R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [3] 4 5 6   Go Down

Author Topic: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)  (Read 22144 times)

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #30 on: September 15, 2005, 04:40:03 PM »

duplicate deleted

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #31 on: September 15, 2005, 04:48:09 PM »

blairl wrote on Thu, 15 September 2005 20:09

bobkatz wrote on Thu, 15 September 2005 08:11

If 192K proves to be the "best fit point", I suspect it will be because some lazy designers didn't do their psychoacoustic homework.


I remember Paul talking about this topic.  During the design of the Oxford console, they did some serious listening tests as they were tweaking the converters.  In the end, those involved with the listening process could not tell any difference between the source audio and the audio running through the Oxford AD /DA process.  You can't get any better than no difference from the source.  This was all done at 48k.  It's all in the filter design.



Were they all men and what were their ages? Anecdotal analysis using only men over 25 is severely flawed. Remember the population studied is the only one that determines the final generalization. Garbage in, garbage out. They may have perceived no difference, but if they didn't include the population that actually listens to music as a randomized sample (say, male and female 0 to 125 years),  then any attempt to do a factor analysis (the bare minimum for a thorough statistical analysis, IMO) will be irrelevant.

howlback

  • Full Member
  • ***
  • Offline Offline
  • Posts: 249
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #32 on: September 15, 2005, 05:54:40 PM »

JamSync wrote on Thu, 15 September 2005 16:21

howlback wrote on Thu, 15 September 2005 05:44


I also wish that MIX would really try to consult with the academics who also go to AES and present papers on this stuff instead of just bombarding engineers with propaganda.  My suspicion is that, despite the possible good intentions of Mr. Calder and yourself KK, this might be used as such propaganda.  
-K. Walker  
Yes, I thought this would bring all the assailants out of their trenches...credentials in physics, math, and statistics, I have them, so kindly stop with the assumptions that I don't...You don't have to listen to the disc or read the article if you're afraid of propaganda.
KK, I read your bio before I posted.  I in no way insinuated that you were creating propaganda (I said you had good intentions), only that your work (which is possibly a valuable bridge between the academic and the real-world) might be USED as propaganda.  This happens all the time, even to the best research.  James Boyk's work is a great example.  He measured some instruments, but some are abusing his academic work as some kind of proof that high resolution is king.  He has no data to show that.  

I am assuming that most of your AES presence is as a recording engineer, not an academic.

Relatively conclusive answers to these issues may come over time or may never come at all.  What I would like to see in MIX is a back and forth between say Stanley Lipschitz and Mike Story - something that is hard to get in the likes of the AES journal.

The reputation of MIX does not rest with your editors or publishers - it rests with people like you!  

-regards

k. walker  
Logged
 

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #33 on: September 15, 2005, 09:17:20 PM »

[quote title=howlback wrote on Thu, 15 September 2005 22:54][quote title=JamSync wrote on Thu, 15 September 2005 16:21][quote title=howlback wrote on Thu, 15 September 2005 05:44]
I am assuming that most of your AES presence is as a recording engineer, not an academic.

As somebody who has studied these issues academically, relatively conclusive answers to these issues may come over time, or may never come at all.  What I would like to see in MIX is a back and forth between say Stanley Lipschitz and Mike Story - something that is hard to get in the likes of the AES journal.
[/quoteI]

You assume incorrectly. I rarely go to the AES. I have taught college and have turned down teaching positions regularly. One does not have to be affiliated with a school to "study issues academically".

I find it amusing that you think people who write papers for the AES are somehow immune to being the progenitors of propaganda. Have you actually looked at their credits? A lot of them ARE manufacturers, or they work for manufacturers, or they consult for manufacturers or their ivory-tower research grants come from manufacturers.

It's fine if you want to see Lipschitz and Story...either suggest it to George Petersen or start your own magazine. It's a free world. But that has absolutely nothing to do with what I was saying and I really resent your tacking the word propaganda onto some critique of something that you haven't read. That's bullshit.

If you simply accept some paper written by an academic as the truth and you can't do the math to check it, then you might as well admit you're taking what they say on faith, something that leaves you prey to believing propaganda.

sdevino

  • Full Member
  • ***
  • Offline Offline
  • Posts: 153
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #34 on: September 15, 2005, 10:19:00 PM »

Here is a sample of one of the latest adc's.

http://www.cirrus.com/en/products/pro/detail/P1024.html


It is multibit sigma-delta as was pointed out to me earlier in the thread by KK or Vincent. For most of the 90's most converters were single bit sigma-delta.

