Native DSD files, HQPlayer & Mytek Stereo192-DSD

[PeterSt]

"NO linux drivers available for the NOS1?"

 

No. And none in the pipeline either.

 

But having heard the affect that the NAA has on the sound, I will push and push to get PeterSt to sort something out. If the NAA can transform the Mytek in the way I've experienced, I can only imagine what it would do with the NOS1. What I'm absolutely convinced of is that I've only heard a fraction of what the NOS1 is really capable of.

 

But I'm not holding my breath.

 

Well, we are talking about some strange apples and oranges of course, out of which I try to make apples. So, once Direct DSD is available for the NOS1 it should be so that only *then* the next question arises : will Linux be necessary.

And FYI : I am in the stupid situation that I can play Direct DSD alright, but can't test it. IOW, I too need DSD in the NOS1 for myself.

What's also not to be underestimated is what can be done with the DSD filtering in -again- advance of the hardware. We can well say that this is a whole new can of worms, which -as I found- seems to start with the DSD data itself (see earlier post). Here some new phenomena seem to be in order, if one only thinks about how "direct" the A/D was taken (opposed to PCM) which leaves no room for others to blame. Still I can see the differences so again something must be in order I never knew. Or we didn't - I'm not sure.

 

Back to Linux ...

When there was a brief talk about this in another thread a few weeks back, I investigated a few things and let's say Linux drivers will be a possibility. But now the same problem as ever occurs ... where is the Linux player doing what I want.

 

All 'n all what we start see happening is that hardware starts to become dependend on software. Not really, but for best performance it is.

No wonder of course when we start to move the software in the hardware to outside of the hardware ...

 

Peter

 

 

[Miska]

What's also not to be underestimated is what can be done with the DSD filtering in -again- advance of the hardware.

 

That's one option too, best combined with DSD upsampling for example from DSD(64) to DSD128. And of course one of the DSD->DSD things for the beta2 is also DSD128 source file to DSD(64) output.

 

But, I'm going to have a "Direct SDM" setting too - just to keep everything optional as before.

 

 

[PeterSt]

But, I'm going to have a "Direct SDM" setting too - just to keep everything optional as before.

 

What I don't get ...

 

Since the Mytek is using the Sabre, I'd say this is nothing like what I tend to call "Direct DSD".

 

Am I wrong ?

 

[k-man]

But, I'm going to have a "Direct SDM" setting too

 

Am I right to assume this will "benefit" dCS/Meitner DACs with their built-in DSD filtering?

 

Congrats Miska!

 

[manisandher]

"... let's say Linux drivers will be a possibility. But now the same problem as ever occurs ... where is the Linux player doing what I want."

 

But I'm not using a Linux player! I'm using HQPlayer on my PC with W7. My feeling is that if it's possible for HQPlayer to stream data to the NAA via ethernet, then the same should be true for any windows player, including XXHE. Why not? Well, you'd need to get the necessary Windows 'code' from Miska I suspect, but this would just open the NAA up to so many more people who have their preferred players. However, I'm not sure if it's totally in Miska's interest to do this.

 

Another point: Of course, XXHE couldn't be used in 'Minimise OS' mode. But the beauty of the NAA arrangement is that it wouldn't make any difference to the SQ anyway.

 

Mani.

 

[Miska]

Am I right to assume this will "benefit" dCS/Meitner DACs with their built-in DSD filtering?

 

Yes, in a way that those can continue to receive pure unprocessed bit-perfect data from a file. Just like in current versions for DSD to DSD playback, where a specific setting is not needed.

 

Of course it is also useful for comparing with/without processing cases.

 

 

[Miska]

Since the Mytek is using the Sabre, I'd say this is nothing like what I tend to call "Direct DSD".

 

Well, the setting is about the software behavior, not about the DAC. I try to be somewhat DAC agnostic.

 

What comes to Mytek DAC, I'd like to refer to the Mytek thread where this topic has been discussed already.

 

 

[Miska]

I wanted to say a bit about the NAA hardware and software to address some of the questions in more detail.

