First multi-channel DSD playback solution with Mytek!

[Miska]

Third, yes native DSD64 cannot be DSP'd so it is very advantageous that I have an almost perfect room acoustically (thanks Jeff Hedback, Ethan Winer, etc), have my identical transducers spaced in a ITU setup (i.e i don't need bass management or channel delays, just trims done in the analog world, and LFE DSP'd for the room via SMS-1). Maybe Miska or others will chime in on what DSD128 could do to remedy this non-DSP issue for others.

 

Technically there's no problem of doing speaker distance and level adjustments (or anything else like digital room correction) DSP for DSD64 or DSD128. I don't know how well CAPS hardware could handle that, maybe someone could try out. I have only one Mytek DAC so I cannot try this stuff out. But I know for sure, that unless you have really simple DRC filter, CAPS won't cope with multi-channel digital room correction for native DSD, you better have some 8-core... My Mac Mini can run DRC for stereo DSD64 though...

 

So far, I support doing these either for 1:1 rate or 1:2 rate, since going from DSD64 to DSD128 while applying digital volume makes some extra dynamic range while applying level adjustment... But doing DSD64 to DSD64 won't suffer too much either.

 

Multichannel DSD is really cool stuff!

[Miska]

This is the first time I heared of this. How does this work without PMC conversion?

 

I won't tell all the details... But why wouldn't it? With computers, we don't have to consider any limitations in processing power, do we? Of course we'll optimize it to run as fast as possible, but Moore's law will take care of the rest.

[Miska]

To my knowledge there are no DSP engines that run in native DSD and offer bass management and EQ. Digital volume control in DSD, yes, at best delay. That's it.

 

Maybe it's time to update that knowledge then...

 

Bass management and cross-overs are on the TODO-list though. Processing is piece of cake, having a sensible simple GUI for configuring such complex systems is bigger challenge.

[Miska]

Please enlighten me.

 

I have my own DSP engine capable of all that.

 

Sounds like my knowledge is still up to date then.

 

Bass management not in the GUI, because I don't consider it important. 5.1 channel is supposed to have sub and most classical material is 5.0 channel and better to be played through full-range speakers without extra cross-overs. Bass management is for cheap home theaters and screws up sound.

 

If you consider digital room correction EQ with phase adjustments etc simpler than bass management then that's fine, I don't mind.

 

My point was that I have practically all the processing that I have available for PCM also available for DSD, that is rate conversion up or down, delay control, volume control and convolution engine (so you can run any kind filter). And conversion PCM <-> DSD. So from my perspective PCM and DSD are equal.

 

So if/once I have bass management it will most likely be available for both. Now it's not available for PCM either.

[Miska]

So it appears my assessment that it is not currently "available (i.e. to the general public)" is correct.

 

I don't know about bass management. I don't offer it, maybe in future if I bother. But a full convolution engine for running digital room correction or what ever similar EQ is there, available immediately.

 

I use it to smooth out bass response in my listening room, equally for PCM and DSD content.

 

What do you mean? Bass management for PCM is available everywhere - an a PC, Mac, in univeral players and processors.

 

I don't offer it in my software, and I don't have specific plans about it either. But, OTOH, I'm not planning to be "universal" because that's already a word that makes me feel sick. (usually means "universally bad on everything")

[Miska]

^ guys. It doesn't matter. Play pure DSD out via analog into pure 5.1 analog inputs into the receiver. And "manage" everything like ted says with your own ears on the analog side. Been doing that for years with my old trusty NAD SACD player. That's all you need to do guys. Cheers. WAP.

 

Wap, I've already long time ago figured out that only way to get what I want is to do it myself.

 

I don't have to wait and beg for someone to do it, nor agree with anybody... Maybe it takes insane amount of time and effort, but eventually I'll get exactly what I wanted.

 

One thing I wanted was to run my DRC filter with DSD files too, so only option was to just implement the damn thing. Same goes also for speaker level/distance adjustment for multichannel.

 

One thing I love the most though, is when someone says something cannot be done...

[Miska]

I think I figured out a way that you could do this with different DACs in the system. So you could use the Meitner DAC for two channels and two Myteks for the remaining channels or some other combination. I have to download a test file, update some apps and dig out some parts from the boneyard...I'll keep you posted.

 

Jesus, Linux has features for doing this stuff, but there are couple of tricky parts:

- You need to have sample clocks synchronized, so one DAC needs to be master and other slave to it's clock, otherwise DACs will eventually drift out of sync. This can be done with word clock output and input.

- More challenging is starting all the DACs sample-accurately at the same time to avoid phase error between channels. This needs a "broadcast launch-control", which I guess can be done with proprietary protocols like the RigiSys' implementation in Mytek.

