Whats the point of DSD

[craigl59]

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

Actually, it's the SACD license that doesn't allow SACD players to export a full quality digital audio signal from an SACD over a digital output that isn't "secure" (HDMI is secure).
However, that requirement is part of the SACD copy protection scheme, and goes with the SACD media in specific..... DSD audio that originates from something else, for example a file, doesn't have that restriction.
(I have no idea how the individual players may feel about it - or whether they make the distinction.)

The discussion has already been done to death, but DSD itself has no compelling technical advantages over PCM.
DSDx1 is superior in some technical details over the 16/44k PCM used on CDs - but not over 24/96k PCM.
In general, as a RECORDING and REPRODUCTION medium, DSD is about equal to PCM, and it's about equally difficult to convert into analog when you want to play it.
However, since it has severe limitations in regard to editing, it has never made it as an editing standard, and has been a commercial failure as a distribution medium as well.
(Sony has made the DSD format license available to all comers for free - we can assume they did this because it wasn't making them any money.)

However, the hardware in a given DAC may be optimized to do a better job with one or the other.
So, for that particular DAC, it may make sense to convert all incoming signals to the format it prefers.
(It should be possible to design a DAC that will work equally well with both - but that doesn't mean that all of them are.)

You also need to understand that ANY time you convert between sample rates or formats you introduce tiny differences to the signal.
Furthermore, there are usually choices involved, which involve tradeoffs, any of which may or may not be slightly audible.
However, in most cases, these changes aren't better or worse - just different.
(And, since certain tradeoffs may suit your personal preference better, or work better with certain source material, choices can be useful.)

Another factor is that SACDs are mastered with a different target audience in mind.
People who buy SACDs and DSD files expect "a more audiophile sound" - whatever that means to them and the engineer who mastered the disc.
There is also different technology involved - SACD mastering offers less flexibility and fewer options.
All of these factors lead to the fact that it isn't unusual at all for the SACD version of an album to be mastered differently than the PCM version.

And all of this leads back to the question of what exactly you mean by "quality".

To answer that original question......

Since information cannot be created, if by "quality" you mean "information accurate to the original", then NOTHING can give you more quality than was there in the original file.
At most, by converting to some specific format or sample rate, you may be allowing a specific DAC to do a better job of causing less damage to the information that is present.
(This is why most modern DACs do internal oversampling; because the oversampled signal can be converted more accurately with less alteration by the reconstruction filters.)
And, since each conversion may alter the sound a little bit, you may simply like the way audio sounds after being passed through a certain conversion process.
(And it's the job of the marketing department at any company to convince you that the differences in their product make it better.
And, if you're the type of person who simply likes more choices, and more options to choose from, then DSD is simply one more option to choose from.

I should also note that you need to be careful when reading white papers..... many of them are quite old and digital audio technology changes quite rapidly.
One of the linked white papers compared DSD with single-bit Delta-Sigma converters.... but virtually all modern DACs use multi-bit Delta-Sigma converters.
Likewise, it used to be common to compare SACDs to CDs (DSDx1 to 16/44k PCM); but high-resolution 24/96k PCM audio files are becoming quite common these days.

It is my understanding that the Oppo CAN receive DSD via DLNA (over Ethernet)...
And this is claimed to work with some DLNA servers (perhaps some versions of Twonky).
However, I've read that it won't work with jRiver.
(If you Google "Oppo DSD DLNA" you can find a lot of discussion and suggestions on the subject...)

You could try HERE for starters:
www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/oppo-bdp-10%5B3-5%5D-d-and-multichannel-direct-stream-digital-playback-26808/

Oct 14, 2016 7:06:47 GMT -5 geebo said:

You can't send DSD over coax or Toslink. DSD licensing doesn't allow for it. The Oppo can send DSD via HDMI, however.

Keith:

This is extraordinarily clear and has explained many questions about DSD I have had for some time.

Would suggest, however, that you amend your consideration of editing happening in the analog world. While this can happen, many of us that have been using 96/24 since 1998 or so have always edited with plugins and DAW tools that maintain that signal rate throughout the entire editing process.

Recently have had several conversations with production people about the value of upsampling. They have agreed with you that, literally, "nothing can give you more quality than there was in the original file." HOWEVER, they claim that the advantage to upsampling is in the down conversion -- that, specifically, a CD file upsampled to 192/24, then D/A converted will have better S/N and this can contribute to more transparency and detail. I have noted this myself using JRiver to upsample CDs and have compared the sound with the original source and can hear a significant improvement.

