256.

Studios are not going to start doing DSD. The new titles on nativeDSD seem to have dried up. They would need to buy Pyramix, which is very expensive. Linn Records was the largest single user of DSD and issuer of SACD, they used it for 15 years, then abandoned it for 24/192 PCM as their standard.

The market for purchased downloads is quickly diminishing, and you can’t stream DSD.

If Octave and Cookie Marenco can make sustainable businesses using DSD, then fine, but it failed as a mainstream format years ago. That’s the reality.

If in reality it’s high rate PCM, and sounds good, what difference does it make? I would hope people are buying for the music rather than the format. My disinterest in DSD is not just because I can’t hear any difference, but because virtually none of the music I listen to is actually produced and issued in DSD.

We can certainly agree on that ^^.

SEE

Not trying to be argumentative, I just think there is a lot of opinion in this post:

That’s not a technical limit. It’s just what dCS offers today.

Why? Storage is free. There are 2TB thumb drives. No one has trouble storing vast amounts of music or HD video.

MSB discrete is less, but that’s beside the point. dCS, MSB and other brands offer products in that price range. Of course you are entitled to your opinion.

This is likely factually untrue, and the heart of the question.

There are many reasons why a product or service can fail at one point in time, and succeed later. Fujitsu, not Apple, had some of the first tablet computers. They failed wildly at the time, because of, among many reasons: poor OCR, poor LCDs, poor network speeds, insufficient computing power, etc. The tablet is now a multi-billion dollar global category.

…but do you think once the DSD recording got hires PCM during editing, bringing it back from there to DSD helps anything? I thought not, that’s why we asked if it makes sense.

1 Like

I think…it depends😝 It is not unlike (in my thinking at this point anyway) having a system that is well set up for analog. You may prefer the sound of that, since that is what you’ve put the effort and money into. You have developed a system synergy over time oriented around that.

If your system is well sorted around DSD, it would be at least as good or “better” (say, because you like the sound of your DSD DAC) than sticking with the DXD. Fortunately in the case of Octave, you now have the option of either, and can decide for yourself.

Some gear/chips may not play DXD natively, if at all, much as some don’t do DSD. As has been pointed out in other posts, if you have certain makers’ DACs that are based on PCM, you may want to lean that way.

I just think that nowadays, at a given high-ish level of system, IT IS ALL GOOD. Better than good. Heck, Redbook is amazing on most systems now.

IF IT WERE POSSIBLE to have all things be equal (which it ain’t) the differences between the playback systems would be very, very small. But you tend to have to buy into an ecosystem, and various makers’ make choices, and their products have inherent sounds - in no small part due to the ANALOG sections, and their attention to power supplies and so on. The same is true in the studio.

2 Likes

So I understand you think there might be DAC‘s which sound better with the PCM edit format and others which sound better with a DSD reconverted format, right? At least that’s what could make sense to me. And you think if one optimized his setup towards DSD, the PCM format, a DSD recording was finally edited in, could sound less optimal? That’s hard to get for me.

Where I come from is:
Ted in the past even stated for his DSD! DAC’s, that it makes more sense to play a native PCM format in PCM (because the DS converts it to DSD anyway) and not to externally convert it to DSD before. That’s why I didn’t have an idea why a back conversion of the DXD/PCM file to DSD should make sense for any DAC (if it even doesn’t for a DSD DAC). Because after editing, DXD/PCM is the new native format. How this can still inherit the DSD advantages although it now is PCM, is still a mystery to me, but I count on it.

I guess if there’d be a clear answer on hand already, we might have seen one from Paul…so let’s wait and see :wink: If it’s not answered here, I might try to knock on Ted’s door.

1 Like

I think there is a lot of truth to this and we have not quite yet accepted that (nearly) everything matters in terms of controlling for variables that can affect the ultimate sound of recorded music through headphones or loudspeakers.

I also speculate that “we” have not yet identified the necessary and sufficient factors that must be addressed to exert “control” over the quality and nature of what is ultimately reproduced.

That said, I believe we are much further along in developing this data base now, then we were even 5 or 10 years ago.

Just my unscientific (technically unqualified) observation…

SEE

To be honest I wasn’t thinking about this at all.

I’m mostly interested in the optimal signal path, losses and provence. I have owned dCS for many years but am not wedded to them. I am a fan of Paul’s, and Octave, and really just want to understand the benefits of DSD.

It makes sense to me that if one records in DXD, and edits/masters in DXD, that one should release the file in DXD, AND state exactly what the file journey was, i.e., provence.

If one records in DSD, and masters in DXD, I have the same question as you @jazznut: (a) why round trip to DSD at the end? AND ALSO (b) are there losses in doing so? and (c) what is gained by doing so, if anything?

: )

1 Like

Yes same for me, this wasn’t my initial thinking…I’d just be interested what sense this forth and back conversion makes.

I’ll come back to these questions later - trying to get some stuff done.

One answer is that people want stuff in DSD format.

Let’s get real for a second.

On nativeDSD, and there may be other DSD sources, there are 260 titles that originated in DSD256. Anything issued higher has been upsampled. The vast majority are DSD64, which is the limit of the Sonoma system used by Octave.

Screenshot 2022-08-07 at 18.50.17

Labels like Linn Records and Alia Vox, that have large DSD archives, don’t even see any point of making them available to purchase.

To put this into context, Amazon Music have almost 100 million tracks and Qobuz has almost 500,000 HD albums in PCM.

So DSD is the literal grain of sand in the dessert, and it’s not new, it’s been around for 20 years.

So why would any manufacturer waste any money investing in high rate DSD capability?

Does the number of titles suggest to you anything about the general interest in DSD in the recording industry?

How many services will want to store multiple HD versions of files consuming vast gigabytes that next to no one will ever buy?

As fas as “lossy” is concerned, for most people 16/44 and above is lossless and anything below is lossy. If you have a problem with one of the 260 commercially available DSD256 recordings being downsampled to DSD128 is lossy, then I can’t help you. Life is too short.

1 Like

Life IS too short, furthermore, it is too important to be taken seriously!

But back to DSD; not all DSD (or any other format) are not created equally. The care taken during recording, as noted above, is key. Well recorded music is always a joy to listen to.

Now, back to life.

3 Likes

Well performed live music is even better!

2 Likes