Multi-bit DSD versus PCM

[Fitzcaraldo215]

You say you understand how modern SDM DAC chips work, and you never even tried Hqplayer, but you don't agree that it's a better way to do things.

Do you not see a problem of why it would be hard to put much weight on your opinion of this matter?

No, I see a problem putting any weight on your opinion. Look, I am not trying to convince you that I am right, as you seem to be trying to do with me. HQPlayer might, indeed, be the greatest thing ever in certain circumstances. I stated that before, and I have no proof it is not. But, I do not find your opinion or arguments about that to be convincing. You have not proven, try as you might, that it is as outstanding as you claim. Someday, I might elect to give it a try, if I see fit. But, neither you nor the product specs and capabilities nor other's listening opinions have yet convinced me.

Others will have to decide whether they accept your opinion, my opinion, or some other opinion entirely. So, let's agree to disagree. Happy listening!

 

[Blizzard]

No, I see a problem putting any weight on your opinion. Look, I am not trying to convince you that I am right, as you seem to be trying to do with me. HQPlayer might, indeed, be the greatest thing ever in certain circumstances. I stated that before, and I have no proof it is not. But, I do not find your opinion or arguments about that to be convincing. You have not proven, try as you might, that it is as outstanding as you claim. Someday, I might elect to give it a try, if I see fit. But, neither you nor the product specs and capabilities nor other's listening opinions have yet convinced me.

Others will have to decide whether they accept your opinion, my opinion, or some other opinion entirely. So, let's agree to disagree. Happy listening!

How do you prove anything in audio? If you show measurements, it doesn't guarantee great sound, and if 999 out of 1000 prefer it subjectively, it still is no guarantee you will feel the same.

Looking at a modulator/filter algorithm plot on paper will never do much for understanding how it will sound. Unless you're an expert on that sort of thing. Kinda like an auto mechanic examining a CT scan of a human brain for a rare cancer. It's very complex how these algorithms work. So the only way to truly judge this is by trying and listening. Until then, there's nothing that will ever convince you.

But if you want to try to understand, here's some guides to get you started:

SDM modulation:

http://www.numerix-dsp.com/appsnotes/APR8-sigma-delta.pdf

Digital filters:

http://www.resonessencelabs.com/digital-filters/

Happy listening as well!

 

[Blizzard]

More examples of a DSD upsampling DAC's. Same principal only they use a low powered FPGA to do the algorithms instead:


http://www.psaudio.com/directstream-dac/#lightbox/2/

All of Emm Labs/Meitner DAC's upsample all PCM to double DSD:


http://www.emmlabs.com/html/audio/dac2/dac2.html

 

[Blizzard]

I found a good article explaining multibit DSD or "DSDwide" and how it's not the same as PCM. This is what the modulator section of a modern SDM chip converts the DSD to to allow DSP for volume control. It's also what HQplayer does to apply the DSP to DSD:

"To address some of these issues, a new studio format has been developed, usually referred to as "DSD-wide", which retains the high sample rate of standard DSD, but uses an 8-bit, rather than single-bit digital word length, yet still relies heavily on the noise shaping principle. It becomes almost the same as PCM—and is sometimes disparagingly referred to as "PCM-narrow"—but has the added benefit of making DSP operations in the studio a great deal more practical. The main difference is that "DSD-wide" still retains 2.8224 MHz (64Fs) sampling frequency while the highest frequency in which PCM is being edited is 352.8 kHz (8Fs). The "DSD-wide" signal is down-converted to regular DSD for SACD mastering. As a result of this technique and other developments there are now a few digital audio workstations (DAWs) that operate, or can operate, in the DSD domain, notably Pyramix and some SADiE systems."

http://www.di.unipi.it/~romani/DIDATTICA/AD/Direct Stream Digital.pdf

 

[amirm]

I found a good article explaining multibit DSD or "DSDwide" and how it's not the same as PCM. This is what the modulator section of a modern SDM chip converts the DSD to to allow DSP for volume control. It's also what HQplayer does to apply the DSP to DSD:

"To address some of these issues, a new studio format has been developed, usually referred to as "DSD-wide", which retains the high sample rate of standard DSD, but uses an 8-bit, rather than single-bit digital word length, yet still relies heavily on the noise shaping principle. It becomes almost the same as PCM—and is sometimes disparagingly referred to as "PCM-narrow"—but has the added benefit of making DSP operations in the studio a great deal more practical. The main difference is that "DSD-wide" still retains 2.8224 MHz (64Fs) sampling frequency while the highest frequency in which PCM is being edited is 352.8 kHz (8Fs). The "DSD-wide" signal is down-converted to regular DSD for SACD mastering. As a result of this technique and other developments there are now a few digital audio workstations (DAWs) that operate, or can operate, in the DSD domain, notably Pyramix and some SADiE systems."

http://www.di.unipi.it/~romani/DIDATTICA/AD/Direct Stream Digital.pdf

DSD wide is PCM. It is not DSD. Just because it has the same sample rate as consumer DSD, it doesn't mean it is DSD.

His mistake and that of yours stems from thinking PCM must have the handful of set sample rates that are standardized for consumer use. Such is not the case. PCM as a signal processing format has no set sampling rate. It just says how the bits are coded. Therefore an 8-bit system sampling at 2.8 Mhz, *is* PCM. It is not a standardized PCM sample rate in our consumer systems. But then again, neither is 8-bit "DSD." Your DSD Dac would not play that stream either. It is yet another non-standard (in consumer field) format. This is why this is correctly called "PCM narrow."

Regardless, DSD wide is NOT quad DSD which you advocate. DSD wide is a proper PCM format that as he mentions, can be edited. And because it is multi-bit, it can be dithered and hence does not have the problems of normal consumer DSD. I would happily get behind DSD wide just as well as PCM. Very different animal than consumer DSD, no matter what its sample rate.

 

[Blizzard]

He is mistaken in his explanation. PCM as a signal processing format has no set sampling rate. It just says how the bits are coded. An 8-bit system sampling at 2.8 Mhz, *is* PCM. It is not a standardized PCM sample rate in our systems. But then again, neither is 8-bit "DSD." Your DSD Dac would not play that stream either. It is yet another non-standard (in consumer field) format.

Regardless, DSD wide is NOT quad DSD which you advocate. DSD wide is a proper PCM format that as he mentions, can be edited. And because it is multi-bit, it can be dithered and hence does not have the problems of normal consumer DSD.

I just realized that's direct from Wikipedia

https://en.wikipedia.org/wiki/Direct_Stream_Digital

Anyways, the multibit DSD or Multibit SDM allows the DSP to be applied without downsampling to PCM resolutions. So it has advantages over converting to PCM rates like DxD 24/352.8

I never said DSD wide was quad DSD. It's multibit DSD, and can be done to any DSD sample rate without downsampling it.

Anyways whatever you want to call it, it's better than downsampling to a PCM format like DxD to apply the DSP.

 

[Bruce B]

DSD wide (PCM narrow) is the editing format in Sonoma and SADiE
DXD (which is really PCM at 352.8/24) is the editing format in Pyramix at DSD64fs. Pyramix edits DSD128fs at 705.6kHz and edits Quad DSD at 1,411.2kHz (1.4Mz) PCM rate.

 

[amirm]

Anyways, the multibit DSD or Multibit SDM allows the DSP to be applied without downsampling to PCM resolutions. So it has advantages over converting to PCM rates like DxD 24/384.

There is no disadvantage in PCM in that manner. The numbers do not have the meaning you are assuming. That is, the higher sample rate of DSD does not at all mean that it has higher information capacity.

In signal processing there are two axis, sample rate and bit depth. You can trade one for the other and have the same amount of data capture as a matter of information theory. 16 bit 44.1 is the same as 15 bits at 176 Khz (post analog filtering down to 22.05 Khz). Let me repeat, even though the latter has less bits, it has the same signal to noise ratio of the 16 bit system. We simply traded bits for speed. I can keep climbing the sample rate and at the same time reduce the need for that number of bits. Or vice versa.

DXD at 24 bits/384 Khz sampling has exceptionally wide envelop. If we multiply 384 by 4 we get 1.5 mbit/sec. So 24 bit/384 is the same as 23 bits and 1.5 Mhz sampling. Quadruple that again and we 22 bits at whopping 6 Mhz. Double rate DSD runs at 5.6 Mhz but it only has a single bit of resolution at that sample rate! It is short 21 bits compared to DXD. Quad DXD suffers for the same reason.

So no way do you lose information when you convert to DXD. DXD has far more information capacity than DSD at any sample rate (when filtered to the audio bandwidth).

