DSD mastering and quality

I’ve been watching Paul’s videos on YouTube for years and have both enjoyed them and learned a lot - thank you!

Also, I am a fan of DSD and love what smaller recording studios like Octave are doing to push the envelop on quality and sound.

I have a couple questions for @Paul and apologies in advance if these have been answered elsewhere:

(1) is it true that all DSD recordings need to be converted to DXD for editing and mastering?
(2) if (1) is true, is this conversion lossy or lossless?
(3) if (2) is lossy, why not leave the final mastered copy in DXD and not have additional losses on the roundtrip back to DSD?

Thank you!

Already been answered here:

See also:



DXD is just high rate PCM.


You might be interested in this label, too.

Todd the owner says, he usually doesn’t master/edit at all, no matter if he records in PCM or DSD. I think he mainly uses PCM format. He rules out everything at the recording stage. So his recordings should be straight through…and they usually are 2 mic recordings where possible as far as I remember, which seems to make a very natural soundstage/ambiance.

But as I said, no mainly DSD label.


1 Like

Hi R, thanks for the questions.

First, DSD cannot be mixed or volume adjusted. It is what it is. So when we record the input microphone preamplifiers are used to set the gain.

Once recorded we have two choices for mixing: DXD or analog. In the past, we have gone analog but that has a number of disadvantages (minor). In our modern methods we chose DXD. Converting to DXD in the way we do it (on Pyramid or Zephiir) is almost (if not completely) indistinguishable from the original. I can’t hear any difference.

When we’re ready to master, we leave DXD as the master. In fact, as of yesterday, you can now purchase Octave masters at 352.8kHz (DXD). We then convert back to DSD by going analog and then back to DSD at 4X and 1X.


So you mean your DSD downloads are one or two conversion steps away from the PCM master? Why then offer DSD download format at all?

It’s strange that DSD seems to be such an advantage although just present at the recording stage.

1 Like

Thank you @Paul. I really appreciate you taking the time to respond directly. I own a business too (smaller!) and recognize that it’s a big additional effort after all of the other things that running a company entails.

My questions are from the position of a supporter, and just trying to understand.

Similar to @jazznut below, why bother to go back to DSD if it’s only the recording/capture format? It would seem that the master DXD would be the highest resolution end file?

Thank you,

I bought Sera Una Noche CDs and Vinyl from him. Todd Garfinkle is an expert in high quality recordings. For him it seems an art that needs perfection. The acoustics of Sera Una Noche, recorded in a Church close to Buenos Aires are captured fantastically.

Like @jazznut says he tackles every acoustics issue at the recording stage. With the right gear, you won’t get much closer to reproduction of a life experience. Although I would love to visit Argentina once.

He is the proof that, when recorded correctly and not “mastered” endlessly, the sound quality is much less depending on the format (PCM vs. DSD) than on the art of recording.

For vinyl Todd looks at the quality of the pressing process too. I bought 2 copies of the Nublado 12” 45 rpm records on red and black vinyl. Both were pressed in the Philips, later Sony record manufacturing in Haarlem, The Netherlands. As Todd ordered them when he lived in California then moved to Japan, where I ordered them. Those records travelled around the world. Funny fact!

Todd if necessary washes and flattens them manually as he wants to deliver a product that’s more perfect than it comes even out of the better pressing plants.

I wish he had a tape machine, too. His recordings would be great to have all analog, but great as they are, too.

I have all those 3-4 meanwhile rare Sera una noche LP’s/DLP’s, dealt around 500-600$ for the bunch now I guess.

Would still be interested in the above format thing…maybe if others or a few more ask, too…

1 Like

I wish people would stop using the term DXD. It’s 24/352 or 24/384 PCM.

The reality is that the vast majority of music sold as DSD has been PCM at some time, just as it turns out all analogue MoFi vinyl had been digitised.

There are all ivory tower arguments, personally I can’t hear any difference with these high resolution formats.

1 Like

While I don’t agree with Steven about the hearing I do agree about the use of the term DXD (a term which I too often use).

