Native DSD files, HQPlayer & Mytek Stereo192-DSD

[PeterSt]

So (hoping that I don't sound rigid now :-) why does changing from the NAS to the internal disc change sound so much you think ?

 

(all in the context of me always failing on this stuff)

 

[Miska]

So (hoping that I don't sound rigid now :-) why does changing from the NAS to the internal disc change sound so much you think ?

 

Noise leakage through the switch? Although I find it unlikely. More likely would be noise through mains. Or powerline ripple inside the NAA computer. One option is to put a scope on main board powerlines to see how clean those are. In any case with NAA it is easy to put emphasis on optimizing that piece of hardware since it doesn't have run all kinds of stuff. One can design a small optimized-for-audio ARM-based computer mainboard to run just the NAA and nothing else.

 

Sound doesn't change here, player being MacMini connected to 5.6 GHz 802.11n WLAN, playing either from a "NAS" (this desktop, Intel Core i5 running Linux&Samba) or from a local HDD connected using FW800. Playback over that same WLAN.

 

Also same sound when playing from this Linux desktop using Linux version of the player.

 

Also same sound when playing from a Core i3 laptop over 2.4 GHz 802.11n WLAN.

 

Ethernet adapters are all Intel gigabit adapters. Central switch where WLAN APs and NAA are connected is a managed gigabit Cisco (with QoS) and the Linux desktop is in addition behind an unmanaged gigabit Allied Telesyn switch.

 

Ethernet cabling is STP CAT-6 built into walls and from wall-sockets to computers UTP CAT-6. Switches are in a steel central connection closet built into a wall. (switches have stell casing and their own power feed with grounded IEC sockets)

 

Power filtering for audio is through a Siemens power filter.

 

NAA tested on three devices, this Core i5 desktop, Atom mainboard and ARM-based SoC.

 

 

[PeterSt]

Thanks for all that (hey, you keep on extending that post :-).

 

Wouldn't avoiding any switches imply a more "lean" (for noise) solution ? I mean, for people to try ? So, right from the Ethernet socket on the mobo to the NAA ?

(make a special Ethernet cable of course -> the one going from the (A)DSL socket to the router can be used I think)

 

Of course, there's still no NAS around then, but at least it eliminates some doubtful sources. So, just to try / start with ...

 

 

 

[elcorso]

Hi Ted,

 

From the booklet.

 

Mine is PTC 5186 067 (Beethoven • Piano Sonatas Nos. 1,2 & 3)

AD Converter: EMM Labs mk III, 8 channel DSD converter

 

Yours PTC 5186 024

 

I'm sorry I gave you a wrong link to Amazon.

 

But, who knows? Maybe you remember my issue with Zen Mastering (Ref. Recordings SACD) who corrected me, even if the booklet tells another history.

 

Roch

 

PS/ Mary Kodama on PentaTone Classics:

http://www.pentatonemusic.com/bios/kodama_d.pdf

 

EDITED, added

 

This is the correct link to Amazon.UK

 

http://www.amazon.co.uk/Beethoven-Piano-Sonatas-Nos-Hybrid/dp/B001HBX936/ref=sr_1_fkmr0_1?s=music&ie=UTF8&qid=1328283857&sr=1-1-fkmr0

 

This is the correct link to Amazon.US

 

http://www.amazon.com/Beethoven-Piano-Sonatas-Op-Nos/dp/B001HBX936/ref=pd_sim_m_5

 

[Miska]

Wouldn't avoiding any switches imply a more "lean" (for noise) solution ? I mean, for people to try ? So, right from the Ethernet socket on the mobo to the NAA ?

 

Sure, that's one possibility. Although then it needs either DHCP server running on the PC or manual IP address configuration on the NAA.

 

Of course, there's still no NAS around then, but at least it eliminates some doubtful sources. So, just to try / start with ...

 

Having two ethernet interfaces is fairly trivial too. Either by adding a PCI/PCIe interface card or using a USB ethernet adapter. This way it is possible to have both NAS and NAA, both using their own ethernet interfaces on the PC.

 

But I would try to track this down on the NAA side, because that's where the difference is made. And it should be easier, since it's now much much simpler than the player computer.

 

Switch-type solution is also much better once I get the multi-room support fully in place.

 

 

[manisandher]

I finally got around to trying out the Teac DV-RA1000HD recorder I bought a few months back, but which has just sat in its box until yesterday. I'm very impressed with its DSD recording capability! Using headphones, I just can't distinguish between the original vinyl going into the Teac (using its Monitor function) and the same DSD recording being played back from its HDD through its headphone output.