 

Usually on Windows ASIO driver takes care of this, on Linux ALSA can take care of the same (all the needed stuff is already there).

 

In the old days, over ten years ago, I used to have an M-Audio Delta1010 interface (still in production), and it's ASIO driver has supported bonding four of those interfaces together to form a 32-channel in/out system.

[Miska]

My test today confirmed that playback via J-River and the Lynx card is resulting in a multi channel DSD to PCM conversion at 176.4.

 

BTW I think the USB solution is very cool and I was just trying to see if I could figure out a way to use your Meitner and two Mytek...

 

I've been playing multichannel DSD with DoP on Linux and Windows (since I added DoP support), as well as DSD->PCM converted multichannel to various different 8-channel pro-interfaces. Any bit-perfect multichannel AES card should work fine with stack of Mytek's.

 

Only new thing to me in this set up has been linking of multiple USB devices.

[Miska]

The USB ASIO driver does not care whether its a preamp or mastering version; same USB, same outputs. But...what I documented above was that the USB driver will give the fronts (left and right) to the DAC with the oldest serial number, and then the center/LFE channels to the next newest DAC (and if a third DAC, it would be the newest s/n and get surround left/right).

 

I'm just curious, do you use word-clock or how the sample clocks are synchronized between units? IOW, which unit has the master clock?

[Miska]

Miska, it's handled by the ASIO driver (USBMLink) and likely DAC1.

 

I repeat what I first started this thread off with: three DACs, a USB hub, a knowledge of the serial numbers in order to cable the channels correctly, and a pre-beta release of J River (was 130, now using 138)...PERIOD!

 

Doesn't sound like asynchronous USB then... I mean somehow they must use the same sample clock, otherwise the DACs will eventually drift out-of-phase. IOW, won't be in-sync at sample level. USB is not exactly good vehicle for clock distribution.

[Miska]

By the way, if you bought a new Mytek, your current one would become the fronts in this example, of course.

 

Unless the player allows free channel assignments, like I do for ASIO outputs. So any of the source channels can be assigned to any output channel. This kind of functionality is typically needed in studio setups for audio interfaces with more than eight output channels.

[Miska]

Ted, have you looked into .dsf as a multi-channel format? It's specified as stereo only, and I have not tried to convert a 5 channel .dff file with Audiogate to see what it does with it. It's on a machine currently taken apart.

 

No it's not stereo-only, DSF supports mono, stereo, 3.0, quad, 3.1, 5.0 and 5.1... DFF can support the same, but also in addition 2.1.

 

So both DSD file formats are perfectly ready for any pretty much any channel configuration up to 5.1. (so all the same as SACD)

[Miska]

You know that DSD alone doesn't support 5.1? SACD uses lossless compression called DST to achieve that.

 

Don't dare to tell that to my software and uncompressed 5.0 and 5.1 channel DSD files! I'm afraid it could stop working!

[Miska]

Could you use three inexpensive DSD DAC’s for multi-channel DSD playback?

 

Something like three Schiit Loki’s, or three iFi nano iDSD’s?

 

No... Reason is that it is very difficult to get multiple devices started precisely at the same time within one sample's time, so that there wouldn't be phase differences between those devices. And if you can start those at the same time, over time phase difference will develop because you'll never have three clocks running precisely at the same speed.

 

Pro-gear has word-clock inputs for synchronizing the clocks. Interfaces like AES and S/PDIF are better than USB for this purpose because master clock is at the sender side so there's no clock synchronization issue and you can use interfaces like the multi-AES cards from RME or Lynx to output synchronized multichannel. Asynchronous USB goes precisely against you in this kind of setup because master clock is at the DAC side, and there cannot be three master clocks for one content.

 

At technical level, Linux audio system supports linking multiple audio interfaces together and attempting to start those simultaneously.

[Miska]

I believe one could compensate on Acourate for the different starting times Miska refers to.

 

Problem is that you don't know how much there is device-to-device start delay, it will be different on every start...

[Miska]

None of this would be DSD of course. Acourate is PCM

 

But the filters created by Acourate can be used on DSD with other software...

 

Plus of course channel delays/levels are available without Acourate or such.

[Miska]

I believe that would be solved thru measuring the response from each speaker and adjusting inside Acourate. Similar to the above, I understand for what I'm after that can be done, and was extrapolating to a full-range multichannel situation. But maybe it doesn't work there.

 

You would need to re-measure it every time you hit "Play", because you get different delay every time... Plus the relative delay drifts over time during playback too.

 

So no, Acourate doesn't help synchronizing multiple DACs.