To DSD or not to DSD?

[Hiro]

The "human ears sense sounds at frequencies between 20 an 20,000 cycles per second (Hz)" canard does not change the fact noise above 50kHz is massively amplified by 7th order Delta-Sigma modulation. In the case of an 1-bit system, which any system that is outputting an 1-bit serial bitstream (DSD) still is, the high-frequency noise is so much of a problem that it needs to be very aggressively filtered, i.e. by using a filter that is even more aggressive than the decimation filter DSD fans always keep crying about.

 

PCM uses far more processing than the aggressive digital LPF and decimation filter. And in most cases it uses delta sigma modulation and noise shaping too.

 

On top of that, proper dither of an 1-bit system is simply not possible in any way at all. (And I really very much hate to keep sounding like a broken record on this, but it has long been established by Lipshitz, Vanderkooy & Wannamaker in a series of IMO fairly groundbreaking AES papers, so I still think it is safe for me to completely ignore pretty much whatever it is that Andreas Koch would like me to believe instead).

 

Miska's real-live measurements of Sigma Delta modulators are far more informative than theorizing on paper.

[spdif-usb]

I do a lot of listening too. I demand both, good sound and good theoretical and technical performance (measurement results).

 

If measurement results are considered a marketing ploy, I'm fine with it. Anybody is free to listen with their own ears and make their judgement. I'm just sharing the information I have. I don't have a particular agenda, I support both PCM and SDM outputs and conversion from any input format to any output format, so I'm kind of format agnostic in that sense. HQPlayer can do 1536/32 TPDF-dithered (or noise shaped) PCM output too.

I am glad we can put to rest the old adage that people from the likes of Andreas Koch and Paul McGowan are necessarily being correct when it comes to the tech explanations given by them on the subject of DSD vs. PCM. In pure terms of listening with my own ears, I can only echo Barry Diament's sentiment IIRC, i.e. that properly done PCM at 4x rates sounds audibly transparent, whereas properly done PCM at less than 4x rates doesn't, and that properly done DSD doesn't do it either.

 

That is not at all the most important reason why I hardly ever listen to DSD music for my personal enjoyment, though. I could easily live with DSD128fs because IMO it does sound close enough to 24/96 PCM, all else being equal, and besides, the vast majority of music I listen to is still only 24/96 PCM. I find that truly very excellent sounding 24/192 PCM is still a bit of a rarity, to say the least. The bottom line to me, however, is just the fact that music I like seems virtually nonexistent in DSD128fs, and completely nonexistent in DSD256fs.

[spdif-usb]

PCM uses far more processing than the aggressive digital LPF and decimation filter. And in most cases it uses delta sigma modulation and noise shaping too.

The fact that it uses far more processing is largely irrelevant in human audible terms because of too many other factors (e.g., the various filter characteristics and the use of dither, etc.). I am guessing that, if it were possible to prove beyond the shadow of a doubt that having less processing necessarily always implies better results, Meitner would probably not be advocating MDAT upsampling either.

Miska's real-live measurements of Sigma Delta modulators are far more informative than theorizing on paper.

Whatever you choose to believe is fine with me. Personally, I, think real-live measurements often can be terribly misleading as they hardly ever tell the whole story in the first place so it is kind of like judging marriage by looking at a wedding picture. On top of that, if you really think theorizing on paper is not so informative then I guess you conveniently forgot to mention that this also applies to what people like Andreas Koch have theorized on paper, and I guess you also conveniently forgot to mention that it also applies to your own theorizing on paper about whether or not interpretations of Miska's real-live measurements are far more informative. Did I miss something?

[bogi]

That is not at all the most important reason why I hardly ever listen to DSD music for my personal enjoyment, though. I could easily live with DSD128fs because IMO it does sound close enough to 24/96 PCM, all else being equal, and besides, the vast majority of music I listen to is still only 24/96 PCM. I find that truly very excellent sounding 24/192 PCM is still a bit of a rarity, to say the least. The bottom line to me, however, is just the fact that music I like seems virtually nonexistent in DSD128fs, and completely nonexistent in DSD256fs.

 

That's your personal listening experience, mine is other. We use different equipment, we have different ears and may be our brains don't process music quite in the same way in all details.

 

On my relatively cheap setup, using on the fly conversion of Redbook material to DSD128 brings clearly better result to my ears than doing anything else: playing bit perfect 44.1/16, or any PCM upsampling up to 384/32. I mean that's mostly about my DAC. From the point I discovered that with my DAC I listen ripped CDs this way. I convert all to DSD128, including 96/24 and 192/24 PCM recordings, because to my ears they also sound better this way.

 

This fact turned my interest into DSD world. Not planned or wished, it simply happened. I couldn't think previously that listening of CD material could be improved (with some DACs) by converting it to DSD.