Wish I remembered the story about its origin. Gus Skinas had shared with us that way back when, when Sony was launching DSD, Pyramix was touting something about 352.8 PCM as being identical to which Sony came down upon them hard. I think it was the folks at Pyramix that coined the phrase DXD.

Agreed on all of that, Steven. DXD was developed by Merging Technologies and Philips to edit DSD in PCM. It takes special gear to record and edit in it. Then when very high res PCM files started to become available to the end user, it sort of turned into this other “format” like DSD.

I guess I think of DSD as the “tape” of the digital world. It is a digital format that is not PCM, and so is unique and has its own sound (or lack thereof). So there are reasons to use it as both an acquisition format and delivery format, much as one would use tape as an acquisition and delivery format in analog. But without the analog issues. But if you want to edit it, this is how the sausage is made.


If those pursuing some imagined state of audio minimalist perfection feel that digital is the way to go, I imagine such a purist being horrified to find that sometimes people take digital recordings and run them through various sorts of analog equipment in the midst of the process in order to inject some “vibe” into what can be a very sterile-sounding process🤠

1 Like

you’re mean


I have no problem with the term DXD. We know it is 352.8/24. It is merely a name.

We similarly know Red Book is 44.1/16. (Yes, Red Book is also a great deal more as it specifies a format as well.)

That’s my recollection as well, Paul. Pyramix is a DAW product by the Swiss company Merging Technologies. They do a lot of Egyptian-themed product names; Horus, Anubis, Pyramix, etc.

My and @jazznut’s question remains unanswered:

If all DSD is brought into DXD for editing and mastering, why round trip back to DSD?

There is massive conflation between the use of high data rates for (a) capturing and (b) processing.

16/44 was chosen in the 1970s as able to capture the limits of human hearing, and streaming has matured to the point that from my perspective the most popular rates for recording, mastering and streaming classical and jazz is 24/96 and more “popular” music seems to be streamed more at 24/44. This of course is for the 3%, the other 97% stream at 16/44 or less (mostly mp3).

24/96 presents no bandwidth issues, nor does 24/192 these days, but 24/192 files are a bit large for storage.

dCS developed 24-bit A/D and D/A converters from the late 1980s, for professional use, mainly working with Bob Ludwig, introducing higher sample rates, first 24/192 in 1993, 24/352 and 24/384 PCM. They also developed DSD processors for the first SACDs produced, again by Bob Ludwig.

With these many sampling rates in professional use, they (and others) developed the first digital to digital converters (DDC), again for studio use. At this time, in 2000, by accident it was found by a dCS distributor that using a DDC with a 24/96 DAC improved sound quality, so dCS released the 972 model as the first upsampler for consumer use.

Of course upsampling is commonplace these days, whether internal or external, and my own unit upsamples everything to 40/384 PCM. It is no surprise to me that external upsamplers tend to be made by manufacturers using proprietary D/A conversion (dCS, Chord and Denafrips come to mind).

The first external upsampler I heard was the dCS Vivaldi in 2013, and I was sold on what it can do to humble 16/44 from a CD.

So my experience is that you have to be very clear to delineate the benefits of using high data rates for signal processing, which can be significant, and high data rates for source files, which I’m not convinced has any benefit.

“If all DSD is brought into DXD for editing and mastering, why round trip back to DSD?”

Well, Paul’s the guy you’re asking, and the guy with the record label, but my two cents is that as mentioned above, it has a sound that is different from PCM. PCM has had issues from the getgo sonically, though they get fairly well ironed out in the higher resolutions and with modern chips. In the digital realm is is about the sound of the conversions from A to D and D to A. And DACs like Ted’s are about DSD.


I pretty much agree with this as well. I had a friend ask me about getting into playing higher res stuff, and I pointed to the sound of the Redbook Malia “Convergence” album your dealer uses as a demo. I’m fine with the sound of that for delivery. And that was streamed on Qobuz!:stuck_out_tongue_closed_eyes:

Though as always - this gets back to it just being a good recording of good performances. I don’t actually know how it was recorded, but Boris knows what he’s doing.

1 Like