There is lots of literature out there, but a multibit sigma-delta essentially trades complexity (added internal DAC in feedback loop) for a lower oversampling speed. The resulting output bit stream is still a serial bit stream which is then decimated into pcm for storage or processing in the digital domain.

So Sigm Delta modulation is alive and well and is a fundamental building block of most high performance converters built since the early 90's ( I worked on Motorola's 1st 20 bit ADC in 1989). Multibit provides higher resolution at a lower clock speed but results in single bit stream (if I read the literature and block diagrams correctly)

DSD never made sense to me, to me its always been a marketing ploy.

I agree with Bob that better filter design at 44.1 or a little above should provide a great deal of improvement.

I have also never seen any study published which showed any correlation between > 48kHz sample rates and perceived improved sound. On the other hand I attended at least one paper presented by a Japanese group at the last NYC AES which used 30 subjects and documented the inability of the group to tell the difference between 44.1k and 96k.  I am not saying its not there, all I am saying is no one has proven it yet, and current physics says it doesn't matter (nyquist).

So for me when someone does their homework and can either:
1. prove nyquist wrong

or

2. assemble a group in a controlled process and show some correlation between higher sample rate and perceived better sound quality then I will embrace the concept.

I can hear a difference on some gear between 44.1 and 96, but I cannot tell you whether it is due to the sample rate or something else.

Steve
Logged
Steve Devino

Granite Rocks Recording Studios
Studio gear design and setup

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #35 on: September 15, 2005, 10:42:07 PM »

howlback wrote on Thu, 15 September 2005 05:44

 Unless they used a custom set of converters designed along the lines of the recommendations of James Dunn, all of the tracks are going to sound completely different.


-K. Walker  


I'm not sure which reference you mean regarding "James Dunn".

However, here's one of my favorite things by Julian Dunn (RIP) who created some of my favorite tools.

http://www.nanophon.com/audio/antialia.pdf


howlback

  • Full Member
  • ***
  • Offline Offline
  • Posts: 249
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #36 on: September 16, 2005, 12:47:57 AM »

JamSync wrote on Thu, 15 September 2005 21:17

One does not have to be affiliated with a school to "study issues academically".
You are right.  One has to collect data analyze it and present it to the community.  Teaching has nothing to do with it.
JamSync wrote on Thu, 15 September 2005 21:17

I find it amusing that you think people who write papers for the AES are somehow immune to being the progenitors of propaganda. Have you actually looked at their credits? A lot of them ARE manufacturers, or they work for manufacturers, or they consult for manufacturers or their ivory-tower research grants come from manufacturers.
You are right, there is very biased research out there, particularly on this issue.
JamSync wrote on Thu, 15 September 2005 21:17

I was saying and I reallly resent your tacking the word propaganda onto some critique of something that you haven't read. That's bullshit.
I respectfully suggest that there is miscommunication between us.  At no time have I meant that you are writing propaganda.
JamSync wrote on Thu, 15 September 2005 21:17

If you simply accept some paper written by an academic as the truth and you can't do the math to check it, then you might as well admit you're taking what they say on faith, something that leaves you prey to believing propaganda.
The math isn't the problem.  Reproducing the results from listening experiments is.
JamSync wrote on Thu, 15 September 2005 21:17

I'm not sure which reference you mean regarding "James Dunn".However, here's one of my favorite things by Julian Dunn (RIP) who created some of my favorite tools. http://www.nanophon.com/audio/antialia.pdf
You are quite right. I have a problem remembering names, initials are much easier.  If we ever meet in person I don't think I would have this problem with you.  Smile  Presentations like this one are what keep me interested in this subject.

I look forward to reading your article.  I just hope that it doesn't get abused.  

-k. walker
Logged
 

blairl

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 376
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #37 on: September 16, 2005, 01:08:09 AM »

When people claim that a converter sounds better because of the higher sampling rate, you will find a substantial number of people that disagree.  Some people will dismiss altogether the notion that we need higher sampling rates to capture information that lies beyond our ability to hear or "perceive" and instead focus on the science of the ear in developing converters.  In short, some people with considerable knowledge and experience find that higher sampling rates are completely unnecessary.

There is a line of thinking that says 96k sounds better than 48k and therefore 96k is better.  Another camp says that 96k sounds better than 48k, not because of it's ability to capture super harmonic information, but because the filter design is easier to work with.  They claim that a correctly designed filter on a 48k system can sound just as good.

