Does DSD sound better than PCM?

[Miska]

Then you are assuming anti-image filters are inadequate and are assuming your definition of "best reconstruction" is the best there is. I am not sure it is possible to not assume something, somewhere, at a very basic level.

I was earlier demonstrating that they are not perfect and how to make them tens of dB's better.

In SDM DAC that is only part of the reconstruction anyway...

I said nothing about computing limitations; not my field, decades since my digital filter design classes, and I had no thought that modern CPUs and/or DSPs couldn't handle the load.

The tiny little DSP inside a ~$10 DAC chip is not so great... Cutting lot of corners...

But of course just digital filter is only part of the picture, because performance also involves modulator design, etc.

 

[captain paranoia]

White paper by Bruno Putzeys and Eelco Grimm starts with "DSD is another word for 1-bit PCM sampled at 2.8224MHz"

Pulse Code Modulation and Pulse Density Modulation are two entirely different sampling methods.
PDM uses a one-bit element for both sampling and reconstruction, but it is not PCM.

https://en.wikipedia.org/wiki/Pulse-code_modulation
https://en.wikipedia.org/wiki/Pulse-density_modulation

Note, in particular
https://en.wikipedia.org/wiki/Pulse-density_modulation#Applications

 

[pjug]

Pulse Code Modulation and Pulse Density Modulation are two entirely different sampling methods.
PDM uses a one-bit element for both sampling and reconstruction, but it is not PCM.

https://en.wikipedia.org/wiki/Pulse-code_modulation
https://en.wikipedia.org/wiki/Pulse-density_modulation

Note, in particular
https://en.wikipedia.org/wiki/Pulse-density_modulation#Applications

I'm not sure if you are disagreeing with what Bruno Putzeys wrote or not. But he knows what he is talking about. As does @mansr. As decent as Wikipedia is on some matters, it would not be my go-to source for information on highly technical topics.

 

[Miska]

From practical perspective, you can consider SDM (DSD) as PWM (difference between PDM and PWM is rather minor detail when digitally generated).

For some people, anything is PCM and if they like to think that way, I don't mind.

PCM needs only quantizer, SDM needs a modulator.

Bruno calls the Mola-Mola DAC he designed a 1-bit PWM DAC. One could as well call it a DSD DAC. But since "DSD" is just a marketing term created for SACD rather than a technical term, I'd rather avoid using it and talk about something else instead.

 

[pjug]

From practical perspective, you can consider SDM (DSD) as PWM (difference between PDM and PWM is rather minor detail when digitally generated).

For some people, anything is PCM and if they like to think that way, I don't mind.

PCM needs only quantizer, SDM needs a modulator.

Bruno calls the Mola-Mola DAC he designed a 1-bit PWM DAC. One could as well call it a DSD DAC. But since "DSD" is just a marketing term created for SACD rather than a technical term, I'd rather avoid using it and talk about something else instead.

I thought that the distinction that some are drawing is that PCM has a specific sample rate. Whereas with true PWM or PDM, the PW or PD varies on a continuum. And so if we define in that way then DSD is PCM. I can see where the terminology might not matter so much. But on the other forum the whole argument started from someone claiming that DSD was entirely different from PCM.

 

[Miska]

I thought that the distinction that some are drawing is that PCM has a specific sample rate. Whereas with true PWM or PDM, the PW or PD varies on a continuum. And so if we define in that way then DSD is PCM. I can see where the terminology might not matter so much. But on the other forum the whole argument started from someone claiming that DSD was entirely different from PCM.

Discrete digitally generated SDM/PWM/PDM has a sample rate. It is just "non-Nyquist sampling". So the sampling rate is not directly related to wanted bandwidth. Instead the audio bandwidth is vaguely something where "local SNR" becomes close to 6 dB.

But overall, most PCM you see these days has been converted from SDM. Number of pure PCM ADCs is going down with pretty much just some Pacific Microsonics Model Ones and Twos around. For DACs you can still find some PCM DACs based on TI's PCM1704 (should be paired with DF1704 or DF1706 digital filter) and some new discrete R2R implementations. And some very limited stock editions based on Philips TDA154x or Analog Devices chips. I have recycled my old Marantz CD-60 long time ago that had the classic SAA7220 digital filter + TDA1541A pair (maybe I shouldn't have, because it was still operational).

