Direct Stream Digital

[manisandher]

There seems to be quite a bit of a buzz around DSD in audiophile circles nowadays. Many DAC manufacturers are placing DSD capabilities into their DACs. Some reviewers felt DSD gave them the best sound at CES2013 with true DSD surround sound recordings (http://www.stereophile.com/content/my-epiphany-pass-labs-sony-kimber-and-dsd-files).

I've embarked on my 'vinyl digitization' project and hope to gain some insights on the whole DSD vs. PCM debate. I have a true 1-bit DSD recorder and also a true 24-bit PCM recorder. But comparing the two is more easily said than done because I can't use the NOS1 to listen back to the DSD recordings natively. Right now, I'm converting DSD128 to PCM24/88.2 using Korg's Audiogate software and using these files with the NOS1. The results are already very, very good. I've not yet had the time to compare them to native PCM24/88.2 files, but will do this soon and post the files for anyone to listen to.

So a simple question: will XX ever be able to play DSD files?

A more involved question: what's your gripe with DSD in general (I'm pretty sure you have one)?

Mani.

[PeterSt]

Hey Mani,

First question which comes to mind is why you'd ever want to listen to DSD through the NOS1.
Unexpected question ?

My point is, putting it that way you start out with "mixing" things, and although you may expect that the result through the NOS1 will be better than your Mytek I don't see the reason for comparing it that way. I mean, when DSD would be better all the way, then you'd need a PCM D/A converter to compare it with. Of course you can say something like "yea, but that Mytek isn't the best to begin with", but then what ? Wouldn't all be apples and oranges then ?

But this (my thinking) is also implied by the workflow you talked yourself into. So, starting out with genuine DSD and converting that to PCM will just never be the best. It makes no sense. But also, you start out with your post with recording itself, and when it were about *that*, it could be a whole different story. For example, we could go as far as stating that native DSD recordings come close to 24 multibit recordings or even outbetter it. But then your post mixes up with native DSD playback, up to XXHighEnd being able to play DSD with the sort of obvious suggestion that it still needs to play through the NOS1 and without the suggestion that the NOS1 should better do DSD because probably the Mytek isn't the best at it and ...
Anything else ?

haha

I think (from elsewhere) my statement about DSD is a clear one :
If you try to count OUT the apples and oranges and only hunt for the best possible and where NATIVE DSD is your only option for that, then DSD is useless. There's just no native DSD around (less than our so called hires) or we can't judge it to be native or not in the first place.
I am NOT saying that this is the reason that XXHighEnd does not support DSD playback, because this is only not so because I didn't give it priority, PLUS I see the flaws in DSD which is another subject. But XXHE will. And even through the NOS1 and in a better way than ever done (well, that's what I foresee - spent quite some time on it already).
But useful I see it not.

Regarding the latter, let's say that half of the material we own has been converted by whatever means from DSD already. So, nothing different from what you (and many) are trying to do yourself.

But let me remind you : it merely seems that you are after the recording part of it all. To me this would be

DSD -> Mytek (native DSD DAC - ehm, if so at al).
or
PCM (like with PMII) -> Native PCM DAC (= NOS1).

No combinations are allowed, because it will degrade or be moot in the first place. UNLESS we make this combination :

(record in DSD because we think it will be the best - no filters !)
DSD -> XXHighEnd -> PCM-sort-of -> NOS1.

Is it now clear what your question is about ?

Peter


PS: Notice that this all is still useless to all but you, because we are not in control of the recordings. So again : There's no native DSD around. Maybe a rare case of a Cookie M. live recording without any mixing.

[manisandher]

Edit: Really, the biggest issue using the Mytek for playback of DSD is that I'd have to introduce a preamp into the chain. I'm sure you'd agree that this is a total no-no...

Yes it's true that I don't 'trust' the Mytek to be able to get the best out of the native PCM24/88.2 recordings. So using it to compare native-DSD vs. native-PCM is a no-go - I'm sure the DSD would win hands down. But that's not to say that the DSD recording is actually better (but it could be).

The DSD-to-PCM conversion with Audiogate is just an experiment to see what it sounds like. But I agree, ideally there should be no conversions whatsoever (except done 'properly' in XX).

OK, so there are 3 options:

1. DSD -> Mytek

2. PCM -> NOS1

3. DSD -> XXHighEnd -> DSD-to-PCM -> NOS1

As you mentioned, 1 and 3 are attractive simply because of the lack of filtering during the ADC process (Edit: which answers your first question as to why I'd want to listen to DSD on the NOS1).

Now the BIG problem I have is in comparing 1 with 2. I don't have a totally transparent preamp, and without this making a comparison is almost impossible. But if 3 were available, I could compare 2 to 3 easily.

I think I've stated before that the way I will proceed is to record everything twice; once in DSD128 on the KORG and once in PCM24/88.2 on the PMII. (Why 88.2 and not 192? Well because I can't use the PMII in 'master clock' mode with the Tascam recorder at 176.4/192 rates. I could revert to using the RME or Weiss interfaces at 176.4/192 rates, but really don't want to use a PC for recording. In any event, I think 24/88.2 should be plenty good enough for vinyl recordings.)