 

Now I certainly don't hold vinyl up as any sort of absolute reference - I personally think it's too flawed. However, in my experience, converting the essence of vinyl into digital is fiendishly difficult, for some reason. My previous Pacific Microsonics Model Two was capable of doing this. But I still never managed to capture it's magic on HDD. Using it as a simple ADC/DAC pass-through, it was totally transparent at 24/192, preserving the essence of the vinyl. However as soon as I interfaced it to a computer and tried to capture its ADC output on the computer's HDD, everything seemed to go to pot. And I tried a bunch of computers with different software/hardware configurations. The Teac's a computer with an HDD as well of course, but I guess it's a stipped down Linux implementation designed just for the job at hand.

 

What I haven't yet tried is comparing the recorded dff files being played back on the HQPlayer/NAA/Mytek to the original vinyl. I think the level of difference between the two should be a good indicator of the tranparency of the former.

 

Mani.

[ted_b]

I just wanted to bump this thread with addtl info about Miska's NAA (network audio adapter) and the folks at Auraliti.

 

As i posted elsewhere, Ray and Demain at Auraliti have been extremely helpful getting me the latest Linux MPD code to do native DSD with Jurgen's latest code (full DSF tagging, etc). Since it has solidified and stabilized (read: 24/7 rock solid with no hiccups and no player issues whatsoever) I thought I would try another nice feature of the Auraliti PK90USB, and that is the included Signalyst NAA setup. This would "remove" MPD as the music server and simply use the PK90USb as a FIFO buffer between the Windows-based Signalyst HQPlayer and my DSD-enabled DAC (Meitner MA-1 in this example; I'll confirm the Sonore/exD too. I don't have firewire for linux so my Mytek is not a candidate here). And I was assured MPD could be reinstituted easily if I decide (summary: simply stop one and start the other...couldn't be more easy).

 

Sounds complicated, huh? Nope. Just install a trial version of HQplayer on Windows (on the same home LAN as my PK90USB), set it for DoP, point it to my NAS library(s) and voila. I am using my ipad's LogMein VNC app to control HQPlayer from my listening position, and will report back on sonic differences vs MPD (which uses mPad as GUI). In this architecture, things like VNC apps on Windows (or even email browsing) is not a sonic issue cuz the NAA buffers all that. OK, MPD is clearly much more friendly right now, with its plethora of GUI and album art features, but sonics is number one, so I thought I'd try the interestingly-simple NAA architecture too. Good job Miska (and Auraliti).

[manisandher]

Ted, I'm very interested in the SQ differences between these two setups. Looking forward to hearing what you have to say...

 

Mani.

[ted_b]

Ted, I'm very interested in the SQ differences between these two setups. Looking forward to hearing what you have to say...

 

Mani.

 

First impressions were interesting. At first I thought that maybe the HQPlayer/NAA setup (aka NAA) produced slightly higher gain, but both Miska and later measurements proved that both MPD and NAA (running no upsampling,etc) are bit-perfect and identical output volume levels. But.....the NAA setup sounds more immediate. On Mari Kodama's wonderful DSD recorded Beethoven piano sonata cycle (Pentatone DSF files) the piano sounds more like it is in the room than on MPD. Weird. On BlueCoast DSD tracks the timbre is ever-so-slightly richer with NAA. Weird again. All we can think of is that the Auraliti has much less work to do in NAA mode (i.e no storage access, etc) or maybe the HQplayer just has a different tonality, just like how all the MAC players differ in sound on a bit-perfect Mac Mini setup. Any ideas?

 

(Note: No real serious PCM a/b listening done yet. A short listen to Charlie Haden's incredible Heartplay 24/96 recordings reveal a slight bass bloominess in MPD, or bass-light presentation in NAA...very prelim and done too late to really know)

[manisandher]

Any ideas?

 

No, not really. I think we'll prove the Standard Model before we fully understand computer audio :-)

 

When you have a bit of time, try this: Copy a track from your NAS to the HDD/SSD of the Windows machine you're using with HQPlayer installed. Play the track from the NAS and then from the HDD/SSD. If your experience is anything like mine, there will be a difference in SQ. Strangely, I prefer the sound from the NAS - it's more direct and immediate than from my SSD. There simply shouldn't be a difference, but there is to my ears (and anyone else's who's heard it).

 

Mani.

[ted_b]

No, not really. I think we'll prove the Standard Model before we fully understand computer audio :-)

 

When you have a bit of time, try this: Copy a track from your NAS to the HDD/SSD of the Windows machine you're using with HQPlayer installed. Play the track from the NAS and then from the HDD/SSD. If your experience is anything like mine, there will be a difference in SQ. Strangely, I prefer the sound from the NAS - it's more direct and immediate than from my SSD. There simply shouldn't be a difference, but there is to my ears (and anyone else's who's heard it).