 

Hardware:

One of the ideas is to make it possible to keep the hardware talking to the DAC as simple, quiet (acoustically and electrically) and low power as possible. Possibly powered from a linear PSU, etc. And to isolate it galvanically from the computer performing all the processing intensive activity.

 

Software:

On the software side idea is to minimize amount of computational activity on NAA, both from the OS and from the audio specific process. The asynchronous FIFO also provides logical timing isolation between the two sides. This is why there's a small optimized software module running on a stripped-down optimized Linux OS.

 

 

[TimDH]

Hi Mani,

Thank you (and Miska) for pushing us forward here. I would feel better if we had a firm theory for why having a lot of processing on the computer connected to the DAC affects sound quality (I know there are a bunch of ideas, but I haven't seen any verified), but I accept that it's so.

 

Your comparisons with DSD on the Mytek after streaming via NAA leaves the comparison to the NOS1 a little apples and oranges. What would you say about the comparison using RB material. Granted, you have HQPlayer vs. XXHighend also as variables, but this might isolate the benefit that the NAA brings to the table, separate from whatever qualities DSD has.

 

Thanks!

Tim

 

[PeterSt]

I would feel better if we had a firm theory for why having a lot of processing on the computer connected to the DAC affects sound quality

 

Nothing much for an answer Tim, but I wonder to what degree Mani really can judge that nothing changes "anymore". I mean, one player only to check, no explicit "SQ dials" in there ... that seems no good base for judgement.

 

Let's say it maybe isn't different from my own early judgement that sound couldn't change anymore with the NOS1-USB you have too. It did. But it took a couple of DACs to be out - people of course dialing more than I can imagine myself - to find out.

And now there are no dials ...

 

Not that it can't be true of course !

Regards,

Peter

 

[falcongate]

The NAA is very interesting indeed. But there are a couple of things I don't quite get. Would you mind to shed a bit of light?

 

- Generally, we want our music-playing computers to do as little unrelated things as possible so we often wind up with dedicated computers. With the NAA approach, does it matter if we may use a general purpose pc to act as the host? Or are we still better off shutting down unrelated activities on this host computer when we use HQ Player?

 

- What are the advantages/disadvantages over the embedded version?

 

Thx

 

[Miska]

- Generally, we want our music-playing computers to do as little unrelated things as possible so we often wind up with dedicated computers. With the NAA approach, does it matter if we may use a general purpose pc to act as the host? Or are we still better off shutting down unrelated activities on this host computer when we use HQ Player?

 

It doesn't matter, since the player and audio output are isolated from each other. All the sensitive tasks are isolated into the NAA (and NAA itself consumes very little CPU). So no need to do anything special about the computer running HQPlayer.

 

- What are the advantages/disadvantages over the embedded version?

 

Embedded version can also be used with NAA. The main difference between Desktop and Embedded versions is that Embedded splits the player itself into two parts - player server and control GUI client. So the playback system could be distributed into three separate devices. Since some of the DSP tasks can be processing-intensive, it may be useful for some setups to keep powerful computer away in a separate room or such. Although similar result can be achieved with Desktop versio too, using Remote Desktop client or such for controlling it, for example from a MacBook Air or similar.

 

Another difference is that Desktop GUI is optimized for use with mouse, touchpad or other accurate pointing devices. Embedded GUI is optimized for capacitive touch screens.

 

 

[manisandher]

"I would feel better if we had a firm theory for why having a lot of processing on the computer connected to the DAC affects sound quality (I know there are a bunch of ideas, but I haven't seen any verified), but I accept that it's so."

 

Hi Tim, quite a while ago, I did some experiementing with underclocking/overclocking my PC running XXHE. IIRC, underclocking seemed to 'smoothen' the sound, whilst overclocking gave it a more 'lively' feel.

 

More recently, I've done some experimenting with power supplies. Interestingly, SMPSs seem to sound more 'lively' than LPSs. I don't have any way of measuring this stuff really, but could it be the case that the harder the CPU is working, the more current it draws from the power supply, which itself injects more noise into the mains and perhaps air too, which causes a change in sound? And perhaps there's more noise in the computer box itself that gets transmitted along the DAC interface?