Thanks again for your clear thinking about DSDs.

Craig

[solidstate]

Oct 17, 2016 15:43:12 GMT -5 yves said:

Oct 17, 2016 14:09:15 GMT -5 solidstate said:

hmm... interesting point.

Somebody needs to design a PCIe based AES/EBU device with DoP that connects to a PC via Thunderbolt or USB 3.1 (supports two PCIe lanes) perhaps negating this issue.

Great thing is the XMC-1 has AES/EBU input!

MADI is also an interesting possible solution.

I'd like to see processor makers design a pre/pro that has MADI out and a second chassis with the converter section. That or simply pass IIS/I2S out to a box specifically to be used as a converter only. The second converter/analog section chassis would allow much more space on PCB to have proper balanced outs on all channels and a really great I/V and line driver for each channel.

There is no reason why you can't pass multiple IIS/I2S buses a short distance via an HDMI connector between these two proposed devices.

PS I heard recent testimony here that a certain DAC sounded much better via AES/EBU transport vs. USB

The problem of HDMI-CEC Audio Clock Regenreration (ACR) has been explained in this presentation:
Audio Transport over HDMI—How good do you want it to sound?
www.aes-media.org/sections/uk/Conf2011/Presentation_PDFs/14%20-%20john%20dawson%20-%20Audio%20Transport%20over%20HDMI%20-%20AES%202011.pdf

Asynchronous USB already solves a similar problem inherent of AES/EBU and S/PDIF... but IMHO asynchronous USB has been given a serious bad name by a lot of manufacturers who either can't understand how electric noise and electromagnetic interference (EMI) can propagate from the USB input interface module into the more sensitive parts of a DAC unit, or basically just want to skimp on the USB implementation because doing that will save them some extra money.
Is the AES/EBU/SPDIF Digital Audio Interface Flawed?
www.aes.org/e-lib/browse.cfm?elib=6773

Oh wow thanks for the insight and presentation/papers!

Definitely will be checking that stuff out.

[solidstate]

Oct 17, 2016 16:42:11 GMT -5 craigl59 said:

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

Actually, it's the SACD license that doesn't allow SACD players to export a full quality digital audio signal from an SACD over a digital output that isn't "secure" (HDMI is secure).
However, that requirement is part of the SACD copy protection scheme, and goes with the SACD media in specific..... DSD audio that originates from something else, for example a file, doesn't have that restriction.
(I have no idea how the individual players may feel about it - or whether they make the distinction.)

The discussion has already been done to death, but DSD itself has no compelling technical advantages over PCM.
DSDx1 is superior in some technical details over the 16/44k PCM used on CDs - but not over 24/96k PCM.
In general, as a RECORDING and REPRODUCTION medium, DSD is about equal to PCM, and it's about equally difficult to convert into analog when you want to play it.
However, since it has severe limitations in regard to editing, it has never made it as an editing standard, and has been a commercial failure as a distribution medium as well.
(Sony has made the DSD format license available to all comers for free - we can assume they did this because it wasn't making them any money.)

However, the hardware in a given DAC may be optimized to do a better job with one or the other.
So, for that particular DAC, it may make sense to convert all incoming signals to the format it prefers.
(It should be possible to design a DAC that will work equally well with both - but that doesn't mean that all of them are.)

You also need to understand that ANY time you convert between sample rates or formats you introduce tiny differences to the signal.
Furthermore, there are usually choices involved, which involve tradeoffs, any of which may or may not be slightly audible.
However, in most cases, these changes aren't better or worse - just different.
(And, since certain tradeoffs may suit your personal preference better, or work better with certain source material, choices can be useful.)

Another factor is that SACDs are mastered with a different target audience in mind.
People who buy SACDs and DSD files expect "a more audiophile sound" - whatever that means to them and the engineer who mastered the disc.
There is also different technology involved - SACD mastering offers less flexibility and fewer options.
All of these factors lead to the fact that it isn't unusual at all for the SACD version of an album to be mastered differently than the PCM version.

And all of this leads back to the question of what exactly you mean by "quality".

To answer that original question......

Since information cannot be created, if by "quality" you mean "information accurate to the original", then NOTHING can give you more quality than was there in the original file.
At most, by converting to some specific format or sample rate, you may be allowing a specific DAC to do a better job of causing less damage to the information that is present.
(This is why most modern DACs do internal oversampling; because the oversampled signal can be converted more accurately with less alteration by the reconstruction filters.)
And, since each conversion may alter the sound a little bit, you may simply like the way audio sounds after being passed through a certain conversion process.
(And it's the job of the marketing department at any company to convince you that the differences in their product make it better.
And, if you're the type of person who simply likes more choices, and more options to choose from, then DSD is simply one more option to choose from.