 

Mani.

 

Will do. My NAS (Synology) is the file source for both setups (MPD and HQplayer) but you likely know that.

[Jud]

First impressions were interesting. At first I thought that maybe the HQPlayer/NAA setup (aka NAA) produced slightly higher gain, but both Miska and later measurements proved that both MPD and NAA (running no upsampling,etc) are bit-perfect and identical output volume levels. But.....the NAA setup sounds more immediate. On Mari Kodama's wonderful DSD recorded Beethoven piano sonata cycle (Pentatone DSF files) the piano sounds more like it is in the room than on MPD. Weird. On BlueCoast DSD tracks the timbre is ever-so-slightly richer with NAA. Weird again. All we can think of is that the Auraliti has much less work to do in NAA mode (i.e no storage access, etc) or maybe the HQplayer just has a different tonality, just like how all the MAC players differ in sound on a bit-perfect Mac Mini setup. Any ideas?

 

(Note: No real serious PCM a/b listening done yet. A short listen to Charlie Haden's incredible Heartplay 24/96 recordings reveal a slight bass bloominess in MPD, or bass-light presentation in NAA...very prelim and done too late to really know)

 

Count me as someone else interested in reading more about this comparison as you learn more.

 

As for ideas, are you able to play with the HQPlayer settings re filters, etc.? It would seem to make sense in my layperson's concept of how this all works that HQPlayer would have a different tonality if it is using different filters than are used in the MPD setup.

[ted_b]

Count me as someone else interested in reading more about this comparison as you learn more.

 

As for ideas, are you able to play with the HQPlayer settings re filters, etc.? It would seem to make sense in my layperson's concept of how this all works that HQPlayer would have a different tonality if it is using different filters than are used in the MPD setup.

 

Jud, as I said in my post, I am running no upsampling, etc (meaning no filters, nothing). i wanted to get a feel for bit-perfect identical playback, without any more variables.

[Jud]

Deleted.

[Jud]

First impressions were interesting. At first I thought that maybe the HQPlayer/NAA setup (aka NAA) produced slightly higher gain, but both Miska and later measurements proved that both MPD and NAA (running no upsampling,etc) are bit-perfect and identical output volume levels. But.....the NAA setup sounds more immediate. On Mari Kodama's wonderful DSD recorded Beethoven piano sonata cycle (Pentatone DSF files) the piano sounds more like it is in the room than on MPD. Weird. On BlueCoast DSD tracks the timbre is ever-so-slightly richer with NAA. Weird again. All we can think of is that the Auraliti has much less work to do in NAA mode (i.e no storage access, etc) or maybe the HQplayer just has a different tonality, just like how all the MAC players differ in sound on a bit-perfect Mac Mini setup. Any ideas?

 

(Note: No real serious PCM a/b listening done yet. A short listen to Charlie Haden's incredible Heartplay 24/96 recordings reveal a slight bass bloominess in MPD, or bass-light presentation in NAA...very prelim and done too late to really know)

 

Had a chance to roughly compare mpd on FreeBSD and HQPlayer on Linux on my desktop system this evening. (Would love to try installing the Linux version of HQPlayer on FreeBSD. Might take a flyer in the next day or so and see if it works, unless you tell me it's a waste of time, Miska.) My desktop audio equipment and listening environment are both far more pedestrian than the system you've been listening to, Ted. There's a general purpose (not audio-dedicated) computer under a large glass desk with a Dragonfly USB DAC plugged into it, running into an Audioquest N22 and from there to Wharfedale Diamond 7 speakers on the back corners of the desk, facing me a touch more than a meter away at a little less than 45 degree angles.

 

For some reason tonight I had no trouble at all getting HQPlayer to play music for me. Pilot error previously, I'm sure.

 

I liked the sound from HQPlayer at the default settings better than what I heard from mpd. With no filter and just the default TPDF dither setting, though, I'd have to say I liked mpd slightly more. I did not expect the extent to which turning off all HQPlayer filtering deadened Alison Krauss's voice (test track in all cases was Lie Awake from a CD rip of Paper Airplane). Also, while with Audirvana Plus in the "big" system with the BiFrost I prefer upsampling (though I have not listened without upsampling since getting a new amp), with HQPlayer and the Dragonfly I preferred staying at 44.1kHz resolution (default poly-sinc filtering in both cases, default TPDF dither with 44.1, NS4 dither with 88.2).