See Don's article on oversampling here: http://www.whatsbestforum.com/showthread.php?2487-DACs-102-Delta-Sigma-DACs

 

[Blizzard]

DSD wide (PCM narrow) is the editing format in Sonoma and SADiE
DXD (which is really PCM at 352.8/24) is the editing format in Pyramix at DSD64fs

DSD wide is the closest thing to what SDM DAC chips do to apply volume control to DSD. Same with HQplayer for DSP on DSD. Whatever name you want to call it, what makes it good is you don't need to downsample to do it. It's supposed to be completely lossless. However I've found it to slightly degrade, but not near as bad as offline or online conversion to PCM.

However DSDmaster:

http://dsdmaster.blogspot.ca

has improved it's algorithms since the last time I've tried according to their blog. So maybe it's a bit better than it was. It was the best DSD to PCM conversion software I tried when doing comparisons. Much much better than Jriver on the fly.

Their blog archive is a wealth of information as well for those who want an interesting read:


http://bitperfectsound.blogspot.ca/...-max=2016-01-01T00:00:00-05:00&max-results=37

 

[Blizzard]

There is no disadvantage in PCM in that manner. The numbers do not have the meaning you are assuming. That is, the higher sample rate of DSD does not at all mean that it has higher information capacity.

In signal processing there are two axis, sample rate and bit depth. You can trade one for the other and have the same amount of data capture as a matter of information theory. 16 bit 44.1 is the same as 15 bits at 176 Khz (post analog filtering down to 22.05 Khz). Let me repeat, even though the latter has less bits, it has the same signal to noise ratio of the 16 bit system. We simply traded bits for speed. I can keep climbing the sample rate and at the same time reduce the need for that number of bits. Or vice versa.

DXD at 24 bits/384 Khz sampling has exceptionally wide envelop. If we multiply 384 by 4 we get 1.5 mbit/sec. So 24 bit/384 is the same as 23 bits and 1.5 Mhz sampling. Quadruple that again and we 22 bits at whopping 6 Mhz. Double rate DSD runs at 5.6 Mhz but it only has a single bit of resolution at that sample rate! It is short 21 bits compared to DXD. Quad DXD suffers for the same reason.

So no way do you lose information when you convert to DXD. DXD has far more information capacity than DSD at any sample rate (when filtered to the audio bandwidth).

See Don's article on oversampling here: http://www.whatsbestforum.com/showthread.php?2487-DACs-102-Delta-Sigma-DACs

Maybe you don't lose much information doing a DxD conversion, but you are forced to go through extra filter and modulator algorithms in the DAC chip when the data is converted to a PCM format. This is when the sound quality is compromised, no matter if nothing is lost at all in the transfer.

The algorithms used to preform the transfer aren't all equal as well.

 

[Rodney Gold]

Even if you are right , you will never be right with Blizzard..last word champion ..

At the end of it all , better sounding often does not equate with better numbers.. the audibility of those "better numbers" has yet to be proven. Vinyl and valves make the case..many think they are better sounding than other equipment that is better specced.

All messing with the digital signal will do is change that signal .. not preserve it as original.. it will change the sound , but whether for the better is the users choice.

At the end of it all , what hits the ear is all important .. room mangles sound 100x more than any small digital change...

This is all subjective pontification, being dogmatic about "better" here is useless

 

[Blizzard]

Even if you are right , you will never be right with Blizzard..last word champion ..

At the end of it all , better sounding often does not equate with better numbers.. the audibility of those "better numbers" has yet to be proven. Vinyl and valves make the case..many think they are better sounding than other equipment that is better specced.

All messing with the digital signal will do is change that signal .. not preserve it as original.. it will change the sound , but whether for the better is the users choice.

At the end of it all , what hits the ear is all important .. room mangles sound 100x more than any small digital change...

This is all subjective pontification, being dogmatic about "better" here is useless

This discussion is about the best way to handle things inside a DAC chip. And the best way to handle DSP in conjunction with Roon. I have already tried your MiniDSP way, and found it to be subpar at very best. DSP being preformed in multibit DSD-wide or PCM narrow, or whatever term you want to use to describe it, is the very best way things can be done.