I should also note that you need to be careful when reading white papers..... many of them are quite old and digital audio technology changes quite rapidly.
One of the linked white papers compared DSD with single-bit Delta-Sigma converters.... but virtually all modern DACs use multi-bit Delta-Sigma converters.
Likewise, it used to be common to compare SACDs to CDs (DSDx1 to 16/44k PCM); but high-resolution 24/96k PCM audio files are becoming quite common these days.

It is my understanding that the Oppo CAN receive DSD via DLNA (over Ethernet)...
And this is claimed to work with some DLNA servers (perhaps some versions of Twonky).
However, I've read that it won't work with jRiver.
(If you Google "Oppo DSD DLNA" you can find a lot of discussion and suggestions on the subject...)

You could try HERE for starters:
www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/oppo-bdp-10%5B3-5%5D-d-and-multichannel-direct-stream-digital-playback-26808/

Keith:

This is extraordinarily clear and has explained many questions about DSD I have had for some time.

Would suggest, however, that you amend your consideration of editing happening in the analog world. While this can happen, many of us that have been using 96/24 since 1998 or so have always edited with plugins and DAW tools that maintain that signal rate throughout the entire editing process.

Recently have had several conversations with production people about the value of upsampling. They have agreed with you that, literally, "nothing can give you more quality than there was in the original file." HOWEVER, they claim that the advantage to upsampling is in the down conversion -- that, specifically, a CD file upsampled to 192/24, then D/A converted will have better S/N and this can contribute to more transparency and detail. I have noted this myself using JRiver to upsample CDs and have compared the sound with the original source and can hear a significant improvement.

Thanks again for your clear thinking about DSDs.

Craig

Oh wow!

Thanks for sharing your engineering/production experience.

Talking about death eh...

I think it's media formats like CD that's dead especially when it hurts SQ to the end users.

Digital distro via audio files like flac etc lossless at the same bit/sample rate as the masters is clearly the best solution to get the best audio to the listeners.

FLAC 24/96 is the way to go for distro!

[Boomzilla]

Oct 17, 2016 16:42:11 GMT -5 craigl59 said:

...Recently have had several conversations with production people about the value of upsampling. They have agreed with you that, literally, "nothing can give you more quality than there was in the original file." HOWEVER, they claim that the advantage to upsampling is in the down conversion -- that, specifically, a CD file upsampled to 192/24, then D/A converted will have better S/N and this can contribute to more transparency and detail. I have noted this myself using JRiver to upsample CDs and have compared the sound with the original source and can hear a significant improvement. Thanks again for your clear thinking about DSDs. - Craig

Hi Craig -

I've just spent a day listening to upsampling at the conversion end. The DAC has the ability to upsample "on the fly" any incoming bit-stream at these (user-selectable from the remote) levels:

48K
88.2K
96K
176.4K
192K
352.8K
705.6K
DSD64
DSD128
DSD256

Curiously, my impression of all of these was "not enough difference to bother with." Yes, some had better bass, others better treble, etc. But there was NO clear winner & on average, there was no free lunch either. To get an advantage in one area, you had to sacrifice something elsewhere.

This wasn't my first impression. The initial differences (ooh - this one has more bass slam!) were eventually tempered by "but it isn't as great in the treble..."

So I'm happy that you hear "a significant improvement." I don't doubt you. But my experience is different and I just thought I'd mention it.

Cheers - Boomzilla

[craigl59]

Oct 17, 2016 16:53:18 GMT -5 Boomzilla said:

Oct 17, 2016 16:42:11 GMT -5 craigl59 said:

...Recently have had several conversations with production people about the value of upsampling. They have agreed with you that, literally, "nothing can give you more quality than there was in the original file." HOWEVER, they claim that the advantage to upsampling is in the down conversion -- that, specifically, a CD file upsampled to 192/24, then D/A converted will have better S/N and this can contribute to more transparency and detail. I have noted this myself using JRiver to upsample CDs and have compared the sound with the original source and can hear a significant improvement. Thanks again for your clear thinking about DSDs. - Craig

Hi Craig -

I've just spent a day listening to upsampling at the conversion end. The DAC has the ability to upsample "on the fly" any incoming bit-stream at these (user-selectable from the remote) levels:

48K
88.2K
96K
176.4K
192K
352.8K
705.6K
DSD64
DSD128
DSD256

Curiously, my impression of all of these was "not enough difference to bother with." Yes, some had better bass, others better treble, etc. But there was NO clear winner & on average, there was no free lunch either. To get an advantage in one area, you had to sacrifice something elsewhere.