I still have two old PCM DACs around that I built myself. One with DF1700 + PCM63P pair and another with DF1700 + PCM1700 pair.

I also have the "first DSD DAC" which is Crystal CS4328 - 1-bit 2.8 MHz thingie, also one I built myself.

 

[captain paranoia]

As decent as Wikipedia is on some matters, it would not be my go-to source for information on highly technical topics.

I posted the Wiki links because, as a practicing electronic design engineer, they reflect my understaning of the terms PCM, delta modulation, PPM, PDM and PWM. I'd prefer to go to Wiki than the website of a manufacturer; too much simplification and marketing puff in manufacturers' websites. Calling delta modulation a '1-bit PCM' is one such simplification.

On the whole, I find Wiki very useful for technical topics. It's handy to have an understanding to be able to vet articles, but that's true of anything on the internet, or elsewhere...

 

[andymok]

So I was researching and found this interesting writing from 12 years ago

https://www.gearslutz.com/board/showpost.php?p=1712442&postcount=30

Hi. My name is Peter Scheelke. I am one of the founders of DAD, Digital Audio Denmark. We are the manufacturer of theAX24 AD/DA converter and the Merging Sphynx 2 converter.
In the discussion about DSD/DXD, Sadie/Sonoma/PYRAMIX I can help clarify some of the advantages/disadvantages in the different audio formats.

Further responses regarding DXD and DSD can be found in #34-35, 47, 51, (101, 104, 115)
https://www.gearslutz.com/board/rem...nd-location-recording/30990-dsd-wdsd-dxd.html

Spoiler

The main advantage of high sampling rate is however not the wider frequency band itself since we can not hear frequencies above 20 KHz.

should you care about how different DAWs handles DSD, see #10 as well
https://www.gearslutz.com/board/showpost.php?p=305114&postcount=10

 

[John Deas]

I've just spent the last few hours reading back through this thread from the start (phew...) i've learned a lot and been totally confused. My personal answer to the question based on having read, studied and corresponded with people in the audio recording business is that DSD is a superior way of capturing music but a pain in the arse because of it's inability to be mixed easily i.e without conversion to PCM (which slightly defeats the object of capturing in DSD) or mixing out to analog or mixing prior to capture in analog. The fact that many very reputable sound engineers are willing to go to the hassle of working in DSD tells me that there is something beneficial going on with this format - and don't say "yeah because they can charge you more" because no-one out there is making a fortune out of flogging DSD!
How it sounds on playback is, obviously made clear by all the proceeding posts, totally confusing because of the many variables people have in their set-ups. Many peoples experience of hearing DSD is from SACD and this is not a good basis to comment from; most SACD players convert to PCM and the vast majority of SACD's started as PCM or have been mixed through PCM, also it's only DSD64, now superseded by 128 and 256. So the first thing is get some DSD files that have preferably not been mixed in PCM and try those out to give a true representation of what this format can or cannot do. PCM and DSD do sound different all other factors assumed equal, at the end of the day it is a red or white wine question for the individual, personally I prefer red wine and DSD over white wine and PCM.

 

[pkane]

I've just spent the last few hours reading back through this thread from the start (phew...) i've learned a lot and been totally confused. My personal answer to the question based on having read, studied and corresponded with people in the audio recording business is that DSD is a superior way of capturing music but a pain in the arse because of it's inability to be mixed easily i.e without conversion to PCM (which slightly defeats the object of capturing in DSD) or mixing out to analog or mixing prior to capture in analog. The fact that many very reputable sound engineers are willing to go to the hassle of working in DSD tells me that there is something beneficial going on with this format - and don't say "yeah because they can charge you more" because no-one out there is making a fortune out of flogging DSD!
How it sounds on playback is, obviously made clear by all the proceeding posts, totally confusing because of the many variables people have in their set-ups. Many peoples experience of hearing DSD is from SACD and this is not a good basis to comment from; most SACD players convert to PCM and the vast majority of SACD's started as PCM or have been mixed through PCM, also it's only DSD64, now superseded by 128 and 256. So the first thing is get some DSD files that have preferably not been mixed in PCM and try those out to give a true representation of what this format can or cannot do. PCM and DSD do sound different all other factors assumed equal, at the end of the day it is a red or white wine question for the individual, personally I prefer red wine and DSD over white wine and PCM.