 

What was an absolute m-f-ing revelation to me with HQPlayer was the degree to which the changes in filtering changed the nature of the sound (in line with the descriptions in the HQPlayer manual). Anyone who thinks it's all just digits and that all digital, all DACs, or all software players must sound the same has got to do themselves a favor and listen to this. Yes, I know it's obvious the filter is what reconstructs analog out of the digital signal, but I lacked a gut level appreciation of exactly what that means. (I imagine the same realization is available from any other player that allows a choice of filters.)

[pacwin]

DSD 128fs, Mytek DSD DAC (USB ASIO Driver/Windowsx86/Firmware 1.6.0) and HQPlayer Desktop 2.8.2.

 

First off DSD128 files are not that readily available. Here's a couple of sources. The Japanese One Bit Audio consortium (different than the Sony Phillips alliance) have a website in Japanese but any one with a bit of imagination can find the sample files to download.

 

1bit Audio Consortium Web Page

 

Lets not overload their servers so they shut it down. Also these are wsd ("Wideband Single-bit Data) files which is a Korg format as well. The are the same length as dsf files and might be interchangeable just by changing the filename extension.Don't know didnt try. HQDesktop doesnt know how to open them, but you can easily convert them to dsf files using Audiogate. Some of the samples on offer are DSD256fs and quite a few are 8 channel and the odd one is 3.0 Mhz sample rate. rather than 64fs. The files are classically oriented (harpsichord, violin, guitar, drums, traditional string instruments etc).

 

Note one of the short samples 11.wsd is of some kind of percussion bells etc and will really test your tweeters with the higher order harmonics so not too loud and also it will trigger clipping so recommend you be careful (or use Audiogate to ratchet back the peaks). The samples are fine for testing but they are just that, test samples not fully fleshed out works of art.

 

Additionally if you have a spare $60US you can get a short sample of some 128fs twin microphone Beethoven recorded by Noriaki Kitamura using a portable Korg recorder with the Slovakia Philharmonic. Noriaki Kitamura | Beethoven Symphony No. 7 in A major op. 92 & Strauss "Emperor Waltz" op. 437 | CD Baby Music Store (you also get a full Stereo Hybrid SACD) .

 

You can also find this and another disc of Beethoven Sym 7 at Amazon Japan but its only 25 minutes total of DSD128. There are SACD's and also 176.4 PCM files of the whole performance on the accompanying ROMS.

 

Noriaki Kitamura | Beethoven Symphony No. 7 in A major op. 92 & Strauss "Emperor Waltz" op. 437 | CD Baby Music Store

 

Having got your samples how to setup HQDesktop? . In the settings panel for HQDesktop if you set the bit rate limit to 5644800 this will actually degrade the signal, attenuate the volume, and make you wonder why you bother. It will also generate a large number of FIFO errors in the USB Pal Control Panel under status info. These do not disappear when the bit rate limit is changed but thay are far fewer (Note the sample rate FWIW is running at 88.2kHz) . Still the Mytek display will tell you its playing DSD128

 

 

So set the bit rate limit to 2822400

 

 

Here's the USBPAL Control Panel showing the FIFO errors (Roughly one every 2-6 seconds when playing "correctly"). Not sure of the sonic significance?

 

 

All going well, once you initiate (if you can figure out the "transport" and "playlist" idiosyncrasies of HQ Player Desktop) playing the files HQ Player Desktop, the files sound as expected. Sonically it sounds much improved and all the numbers in HQ Player Desktop as well as the Mytek display indicate that 128FS is playing as well. Presumably there are some bugs in HQ Player Desktop.

 

[Miska]

Having got your samples how to setup HQDesktop? . In the settings panel for HQDesktop if you set the bit rate limit to 5644800 this will actually degrade the signal, attenuate the volume, and make you wonder why you bother. It will also generate a large number of FIFO errors in the USB Pal Control Panel under status info. These do not disappear when the bit rate limit is changed but thay are far fewer (Note the sample rate FWIW is running at 88.2kHz) . Still the Mytek display will tell you its playing DSD128

 

I'd guess that's the Mytek firmware bug related to volume bypass mentioned in the Mytek thread in context of Foobar. I'm not seeing that kind of behavior here (I'm not using bypass mode and I have different driver and firmware versions).

 

That setting is only used to pick up default rate when switching to SDM mode (this is also immediately set to the Mytek).

 

In case you didn't have DSD content on the playlist when the mode was switched (at startup etc), you'll first get DSD64 rate set to Mytek based on the setting and then later DSD128 rate set to match the content...