This wasn't my first impression. The initial differences (ooh - this one has more bass slam!) were eventually tempered by "but it isn't as great in the treble..."

So I'm happy that you hear "a significant improvement." I don't doubt you. But my experience is different and I just thought I'd mention it.

Cheers - Boomzilla

Thanks, Boomzilla for the perspective.

Can you automatically change the bit rate from 16 to 24 as part of the comparison? This is where most of the advantage lies in the DAW files I have previewed.

My checks were done, specifically, by comparing Oppo feeds of CDs at 44.1/16 with JRiver ones at 192/24. Upsampling is done with their SoX option that may or may not (to my ears) add to the quality.

BTW, recently have had a chance to compare a series of piano recordings made at 192/24 with ones made at 96/24. Noone -- the player nor the recordist -- could tell any difference. You are probably aware of the specific file problems associated with 192k conversion and this is why many studios do not use it.

Also, many are still in the ADAT mode at 88k and an interesting example of how good this format can be is in Natalie Merchant's 2014 CD which is available at 88K from HDTracks.

Care to share with us your overall assessment of the Cary?

Thanks.

Craig

[craigl59]

Boomzilla:

Wanted to add that I early voted yesterday and could not get this out of my head:

"In a contest between a blithering idiot and an arrogant liar, only the voters lose." - Boomzilla

Your quotes are the best and point to in insightful mind...

Craig

[KeithL]

Upsampling absolutely can contribute to better sound... especially if you're starting out at 44 k.
(In fact, it's virtually a necessity to achieve good sound quality.)

In order to convert 44k to analog directly, you would need to include, as the last step in the conversion process, a filter that is flat to 20 kHz, yet rolls off VERY quickly above 20 kHz.
(Ideally it would be down something like 80 dB at 22 kHz - which is a VERY difficult filter to design, and is almost certain to adversely affect phase and frequency response in the audible range.)
This is absolutely necessary to remove aliasing that occurs as part of the conversion process - and which ends up as very audible distortion.
This requirement is the reason why most "non-oversampling DACs" have poor high frequency performance.

However, most modern DACs do in fact already oversample internally - usually by at least 8x when playing a 44k source.
(Oversampling is simply another name for "upsampling by an even multiple".)
This means that, when you upsample in something like jRiver, you're simply adding MORE upsampling to that already being done internally by the DAC itself.

As for editing ... virtually everyone I know edits digitally these days.

However, it seems that Cookie Marenco prefers a very... minimalist... approach that includes recording in DSD and very little editing.
Cookie seems to have a few interesting ideas about mastering and editing....
(I was responding to the fact that, as someone noted, she apparently prefers to use DSD recording and analog editing.) 
bluecoastrecords.com/about/about-ese-extended-sound-environment

And here's Mark Waldrep's view on the situation (which is more in line with mine):
www.realhd-audio.com/?p=2026

Oct 17, 2016 16:42:11 GMT -5 craigl59 said:

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

Actually, it's the SACD license that doesn't allow SACD players to export a full quality digital audio signal from an SACD over a digital output that isn't "secure" (HDMI is secure).
However, that requirement is part of the SACD copy protection scheme, and goes with the SACD media in specific..... DSD audio that originates from something else, for example a file, doesn't have that restriction.
(I have no idea how the individual players may feel about it - or whether they make the distinction.)

The discussion has already been done to death, but DSD itself has no compelling technical advantages over PCM.
DSDx1 is superior in some technical details over the 16/44k PCM used on CDs - but not over 24/96k PCM.
In general, as a RECORDING and REPRODUCTION medium, DSD is about equal to PCM, and it's about equally difficult to convert into analog when you want to play it.
However, since it has severe limitations in regard to editing, it has never made it as an editing standard, and has been a commercial failure as a distribution medium as well.
(Sony has made the DSD format license available to all comers for free - we can assume they did this because it wasn't making them any money.)

However, the hardware in a given DAC may be optimized to do a better job with one or the other.
So, for that particular DAC, it may make sense to convert all incoming signals to the format it prefers.
(It should be possible to design a DAC that will work equally well with both - but that doesn't mean that all of them are.)