 

I think PeterSt has a means of measuring all this. And if I understand him correctly (rarely the case ;-)) he doesn't seem to think that any noise emanating from the PC has any affect whatsoever on the performance of the NOS1 DAC. (It's not that there is no noise, but just that the NOS1 'takes care of it'.)

 

But I agree, it would be great if PeterSt, or anyone else with the means, could actually prove (or not) a correlation between CPU power/load and noise once and for all.

 

Mani.

 

[wappinghigh]

Who knows..Miska's efforts might even outperform a Linn Klimax DS...

 

One thing's for sure. It will be wayyyyyy cheaper

 

[manisandher]

"... if I understand [PeterSt] correctly (rarely the case ;-)) he doesn't seem to think that any noise emanating from the PC has any affect whatsoever on the performance of the NOS1 DAC. (It's not that there is no noise, but just that the NOS1 'takes care of it'.)"

 

And if this indeed is the case, why the hell am I messing around with Atom computers and expensive linear power supplies? This afternoon, I got an old 500VA isolation transformer out of the basement. I bought it years ago to power some front-end components but just didn't like how it dullened the sound. But right now I'm using it to power an i7 PC. According to my 'noise sniffer', it has totally stopped any noise being injected from the computer's SMPS back to the 'clean' AC mains.

 

I'm happy, if a little embarrassed I hadn't thought of this a lot earlier.

 

Mani.

 

[ted_b]

is more tolerant of noise and general use setup? is this i7 pc you referenced an HQplayer front end pc?

 

[elcorso]

Hi Mani,

 

I have three ISOs on my music system filtering the three noisy SPSUs feeding the Mac Mini, Monitor & external CD/DVD player. They works like haven (to my ears).

 

The other gear I own, Amp, DAC & Pre I don't care since they have very good LPSUs, but could (are) be contaminated from SPSUs 'feeded' from the same AC mains.

 

I own also the 'noise sniffer', but the best test (to me) is, that something is wrong (when I'm sure I don't have any, or very low, jitter) is to get well defined sub-bass, and 'on time'.

 

I have also a separated line & breaker for the music system, and of course, very good ground.

 

"One, two, free" track from Steve Davies Project "Quality of Silence" is very good test for a noise free playback, on DSD or Redbook.

 

Roch

 

[PeterSt]

But right now I'm using it to power an i7 PC. According to my 'noise sniffer', it has totally stopped any noise being injected from the computer's SMPS back to the 'clean' AC mains.

 

I can't emphasize enough how difficult this all is. Including that everybody should obtain himself an analyser for in his audio rack.

But if a noise sniffer does the job ... great.

But super that you bought such a thing in the first place !

 

Peter

 

[falcongate]

Thx Miska. So the key is to minimize activities in the machine outputting to the audio system.

 

[manisandher]

"... are you finding that the front end pc of a NAA-architectur is more tolerant of noise and general use setup? is this i7 pc you referenced an HQplayer front end pc?"

 

I've tried the NAA being fed by two front end PCs with HQPlayer installed on them. The first is my general work PC - loads of things installed on it and multiple programs running all the time. The second is my dedicated music PC (totally silent i7). Both of these sound identical when used as the front end for the NAA. Also, when I'm using my work PC, I can't hear any changes in sound from the NAA as I work away, starting/stopping all sorts of programs. I agree with Miska when he says that the front end PC just doesn't matter... as long as it has enough grunt to run HQPlayer and whichever of its DSPs you happen to be using, some of which are quite CPU-intensive.

 

Mani.

 

[audiozorro]

"I am primarily proposing a system where the DAC (PCI soundcard, USB DAC etc) is NOT connected to the "primary" computer. The DAC gets connected to a small low power computer running a stripped down very simple linux OS, probably without any hard drives at all. The audio data comes over a normal network connection. By doing this most of the issues about how different players and disk types effect sound go out the window, they are unimportant. This gives you the flexibility to use whatever player software you want to use, you now longer have to choose between one that sounds good and one that has a user interface you like!!!!"