[manisandher]

What was an absolute m-f-ing revelation to me with HQPlayer was the degree to which the changes in filtering changed the nature of the sound (in line with the descriptions in the HQPlayer manual). Anyone who thinks it's all just digits and that all digital, all DACs, or all software players must sound the same has got to do themselves a favor and listen to this. Yes, I know it's obvious the filter is what reconstructs analog out of the digital signal, but I lacked a gut level appreciation of exactly what that means. (I imagine the same realization is available from any other player that allows a choice of filters.)

 

Yes, HQPlayer is an excellent resource to learn how different digital filters and noise shaping schemes sound. In much the same way, I find XXHighEnd an excellent resource to learn how different computer settings sound - e.g. different memory management schemes. You wanna know how deep the computer audio rabbit hole is? Start playing around with different dials in XXHighEnd... and you'll be amazed that even with bit-perfect output, everything makes a difference! Yeah, yeah, it shouldn't... but it does.

 

 

Mani.

[manisandher]

Miska, following on from pacwin's post above, is it now possible to connect the Mytek to the NAA via USB and play DSD128 files? As mentioned by pacwin, DSD128 files are few and far between and although I will at some point start digitizing my vinyl, I'll be using my Tascam recorder at DSD64. So rather moot for me, but I'd like to know nevertheless.

 

Cheers, Mani.

[Miska]

Miska, following on from pacwin's post above, is it now possible to connect the Mytek to the NAA via USB and play DSD128 files?

 

Not yet, since DSD128 currently requires using the USB2 connection and ASIO drivers on Windows. So it works with HQPlayer when the player is running on Windows and the DAC is connected to the same machine. I say yet, because it would be technically possible to make it work with DoP 1.1 channel bonding way over Firewire too. I have experimental support for it in the player, but the needed firmware support is missing.

 

Another option is to use SDIF-3 inputs of the mastering version, but that requires special hardware to work.

[davewantsmoore]

Thanks for your posts everyone. Just letting you know, that there are others out here reading and following your discussion with interest.

 

I should have some time very soon to begin my experiments with HQPlayer/NAA, and perhaps XX.

 

I do not have DSD play, so only PCM for now.

[manisandher]

 

Wouldn't avoiding any switches imply a more "lean" (for noise) solution ? I mean, for people to try ? So, right from the Ethernet socket on the mobo to the NAA ?

 

Sure, that's one possibility. Although then it needs either DHCP server running on the PC or manual IP address configuration on the NAA.

 

Miska, I'd like to make a detailed comparison between my dedicated music PC (running in a near-dead, minimized OS state) feeding the Mytek DAC directly vs. via the NAA. I'd like to eliminate any affect a network switch might be having by simply eliminating it altogether (all music stored on large HDD in PC). Would it be possible to set the IP address of the PC on the NAA manually? If so, how would I go about doing this? (If it's better to discuss via PM, let's do that.)

 

Cheers, Mani.

[jrling]

That's what I would like to do too. Any luck?

Jonathan

 

 

Miska, I'd like to make a detailed comparison between my dedicated music PC (running in a near-dead, minimized OS state) feeding the Mytek DAC directly vs. via the NAA. I'd like to eliminate any affect a network switch might be having by simply eliminating it altogether (all music stored on large HDD in PC). Would it be possible to set the IP address of the PC on the NAA manually? If so, how would I go about doing this? (If it's better to discuss via PM, let's do that.)

 

Cheers, Mani.

[1audio]

Depending on your understanding and skill with networking its not difficult to tie two boxes together with a crossover cable (or autoidx which your PC should have which figures out the connections of the cable automatically) and set both to fixed ip addresses (return to the origins of networking). You will need to set fixed ip addresses on each box (each OS handles this differently) and use the ip address of the other box to find it. As I remember the ip address of the NAA is broadcast to all of the network so it will be found automatically in this setup.

 

If you want a short and possibly irreversable trip to insanity, compare ipV4 sound to ipV6. If you hear a difference please don't share, life is too short. . .

[jrling]

Thanks. My networking knowledge is most certainly not up to that! But thanks for the encouragement.

IPv4 v IPv6 will remain unvisited.

 

 

Depending on your understanding and skill with networking its not difficult to tie two boxes together with a crossover cable (or autoidx which your PC should have which figures out the connections of the cable automatically) and set both to fixed ip addresses (return to the origins of networking). You will need to set fixed ip addresses on each box (each OS handles this differently) and use the ip address of the other box to find it. As I remember the ip address of the NAA is broadcast to all of the network so it will be found automatically in this setup.

 

If you want a short and possibly irreversable trip to insanity, compare ipV4 sound to ipV6. If you hear a difference please don't share, life is too short. . .