You also need to understand that ANY time you convert between sample rates or formats you introduce tiny differences to the signal.
Furthermore, there are usually choices involved, which involve tradeoffs, any of which may or may not be slightly audible.
However, in most cases, these changes aren't better or worse - just different.
(And, since certain tradeoffs may suit your personal preference better, or work better with certain source material, choices can be useful.)

Another factor is that SACDs are mastered with a different target audience in mind.
People who buy SACDs and DSD files expect "a more audiophile sound" - whatever that means to them and the engineer who mastered the disc.
There is also different technology involved - SACD mastering offers less flexibility and fewer options.
All of these factors lead to the fact that it isn't unusual at all for the SACD version of an album to be mastered differently than the PCM version.

And all of this leads back to the question of what exactly you mean by "quality".

To answer that original question......

Since information cannot be created, if by "quality" you mean "information accurate to the original", then NOTHING can give you more quality than was there in the original file.
At most, by converting to some specific format or sample rate, you may be allowing a specific DAC to do a better job of causing less damage to the information that is present.
(This is why most modern DACs do internal oversampling; because the oversampled signal can be converted more accurately with less alteration by the reconstruction filters.)
And, since each conversion may alter the sound a little bit, you may simply like the way audio sounds after being passed through a certain conversion process.
(And it's the job of the marketing department at any company to convince you that the differences in their product make it better.
And, if you're the type of person who simply likes more choices, and more options to choose from, then DSD is simply one more option to choose from.

I should also note that you need to be careful when reading white papers..... many of them are quite old and digital audio technology changes quite rapidly.
One of the linked white papers compared DSD with single-bit Delta-Sigma converters.... but virtually all modern DACs use multi-bit Delta-Sigma converters.
Likewise, it used to be common to compare SACDs to CDs (DSDx1 to 16/44k PCM); but high-resolution 24/96k PCM audio files are becoming quite common these days.

It is my understanding that the Oppo CAN receive DSD via DLNA (over Ethernet)...
And this is claimed to work with some DLNA servers (perhaps some versions of Twonky).
However, I've read that it won't work with jRiver.
(If you Google "Oppo DSD DLNA" you can find a lot of discussion and suggestions on the subject...)

You could try HERE for starters:
www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/oppo-bdp-10%5B3-5%5D-d-and-multichannel-direct-stream-digital-playback-26808/

Keith:

This is extraordinarily clear and has explained many questions about DSD I have had for some time.

Would suggest, however, that you amend your consideration of editing happening in the analog world. While this can happen, many of us that have been using 96/24 since 1998 or so have always edited with plugins and DAW tools that maintain that signal rate throughout the entire editing process.

Recently have had several conversations with production people about the value of upsampling. They have agreed with you that, literally, "nothing can give you more quality than there was in the original file." HOWEVER, they claim that the advantage to upsampling is in the down conversion -- that, specifically, a CD file upsampled to 192/24, then D/A converted will have better S/N and this can contribute to more transparency and detail. I have noted this myself using JRiver to upsample CDs and have compared the sound with the original source and can hear a significant improvement.

Thanks again for your clear thinking about DSDs.

Craig

[Boomzilla]

The Cary, if I understand the manual correctly, automatically revises ALL incoming feeds to 32 bit depth.

I'll post a link when my Cary review is published on the Secrets of High Fidelity & Home Theater website.

[craigl59]

Oct 17, 2016 17:28:20 GMT -5 Boomzilla said:

The Cary, if I understand the manual correctly, automatically revises ALL incoming feeds to 32 bit depth.

I'll post a link when my Cary review is published on the Secrets of High Fidelity & Home Theater website.

Keith and Boomzilla:

Both posts are completely clear. Would suggest that the improvement I hear in JRiver upsampling is mostly from the improved bit rate and that the constant 32 bit rate of the Cary might explain why you heard little improvement among various upsamples.

Craig

[yves]

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

To answer that original question......

Since information cannot be created, if by "quality" you mean "information accurate to the original", then NOTHING can give you more quality than was there in the original file.
At most, by converting to some specific format or sample rate, you may be allowing a specific DAC to do a better job of causing less damage to the information that is present.
(This is why most modern DACs do internal oversampling; because the oversampled signal can be converted more accurately with less alteration by the reconstruction filters.)