 

With this 3 box solution:

- different software players in the music server don't matter

- different OSes on the music server don't matter

- different music servers don't matter

- different hard drives in the music server don't matter

- noise and other pollution in the music server don't matter

- fans in the music server don't matter

- all that matters is small low power computer running a stripped down very simple OS

 

 

Is this a correct interpretation of Swenson's approach?

 

 

[manisandher]

""One, two, free" track from Steve Davies Project "Quality of Silence" is very good test for a noise free playback, on DSD or Redbook."

 

Yes, I've ripped this album to .dff and have been listening to it a lot through the NAA and Mytek. There's no question that this is a seriously well-recorded album, though I suspect not to everyone's musical tastes.

 

But I have a problem. The sound I'm getting through the NAA (with PCM or DSD) is almost 'too clean' and dynamic! I find it really hard to describe in any other way. It's not 'bright', or 'forward' or 'edgy'. Rather it's so clean and dynamic that it's almost tiring to listen to. Not relaxing at all. But it certainly keeps your attention at least.

 

Now of course, the Mytek is very new and needs a lot of burning-in by all accounts. Fine, I get this. However, I don't think the Mytek is the culprit here. I've tried the Mytek connected directly to my CAPSv2, and via both firewire and USB2.0, it sounds waaaay smoother and homogenous (and incidentally duller and less involving) than through the NAA.

 

If it's not the Mytek, it must be the NAA. The obvious difference between playing music through the CAPv2 and the NAA is that the NAA is streaming music through the LAN. So I tried streaming music from my NAS drive to the CAPSv2 to see if the LAN is introducing some sort of noise that might be giving the impression of more 'kick' to the sound. But alas, apart from just hardening of the sound a tad, it made little difference.

 

My worry now is that what I'm getting from the NAA are false dynamics. I've experienced this before when I messed around with grounding arrangements on various components and the result is eerily similar. Hopefully something just needs bedding in because I totally love the sound I'm getting, but just wish it were a bit sweeter and more laid back.

 

Mani.

 

[1audio]

In broad terms using a separate computer for processing from the one that passes the audio to the system will isolate noise in the processing system. However if its noisy enough (EMI, RFI, Conducted noise) it cans till degrade things. If its in another room well away from things it really should not have any affect. Some claim that a NAS won't be as good as a local drive and others who have issues with many aspects of digital. There are those on this forum who can hear if a file was transferred over the internet they say, so don't expect miracles.

 

However I find the NAA approach works very well. For files that are played without any changes its pretty much indistinguishable from playing the same file from a NAS with mpd on the same box as NAA. Add in the tricks Signalyst can do and its very interesting. However some will find the UI completely unacceptable. If that is your priority make sure you try the trial version.

 

[manisandher]

"Is this a correct interpretation of Swenson's approach?"

 

Yes, that's pretty much my understanding. And I agree for a couple of reasons:

 

1) extra noise generated by all the things you list would not pass down from the 'primary' computer through meters of network cable, a network switch, and more meters of network cable to the 'small low power' computer

 

2) of course, other sources of noise could easily enter the 'low power computer', causing jitter - but remember, the 'small low power" computer acts as a FIFO buffer, itself controlled by the clock in the DAC

 

My only concern is that a certain amount of noise will still be present in the firewire interface between the NAA and the DAC. But I can't for the life of me see that this could ever be more than would be present in a firewire or USB interface between the 'primary' computer and the DAC.

 

Mani.

 

[audiozorro]

An approach that works very well is not the same as the best approach.

 

I happen to be quite pleased with the UI of HQPlayer. I am also happy with my Auraliti PK90 USB. Unfortunately the two don’t mix (at least to date).

 

Are you suggesting that a 3 box approach may be sonically superior to your yet to be released L1000?

 

Or that incorporating HQPlayer into your unit would be sonically superior to the existing player?

 

If I were cynical, I would think that you are trying to sell me another two boxes, just kidding of course but I am interested in hearing from the hardy and better funded adventurers down this 3 box path.