However your Devialet is only 24/192 capable anyways. So for you, I would just use Roon together with the PC based Dirac. Then go straight into the Devialet with the USB. Ditch the squeezebox and minidsp box. It will sound light years better. Not only that, the PC version can handle 24/192 rather than max out at 24/96. I have a real hard time understanding your reasons for doing things the way you're doing them. Adding all the unnecessary noisy, jittery, resource constrained boxes into the system when the PC running Roon can do a way better job on it's own.

 

[Rodney Gold]

You once again pontificate that it will sound miles better .. how do you know..? the same way that you know that laurence Dickie compromised the G1's by using non blizzard approved components in the crossover ?
My devialet does not have USB input anyway .. It is a Dpremier
You have no proof whatsoever that your path or your rube goldberg diy assemblies will make stufff sound better.
and how , what with the profound effect of DRC , can you say inserting any other unit or scheme will sound better - or that it is sub par..only sub par for your pie in the sky stuff...
Prove it sounds better or keep schtum.

 

[Blizzard]

You once again pontificate that it will sound miles better .. how do you know..? the same way that you know that laurence Dickie compromised the G1's by using non blizzard approved components in the crossover ?
My devialet does not have USB input anyway .. It is a Dpremier
You have no proof whatsoever that your path or your rube goldberg diy assemblies will make stufff sound better.
and how , what with the profound effect of DRC , can you say inserting any other unit or scheme will sound better - or that it is sub par..only sub par for your pie in the sky stuff...
Prove it sounds better or keep schtum.

I know because I've used the very best MiniDSP stuff, and I understand how DSP chips works, and that their stuff is mid-fi at best. I've used lots of DSP chip based products from a few companies.

And I also understand the shortcomings of SPDIF from a squeezebox into a minidsp box then into the Devialet. It's a ridiculously poor way to do things compared to the alternative I mentioned. Just try it, and you will see. The amount of jitter in your digital signal by the time it makes it to your DAC in the Devialet will be outrageous. Maybe acceptable for a $5000 system, but not a system the caliber of yours.

The Dirac started out as a PC based software. They teamed up with miniDSP afterwards. That minidsp box is mainly made for people who don't use a PC in the system. The PC based version is more powerful, and better. Cheaper too.

http://www.dirac.com/online-store/

If you don't have a USB input, get a good quality USB to SPDIF\AES\EBU bridge then. Lots of them out there.

This one here is really good and perfect for the job. Very well built with very low jitter Crystek clocks. The clocks are very important with SPDIF transports. And the clocks in the squeezebox and minidsp are a joke.

http://audiobyte.net/products/hydra-z

 

[Rodney Gold]

You still have to convince me and many others that your way will sound "better" , whatever your better means , and that all the numbers add up to sonic improvements..nothing you have said does so.

 

[Blizzard]

You still have to convince me and many others that your way will sound "better" , whatever your better means , and that all the numbers add up to sonic improvements..nothing you have said does so.

Not sure how I could do that from here. You just need to try it and see. It's not "my way". I didn't invent the PC based Dirac system, I didn't make the USB bridge. It's simply a better way to do things. I'm just telling you about it. It really should make sense to you why it would be better. It's not my fault if it doesn't.

So far 100% of everyone who has listened to my advice on here has been happy with the results, so I'm not sure where you are getting your info from. You should be grateful that I'm taking the time to tell you this, rather than criticizing me about it.

 

[Rodney Gold]

Im critical of bench racers.. design a car in their minds with a 1000kw motor and all in theory , but have no idea of how it will actually perform on the track..seen a million of em..all hot air

 

[Blizzard]

Im critical of bench racers.. design a car in their minds with a 1000kw motor and all in theory , but have no idea of how it will actually perform on the track..seen a million of em..all hot air

Yeah okay. Listen to your joke of a setup as it is then. Ignorance is bliss.

 

[Rodney Gold]

Ignorance might be bliss but a little knowledge is a dangerous thing..

 

[Phelonious Ponk]

Yes some are slower to catch on than others I don't think theres any dispute from anyone that quad DSD isn't superior. What their end of the story is the most popular formats can be done better (PCM) using their approach. But they are also not using third party software based SDM/SRC when making this claim either. This is assuming a regular native format is sent to the DAC as is. And once again this is based chips from yesteryear as well.

It's good to pay attention to English as well as digital. This double negative actually says that quad DSD is not superior; I don't think that's what you meant to say. For those of us who understand English, but not digital, we all start with Redbook, no? Are you saying that converting it makes it sound better?

Tim