Have you read Paul Frindle's or Dan Lavry's thoughts on this topic?  Before you dismiss their findings you might want to read what they have to say and then ask why many people claim that their 48k converters sound better than some 96 or 192k converters.  You might start to wonder if they are on to something.
Logged

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #38 on: September 16, 2005, 01:26:57 AM »

howlback wrote on Fri, 16 September 2005 05:47

I look forward to reading your article.  I just hope that it doesn't get abused.  

-k. walker



Not a problem. Thanks for reading my bio and bringing that to my attention. I didn't write it; my ex-partner did and it's pretty obvious that it's out of date and written in a style that doesn't really match the way I will re-write it. I just figured out how to blow all the links with Dreamweaver yesterday (!), so editing and re-writing the site is now a big priority.

Sorry if I misread your intentions. I agree that reading re-hashed marketing press releases is very disheartening and cannot, should not be tolerated by the reading public.

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #39 on: September 16, 2005, 02:25:52 AM »

blairl wrote on Fri, 16 September 2005 06:08

When people claim that a converter sounds better because of the higher sampling rate, you will find a substantial number of people that disagree.  Some people will dismiss altogether the notion that we need higher sampling rates to capture information that lies beyond our ability to hear or "perceive" and instead focus on the science of the ear in developing converters.  In short, some people with considerable knowledge and experience find that higher sampling rates are completely unnecessary.

There is a line of thinking that says 96k sounds better than 48k and therefore 96k is better.  Another camp says that 96k sounds better than 48k, not because of it's ability to capture super harmonic information, but because the filter design is easier to work with.  They claim that a correctly designed filter on a 48k system can sound just as good.

Have you read Paul Frindle's or Dan Lavry's thoughts on this topic?  Before you dismiss their findings you might want to read what they have to say and then ask why many people claim that their 48k converters sound better than other manufacturer's 192k converters.  You might start to wonder if they are on to something.



I've read them both, respect them both, use the Oxford EQ all the time, etc., etc.

As for the science of the ear, I think we need to incorporate the science of perception.  We haven't even really nailed down HRTFs. How can we lay down the law about external variables, when we're still dancing around that area?

I like Prism converters, but I'd love to see a really well-designed test of
different converters using a properly selected human sample with the appropriate statistical analysis. Without some attempt at valid perceptual testing (and I don't mean the old A/B test), it's immaterial who thinks what. One would like to think that
certain external variables translate to internal perceptual ones, but without a good experimental design, it's impossible to state that with a reasonable certainty. Math and theory are nice, but give me empirical population data. THEN we can say what does and does not matter.

As a very early student in experimental psychology, I was fortunate to study under a professor who turned around the study of children's language from Chomsky's "generative grammar" to focusing instead on novel linguistic creations by children. He was able to set up test settings with some parameters determined by studying their natural verbal utterances and using them, rather than using only terms that adults decided were the proper ones to test. In other words, using something that had long been in development in linguistic studies ("grammar") turned out not to be the definitive force that others had assumed it was and the actual development of children's language, in respect to novel creation, was operating in surprisingly different ways.

In the same way, we have not really generated a set of descriptive tools that are derived from utterances of a native population that has experienced different converters, even a subset of "trained lay listeners". Just saying something is "better" is a limited and lame way to describe the variety of human aural experience.

I'm fully aware of the discussions about filters, sigma/delta design, etc. etc., but
unless you can move it over to the population that will experience the output, and I don't mean audio engineer/males aged 25 and above, then all that discussion, while amusing, interesting, etc. has very little relevance to a decision about what kind of converter will provide the end user with the most enjoyment. No one has proposed that kind of study, although it's quite common in experimental psychology to design such tests.

On the other hand, if higher frequency sampling rates make it easier to design good-sounding converters that would otherwise be more expensive in a lower frequency sampling version due to the expense of creating better filters, then ipso facto, higher frequency rate converters will sound better and sell better. Otherwise, we're talking about angels on the head of a pin.

What no one has explained to me is the reason a triangle in an orchestral piece recorded by Michael Bishop jumped out of the mix as being the only time in my life (up to that point) I had ever heard a recorded triangle that didn't make me want to grit my teeth. I knew I had to find that clarity in the transient and decay of metallic instruments, so that's the reason I switched to working with converters capable of working at the highest frequency I could afford. I'm hoping to purchase the Prism ADA-8XR that I checked out a few weeks ago as soon as they can confirm it works more or less flawlessly with Tiger. I already have an ADA-8, but I want to capture multichannel audio SFX at 192kHz...just in case y'all are wrong;-). I know one thing--it isn't going to sound *bad*.

bblackwood

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 7036
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #40 on: September 16, 2005, 05:21:08 AM »

JamSync wrote on Fri, 16 September 2005 01:25

I'm fully aware of the discussions about filters, sigma/delta design, etc. etc., but unless you can move it over to the population that will experience the output, and I don't mean audio engineer/males aged 25 and above, then all that discussion, while amusing, interesting, etc. has very little relevance to a decision about what kind of converter will provide the end user with the most enjoyment.