[4est]

Yes, you did. Miska provides much more than measurements. The sound of his software speaks volumes for the net of his abilities. I have yet to see a more universally well regarded software player, or a more knowledgeable poster about DSD than he.

Did I miss something?

[Hiro]

The fact that it uses far more processing is largely irrelevant in human audible terms because of too many other factors (e.g., the various filter characteristics and the use of dither, etc.). I am guessing that, if it were possible to prove beyond the shadow of a doubt that having less processing necessarily always implies better results, Meitner would probably not be advocating MDAT upsampling either.

 

Well, many recordings are done in PCM (and most at 44.1/48kHz at that!) so no wonder he is advocating upsampling them to 2x DSD. If all recordings were done at 2x DSD they could simply be sent to his 2x DSD DACs as is (without downsampling/decimation/digital LPF/upsampling/brick wall fiters/SRC).

 

[spdif-usb]

Yes, you did. Miska provides much more than measurements. The sound of his software speaks volumes for the net of his abilities. I have yet to see a more universally well regarded software player, or a more knowledgeable poster about DSD than he.

That is only your opinion.

[spdif-usb]

Well, many recordings are done in PCM (and most at 44.1/48kHz at that!) so no wonder he is advocating upsampling them to 2x DSD. If all recordings were done at 2x DSD they could simply be sent to his 2x DSD DACs as is (without downsampling/decimation/digital LPF/upsampling/brick wall fiters/SRC).

 

Well, I don't see a reason why all recordings should be done at 2x DSD. At all.

[4est]

Everything is opinion. You are mentioning engineers and papers as evidence, and Miska provides the present SOTA item using the technology.

[Miska]

The "human ears sense sounds at frequencies between 20 an 20,000 cycles per second (Hz)" canard does not change the fact noise above 50kHz is massively amplified by 7th order Delta-Sigma modulation. In the case of an 1-bit system, which any system that is outputting an 1-bit serial bitstream (DSD) still is, the high-frequency noise is so much of a problem that it needs to be very aggressively filtered, i.e. by using a filter that is even more aggressive than the decimation filter DSD fans always keep crying about.

 

That's not true. For proof, just look at my DSC1 DAC, specifically designed to work nicely with 7th order modulator. You can also use PCM1795 from TI or other DSD capable chips.

 

Again, if you would have more than 1-bit in parallel and dither it like PCM, then as the level drops the system would quickly become 1-bit. Quite many multi-level modulators are 5-level. If the output would be like PCM it would become 1-bit below -14 dB level. As the RMS level rarely exceeds that (unless the content is heavily compressed), you would essentially always operate it as a 1-bit DAC.

 

On top of that, proper dither of an 1-bit system is simply not possible in any way at all. (And I really very much hate to keep sounding like a broken record on this, but it has long been established by Lipshitz, Vanderkooy & Wannamaker in a series of IMO fairly groundbreaking AES papers, so I still think it is safe for me to completely ignore pretty much whatever it is that Andreas Koch would like me to believe instead).

 

And that's still not true. Again, number of output bits is no no relation to properly dithered SDM. And just like ESS says "HyperStream multiple-bits are in Time" (nothing really special with that, btw). DSD is 1-bit in amplitude, but doesn't say anything about bit organization in time domain, completely up to you.

 

The paper of those guys is over ten years old and algorithms have marched forward a long way since then. Their paper is just ridiculous in that it presents a naive modulator design and is then trying to prove that it's bad...

 

I have shown you a practical example operating on complex input signal having all kinds of difficult properties simultaneously and there are no errors within 180 dB range of the plot.

[gumby]

This AFAIK has been thoroughly debunked by Lipshitz, Vanderkooy & Wannamaker.

 

Hi spdif-usb,

 

Attached is a paper written by Stanley P. Lipshitz and John Vanderkooy for the 2001 AES Convention outlining the reasons 1-bit Sigma Delta converters are “unsuitable” for high quality applications. Basically, they were saying it cannot be dithered as you have noted. You have probably read this.

Also attached is a rebuttle paper from the same 2001 AES convention. You have probably also read this one. But if you haven’t, you might enjoy this also. This article, in essence, states that a multibit Sigma Delta converter can be dithered and discusses how to design a better converter. I cannot speak for Miska, but I think this is essentially what he has been trying to explain to you.

Why 1 Bit Delta Sigma is Bad_2001 Convention.pdf

Why DSD is Best_2001 Convention.pdf

[davide256]

hmm, a lot of tech talk here that's appropriate inside an engineering office trying to design a better DAC but .... not what I think convinces anyone.

 

My repeatable, observable result... DSD on native DSD recording fed through Prima Luna integrated pre section a nice difference but not one to make you stand up and cheer. Native DSD fed through a Conrad Johnson PV10A pre section an outstanding difference fed to same system. Approaches the difference I experience wandering into a DC Audio show room playing PCM with a $500K system... but my system is only $12k in cost

 

Net conclusion: your mileage with DSD will depend on the quality of your pre and other system limitations.