Sorry this subject is not of much interest to the vast majority of people here. Maybe it will be in the future?

Mani.

[Jud]

Thought I read Cookie M. saying somewhere there's now the means to do mixing in DSD format, but I don't know nearly enough about the process to (1) understand whether or not that was what she was actually saying, and (2) know whether this makes sense as something that would keep a DSD recording in "native" format through the mixing process, or whether it is some sort of "DSD-wide" thing.

I'm also a little confused about what actually happens on the D/A end, i.e., whether 1-bit/DSD is truly the "native language" of most DACs out there (not the NOS1, right?), or whether this is as usual something oversimplified for non-engineers like me.

[manisandher]

Jud, not sure if this will help, but here's my modified version of a diagram originally posted in the A. Koch paper.

The top process is what 99.9% of people are experiencing. And the 'killers' here are the DSD-to-PCM conversion in the ADC and the PCM-to-DSD conversion in the DAC.

Mani.

[PeterSt]

Jud,

DSD mixing is possible (for quite some years now) with a conversion which is "to PCM" like. But let's say this is harmless ...

Then still it is so that this is only from lately, while we can easily see (derive) from hires material brought to market, that this is never new stuff (or count on one hand). And so, no reason to see that there will be new DSD stuff around (read : whatever it's constructed, it was or mixed in PCM and brought back to DSD, or it wasn't DSD in the first place OR it was mixed in the analogue domain and then put back to digital DSD (this is what Cookie is doing IIRC)). Here too : or count on one hand.

Peter

[Jud]

Mani -

Yes, this is the diagram I've seen, and it matches stuff I've read also.  But are the conversion from DSD to PCM in A/D on the chart that you've described as "killer," and the one available in DSD mixing of which Peter says "let's say this is harmless" the same (type of) process?

I'm trying to get a start on figuring out if there is a devil in the details here....

[PeterSt]

Jud, allow me for my take : This depends;

DSD could be mixed for a longer time by blundly converting it to PCM first, and when ready - back. This is an obvious one.
The more modern way needs all kind of trickery and as far as I know this uses a few bits in a PCM way, like sigma delta converters are not true 1 bit anymore just the same (in the beginning they were).
At least I would not be able to decide how deteriorating that is. But mind you, I am also not able to say how deteriorating the SDM->PCM conversion during the A/D process is.

All we know are these sort of facts :

1.
Mani prefers Redbook through the NOS1 over DSD throug the Mytek (while it is said that the Mytek does not convert) but with the notice that DSD of any sort WILL be hires, while Redbook clearly is not. Mani may correct me here, but this is the last I read from him. Keep in mind though, that our Mani is the only one being able to make the comparison while he has both the Mytek and the NOS1.

2.
We can never know how native the DSD is, unless we are pointed to them explicitly and which HAS to be a not after-mixed recording (otherwise we can't tell how deteriorating it will be).
but
2a.
As long as the recording is taken ourselves we do have a good comparison, but again it is Mani who seems to be the only one, but there's a but once again : we may think the PMII records in 24 bits PCM natively ... but we don't know really (here too Mani may come up with new data).

3.
Don't listen to me only, me NOT being the one knowing all about DSD (or SDM converters for that matter). This is how I say that at least it is not "honest" to not have Miska around here. And I will keep on saying this as long as it lasts that I don't know everything about DSD. PCM yes.

4.
That said in #3, I do say that the residual high frequency noise from DSD is correlated to the music, just because I can see that. This, opposed to Miska who claims it is not and who should win this game because of his knowledge on the subject.
Mix this up with the inability to have PCM all the way right, read : have no ringing on one side and no harmonic distortion hence high freqyency noise because of *that* on the other.

I say that the whole thing is damn complex to compare from theories only, which is the case because what you may gain on the one, is lost on the other in another subject. This already is apples and oranges. One solution to this only : listen and compare.

And as a bonus I will give you that all is 200% moot by guarantee, because from one day to the other the sound is so much improved on the PCM side alone - which is, say, my job, while in my view on the DSD side nothing happens. So what could be true 2 years back, is obsolete today. And if true today, then obsolete tomorrow.

And in all add to this that I personally don't even believe the Mytek doesn't convert. But that's another story. Not wait, it is not, because it is just all about it : how to play music with the least number of conversions, each conversion assumed to deteriorate.

And what to do with the latter, knowing that all the DSP sh*t applied by mixing (and recording) engineers does just this in the fist place, also in PCM.

Does anyone still know what we are trying to compare ? Well, I tell you no.
But let me tell you that I started the whole "NOS" project which turned out the way it did, just because "NOS" could not be right but sounded right to me. This is similar; I will do DSD (up to the DAC if needed) only because I can't find truths otherwise. But let me make clear that while at the start of the NOS project I had a clear aim of improving sound, I do not see the need for that today at all. And I hope a few people will agree with *that*.


Peter