Excellent point, KK.
Logged
Brad Blackwood
euphonic masters

bobkatz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2926
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #41 on: September 16, 2005, 04:00:57 PM »

JamSync wrote on Fri, 16 September 2005 02:25



On the other hand, if higher frequency sampling rates make it easier to design good-sounding converters that would otherwise be more expensive in a lower frequency sampling version due to the expense of creating better filters, then ipso facto, higher frequency rate converters will sound better and sell better. Otherwise, we're talking about angels on the head of a pin.




Let's talk hypothetically. Assuming that I am right:

That a well-designed, slightly more expensive filter would make a 96 kHz converter sound identical to or better than a well-designed 192 kHz converter.

Now, look at all the outboard gear that I don't' have to buy or replace. Look at all the Weiss processors I own that "only" go out to 96 kHz. All the TC Electronic reverbs that require two engines to get 96 K performance. All the plugins that give up the ghost or require more CPU load at 96K.

Isn't it a GIVEN that we should research and go for that slightly more expensive converter, and stick with 96K, than to succumb to chucking all that gear? The consequences of standardizing at 192K just because someone didn't do his homework, or because better filters at 96K would be "too expensive" are mind-numbing.

Quote:



What no one has explained to me is the reason a triangle in an orchestral piece recorded by Michael Bishop jumped out of the mix as being the only time in my life (up to that point) I had ever heard a recorded triangle that didn't make me want to grit my teeth.




Usually it's the muted trumpets and acoustic pianos that sound wrong. It's pretty rare that I get to hear a triangle in my work.

Quote:



I knew I had to find that clarity in the transient and decay of metallic instruments, so that's the reason I switched to working with converters capable of working at the highest frequency I could afford. I'm hoping to purchase the Prism ADA-8XR that I checked out a few weeks ago as soon as they can confirm it works more or less flawlessly with Tiger. I already have an ADA-8, but I want to capture multichannel audio SFX at 192kHz...just in case y'all are wrong;-). I know one thing--it isn't going to sound *bad*.




Agreed...

BK
Logged
There are two kinds of fools,
One says-this is old and therefore good.
The other says-this is new and therefore better."

No trees were killed in the sending of this message. However a large number of
electrons were terribly inconvenienced.

JamSync

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 460
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #42 on: September 16, 2005, 05:52:39 PM »

[quote title=bobkatz wrote on Fri, 16 September 2005 21:00]
JamSync wrote on Fri, 16 September 2005 02:25


Let's talk hypothetically. Assuming that I am right:

That a well-designed, slightly more expensive filter would make a 96 kHz converter sound identical to or better than a well-designed 192 kHz converter.

Now, look at all the outboard gear that I don't' have to buy or replace. Look at all the Weiss processors I own that "only" go out to 96 kHz. All the TC Electronic reverbs that require two engines to get 96 K performance. All the plugins that give up the ghost or require more CPU load at 96K.

Isn't it a GIVEN that we should research and go for that slightly more expensive converter, and stick with 96K, than to succumb to chucking all that gear? The consequences of standardizing at 192K just because someone didn't do his homework, or because better filters at 96K would be "too expensive" are mind-numbing.

BK



Bob,

What you're saying to manufacturers is essentially: "no, you should not sell more iron because it can be done more economically (for the end user)."  And, Bob, selling iron is the way manufacturing stays in business. First it makes the thing, then it adds features, then it repackages, then it adds features, then it repackages...

Manufacturers are not in business to save you money on gear.  We may very well "standardize" on 384! And then we'll abandon PCM...

You're thinking like an end user, not a manufacturer, Manufacturers want to sell more for as much as they can.

Same reason the speaker companies are sooooo happy about 5.1!

Lower rate stuff in the ongoing studio biz will be around for a long, long time. We're not going to get past lower bit rates in several distribution media. I *did* unload my 44.1/20-bit Apogees, but I still have the 8-channel 48K Apogee 8000 working in one of my rooms. As long as I work in video and video needs 48K, I'm not trashing something that sounds OK and has interesting metering.