Personal preferences aside, DSD is just an unwieldy format, especially in DSD256 or DSD512. And not just for mastering. Try streaming it over WiFi, like I did, and you'll quickly find out that PCM 192/24 is way superior

 

[Soniclife]

So the first thing is get some DSD files that have preferably not been mixed in PCM and try those out to give a true representation of what this format can or cannot do.

To what end? Without a pcm dual recorded version I can only see how this will add to confusion.

 

[JohnYang1997]

Is there any downside of 384khz 32bit? By any stretch of imagination and any technical standard? (besides file size)
Like bandwidth, jitter, digital filters, processing power etc

 

[John Deas]

Personal preferences aside, DSD is just an unwieldy format, especially in DSD256 or DSD512. And not just for mastering. Try streaming it over WiFi, like I did, and you'll quickly find out that PCM 192/24 is way superior

"Unwieldy" is a polite way of saying pain in the arse which it no-doubt is. I'm not remotely interested in streaming anything so no it's not superior to me in that context

 

[John Deas]

To what end? Without a pcm dual recorded version I can only see how this will add to confusion.

I can't disagree with you there. I'm not aware of any files created from the same source to the two formats at the same time - unless anyone knows different? However at least try to start with pure DSD files not SACD sources.

 

[andreasmaaan]

PCM and DSD do sound different all other factors assumed equal, at the end of the day it is a red or white wine question for the individual, personally I prefer red wine and DSD over white wine and PCM.

I think you'll find most here disagree with this statement. DSD and PCM can sound different, but only when poorly implemented. Done properly, there will be no audible difference.

 

[Frank Dernie]

Is there any downside of 384khz 32bit? By any stretch of imagination and any technical standard? (besides file size)
Like bandwidth, jitter, digital filters, processing power etc

Pointlessness?

 

[John Deas]

Is there any downside of 384khz 32bit? By any stretch of imagination and any technical standard? (besides file size)
Like bandwidth, jitter, digital filters, processing power etc

Don't see why, or as per my comment made about recording engineers preferring to work in DSD why do they bother? - it makes no sense to have to work with bigger files, more cpu power, bigger download storage costs etc I simply do not believe it's because they consider they can massively increase their revenue (wow $5 more... )

 

[John Deas]

I think you'll find most here disagree with this statement. DSD and PCM can sound different, but only when poorly implemented. Done properly, there will be no audible difference.

Sorry I just cannot agree and not based on my personal preference. If there is genuinely no difference then please explain why DSD is used when as I've pointed out it's a pain in the arse - sorry, unwieldy to work with - or do the exponents of it simply enjoy the hassle and losing money?

 

[Miska]

mixing out to analog or mixing prior to capture in analog

Quite many use analog consoles for mixing also PCM. That's why you still have lot of analog consoles being manufactured.
Like the AMS Neve 88RS, SSL AWS and the Rupert Neve 5088.

There are also news sometimes, about it, like for example one local mastering studio installing SSL AWS:
https://www.audiomediainternational...astering-studio-installs-ssl-aws-948-console/

Try streaming it over WiFi, like I did, and you'll quickly find out that PCM 192/24 is way superior

At least I don't have a problem streaming DSD512 over WiFi. And it is not so different in the end. For example if you send from HQPlayer to a NAA 192k PCM stream it consumes 6.144 Mbps per channel. A bit less than 5.6448 Mbps of DSD128. Likewise, 352.8k PCM consumes same amount of bandwidth as DSD256 and 705.6k PCM same amount of bandwidth as DSD512...

 

[andreasmaaan]

Sorry I just cannot agree and not based on my personal preference. If there is genuinely no difference then please explain why DSD is used when as I've pointed out it's a pain in the arse - sorry, unwieldy to work with - or do the exponents of it simply enjoy the hassle and losing money?

I don’t have enough information to comment on the beliefs or motivations of people in the industry, but restricting my opinion to technical grounds, either format is easily capable of performing beyond the limits of what humans are capable of hearing.

Your personal preference may be based on suboptimal performance of devices you’ve listened to, psychological factors (which we are all prey to), slight level differences between devices or the same device replaying different formats (which we tend to perceive as sound quality differences) or, most likely, differences in mastering decisions by the engineers who created the recordings you’ve listened to.