 

Every SACD I own is now winging its way to have the DSD extracted to my harddrive so that they are useful music vs useless audio paperweights.

[davide256]

And if you are in the audio recording industry reading this, take note. Before DSD DAC's I had zero, nada interest in buying crippled SACD discs and was moving to digital downloads only. Given the current weak catalog of downloadable DSD, the prospect of buying SACD physical media and ripping to disc becomes very attractive. Wish you would do like the blue ray industry and include a DSD download code in the purchase, I'd pay a few bucks more if you didn't screw it up with DRM.

[davide256]

apologies on the above... I meant ripping to hard drive. But in saying disc I can see where a manufacturer would want a security solution that would not allow disc copy, only drive download and drive backup.

[Paul R]

Naw - Miska is providing evidence from direct engineering experience and design. The spdif-usb opinions are about as far from the mark as it is possible to be.

 

-Paul

 

 

Everything is opinion. You are mentioning engineers and papers as evidence, and Miska provides the present SOTA item using the technology.

[spdif-usb]

Hi spdif-usb,

 

Attached is a paper written by Stanley P. Lipshitz and John Vanderkooy for the 2001 AES Convention outlining the reasons 1-bit Sigma Delta converters are “unsuitable” for high quality applications. Basically, they were saying it cannot be dithered as you have noted. You have probably read this.

Also attached is a rebuttle paper from the same 2001 AES convention. You have probably also read this one. But if you haven’t, you might enjoy this also. This article, in essence, states that a multibit Sigma Delta converter can be dithered and discusses how to design a better converter. I cannot speak for Miska, but I think this is essentially what he has been trying to explain to you.

Convention paper 5395 (i.e. the first one of the two papers that you posted) was only the very start of what went after it. One year later, in 2002, Philips published convention paper 5616 to argue to the contrary. In convention paper 5619, Jamie Angus criticized Lipshitz & Vanderkooy's paper, to which Liphitz & Vanderkooy later responded, i.e. in convention paper 5620. (Convention papers 5398, 5477, and 6093 are part of the same series "Towards a Better Understanding of 1-Bit Sigma-Delta Modulators", by Lipshitz & Vanderkooy).

 

P.S. - From the second one of the two papers that you posted, I quote:

"In DSD jargon, the SDM which is used to convert a multi-bit signal back to a single bit singal, is called a re-quantizer or re-modulator." To which I will add that the multi-bit signal cannot be converted back to DSD except if this re-quantizer is NOT properly dithered.

[spdif-usb]

Naw - Miska is providing evidence from direct engineering experience and design. The spdif-usb opinions are about as far from the mark as it is possible to be.

 

-Paul

That is only your opinion.

[Maldur]

I prefer Miska's knowledge and experience more than some 14-15 years old "historical papers", written by people with built-in PCM box on their head. So, thats not only Paul's opinion.

[wisnon]

I have lost count of them. The latest one that I have heard was the PS Audio DirectStream DAC.

 

That explains it, especially if it was the older firmware…you have not heard DSD done right. The latest FW is reportedly decent, but I have not heard it myself.

[matthias]

That is only your opinion.

 

Miska is the person on CA I trust the most. AFAIK, his knowledge and experience regarding PCM and DSD is without peer. His contributions on CA are the most important for me and they confirm my opinion about the superiority of DSD128+ over PCM.

 

KR

Matt

[spdif-usb]

I prefer Miska's knowledge and experience more than some 14-15 years old "historical papers", written by people with built-in PCM box on their head. So, thats not only Paul's opinion.

It doesn't matter that it's old. Pythagoras is even much older now, but that doesn't prove that he was wrong, does it? The fact that it's not only Paul's opinion also doesn't matter to me. That's because I am old enough to form my own opinion. :grin:

[spdif-usb]

Miska is the person on CA I trust the most. AFAIK, his knowledge and experience regarding PCM and DSD is without peer. His contributions on CA are the most important for me and they confirm my opinion about the superiority of DSD128+ over PCM.

 

KR

Matt

The difference is in the "AFAIK" here.

[matthias]

The difference is in the "AFAIK" here.

 

Look at your post #1067.

 

Matt

[bogi]

I prefer Miska's knowledge and experience more than some 14-15 years old "historical papers", written by people with built-in PCM box on their head. So, thats not only Paul's opinion.

 

+1

[spdif-usb]

That explains it, especially if it was the older firmware…you have not heard DSD done right. The latest FW is reportedly decent, but I have not heard it myself.

I have heard the DirectStream DAC with the new firmware. Like I said though, it isn't the only DSD that I have heard, and I have heard so much stuff by now that I am unable to recall all the various makes and models because I am a busy person at times.