Staying in business in the media field means buying new stuff every so often. It's a terrible truth...let's just hope the mean time between purchases doesn't get even smaller.

howlback

  • Full Member
  • ***
  • Offline Offline
  • Posts: 249
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #43 on: September 17, 2005, 12:33:11 AM »

JamSync wrote on Fri, 16 September 2005 02:25


As for the science of the ear, I think we need to incorporate the science of perception.
Not only do we need to incorporate the science of perception but we need to incorporate a science which is ecological, which has real world validity.
JamSync wrote on Fri, 16 September 2005 02:25


We haven't even really nailed down HRTFs. How can we lay down the law about external variables, when we're still dancing around that area?
I don't really know what you mean by this.  Binarual measurements and models abound.  Perceptual testing doesn't require a perfect model of the ear or hearing system.  Ecological perceptual testing requires asking the right questions  in the right circumstances.
JamSync wrote on Fri, 16 September 2005 02:25

I like Prism converters, but I'd love to see a really well-designed test of different converters using a properly selected human sample with the appropriate statistical analysis. Without some attempt at valid perceptual testing (and I don't mean the old A/B test), it's immaterial who thinks what.
Beyond the tests done by sound recording engineers in studios and beyond the technical councils at the AES who are working on trying to establish "A/B" practices, I do know of a few people who are working on these kinds of tests.  One in particular is looking at the higher sample rate issue from more or less the perspective you seem to be describing.  

I'm interested to know what you consider to be a properly selected human sample.  Sound Quality testing of this kind has to be done with 2 groups of listeners: trained listeners (listeners who have experience with the task at hand) and expert listeners (listeners who are used to listening for such small differences).  Producing statistically significant results from the general populace is probably impossible when you are dealing with such small differences. Finding a large group of expert female listeners can also be a problem, although I think that is changing.

Also, constructing a test like this using different converters doesn't answer the high-res question.  There are still too many variables which are very difficult to hold constant.  Basically, as far as I can tell, to scientifically answer this question one MUST try to minimize the effects of filter design; this is difficult especially when using a variety of converters out of the box.
JamSync wrote on Fri, 16 September 2005 02:25

In the same way, we have not really generated a set of descriptive tools that are derived from utterances of a native population that has experienced different converters, even a subset of "trained lay listeners". Just saying something is "better" is a limited and lame way to describe the variety of human aural experience.
In fact, ONLY asking somebody to rate a particular converter as better or worse is almost pointless.  It doesn't give you any data as to WHY a particular converter is better and certainly doesn't suggest any physical measures which might be associated with that increase in quality.
JamSync wrote on Fri, 16 September 2005 02:25

No one has proposed that kind of study, although it's quite common in experimental psychology to design such tests.
This general kind of study has been proposed and carried out many times.  It is specifically called Sound Transmission Quality Analysis.  There are papers on this going back to the seventies and eighties.  Guys like: Letowski, Blauert, anybody doing speaker preference testing, etc. BUT I have seen very few studies related to the high sample-rate question that weren't pretty biased in one way or another, or that didn't incorporate only a small population.
JamSync wrote on Fri, 16 September 2005 02:25

 On the other hand, if higher frequency sampling rates make it easier to design good-sounding converters that would otherwise be more expensive in a lower frequency sampling version due to the expense of creating better filters, then ipso facto, higher frequency rate converters will sound better and sell better.
I think I'm with Bob on this one.
JamSync wrote on Fri, 16 September 2005 02:25

What no one has explained to me is the reason a triangle in an orchestral piece recorded by Michael Bishop jumped out of the mix as being the only time in my life (up to that point) I had ever heard a recorded triangle that didn't make me want to grit my teeth.
I too have had such experiences. But I'm not convinced my experiences weren't heavily biased.  Hence my agnostic stance on high-res.

Ultimately, manufacturers could in theory make all digital audio much better by simply agreeing to a matched set of standard filters for A-to-D and D-to-A.  But I don't think I will live to see this happen.  Could you imagine being told to record with a certain A-to-D because it minimized pass-band ripple on play-back?

-k. walker
Logged
 

Bob Olhsson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3968
Re: DSD (Direct Stream Digital) audio compared to PCM (Pulse Code Modulation)
« Reply #44 on: September 17, 2005, 09:13:59 AM »

This is exactly what Pacific Microsonics HDCD does.
Pages: 1 2 [3] 4 5 6   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.058 seconds with 22 queries.