Yes, oversampling increases bandwidth because the upper frequency limit according to the Sampling Theory is called the Nyquist frequency, or the sampling frequency divided by two. Whereas the total noise POWER (not to be confused with noise level) remains the same, but is spread over a larger spectrum due to the increased bandwidth so as a result there is a decline in noise in the band of interest (below 20kHz). Sigma Delta's inherent noise shaping shifts most of the noise towards the upper frequency limit of this spectrum. So the larger this spectrum, the farther the noise is shifted upwards in frequency, away from the band of interest, where it can be more easily filtered with less steep, more gentle/relaxed filters. Modern advanced Sigma Delta DAC relies on heavy oversampling.

In Sigma Delta ADC (Analog-to-Digital Converter) the effectiveness of dither is also enhanced by oversampling because the dither noise that is added before the modulator can be subtracted again after. www.mouser.sg/pdfdocs/src-tutorials/ADC-Noise.pdf

And, since each conversion may alter the sound a little bit, you may simply like the way audio sounds after being passed through a certain conversion process.
(And it's the job of the marketing department at any company to convince you that the differences in their product make it better.
And, if you're the type of person who simply likes more choices, and more options to choose from, then DSD is simply one more option to choose from.

I should also note that you need to be careful when reading white papers..... many of them are quite old and digital audio technology changes quite rapidly.
One of the linked white papers compared DSD with single-bit Delta-Sigma converters.... but virtually all modern DACs use multi-bit Delta-Sigma converters.

Yes, but to be able to output DSD the output of the multi-bit Sigma Delta Modulator must be re-quantized from multi-bit to 1-bit, for reasons that are obvious. The problem with that is a 1-bit quantizer cannot adequately be dithered without sacrificing a lot of Dynamic Range, as John Siau of Benchmark Media Systems has explained in that article that was linked by someone right here in this thread,
emotivalounge.proboards.com/post/852269/thread
...and here,
www.realhd-audio.com/?p=74

Likewise, it used to be common to compare SACDs to CDs (DSDx1 to 16/44k PCM); but high-resolution 24/96k PCM audio files are becoming quite common these days.

Personally, I, find that SACD performs in some ways sonically worse than Redbook CD. Quantization distortion happens to be a DIGITAL type of distortion, which most DSD advocates seem to purposefully tend to "forget".

It is my understanding that the Oppo CAN receive DSD via DLNA (over Ethernet)...
And this is claimed to work with some DLNA servers (perhaps some versions of Twonky).
However, I've read that it won't work with jRiver.
(If you Google "Oppo DSD DLNA" you can find a lot of discussion and suggestions on the subject...)

You could try HERE for starters:
www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/oppo-bdp-10%5B3-5%5D-d-and-multichannel-direct-stream-digital-playback-26808/

Oppo makes for a very good Blu-ray player. As a direct competitor against a dedicated, or "minimalist" DAC unit worth its salt, not so much.

[geebo]

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

It is my understanding that the Oppo CAN receive DSD via DLNA (over Ethernet)...
And this is claimed to work with some DLNA servers (perhaps some versions of Twonky).
However, I've read that it won't work with jRiver.
(If you Google "Oppo DSD DLNA" you can find a lot of discussion and suggestions on the subject...)

You could try HERE for starters:
www.computeraudiophile.com/f22-networking-networked-audio-and-streaming/oppo-bdp-10%5B3-5%5D-d-and-multichannel-direct-stream-digital-playback-26808/

I use jRiver MC 22 and just this weekend I was streaming DSD from my NAS to the Oppo and an turn sending that to the XMC. The XMC display showed DSD so it would seem jRiver (at least ver 22) does stream DSD.

[craigl59]

Oct 17, 2016 18:51:01 GMT -5 yves said:

Oct 17, 2016 15:52:47 GMT -5 KeithL said:

Yes, oversampling increases bandwidth because the upper frequency limit according to the Sampling Theory is called the Nyquist frequency, or the sampling frequency divided by two. Whereas the total noise POWER (not to be confused with noise level) remains the same, but is spread over a larger spectrum due to the increased bandwidth so as a result there is a decline in noise in the band of interest (below 20kHz). Sigma Delta's inherent noise shaping shifts most of the noise towards the upper frequency limit of this spectrum. So the larger this spectrum, the farther the noise is shifted upwards in frequency, away from the band of interest, where it can be more easily filtered with less steep, more gentle/relaxed filters. Modern advanced Sigma Delta DAC relies on heavy oversampling.

yves:

Thanks for this explanation; it clears up a topic I have wondered about for some time.