PCM vs. DSD

[Mr Peabody]

Interesting article and be sure to read the comments on the page this link takes you to. The actual article is written by Charlie Hansen of Ayre. Also there are tracks you can sample in both PCM & DSD for your own comparison.
Ayre's PCM and DSD Comparison | AudioStream (http://www.audiostream.com/content/ayres-pcm-dsd-comparison)

[Feanor]

Interesting article and be sure to read the comments on the page this link takes you to. The actual article is written by Charlie Hansen of Ayre. Also there are tracks you can sample in both PCM & DSD for your own comparison.
Ayre's PCM and DSD Comparison | AudioStream (http://www.audiostream.com/content/ayres-pcm-dsd-comparison)
Objective comparison of DSC vs. PCM has always favored PCM on account of DSD's noise level climbs quite high at high frequencies. My layman's surmise is that DSD and SACD were always mainly Sony hype and largely based on Sony's desire to create a copy-protected and proprietary format, (which is the perennial Sony thing).

Personally I doubt I could hear a difference between, say PCM 24/88.2 or 24/96 versus DSD64, if only because I'm pretty deaf at high frequencies, but I will not be going out of my way to acquire or accommodate DSD content.

Unless my recollection is quite wrong, (possible), Sir Terrence strongly favors the DXD format for recording over DSD "Wide" or "Pure". DXD is a PCM format at 24 bits, 352.8 kHz.

[melgross]

Objective comparison of DSC vs. PCM has always favored PCM on account of DSD's noise level climbs quite high at high frequencies. My layman's surmise is that DSD and SACD were always mainly Sony hype and largely based on Sony's desire to create a copy-protected and proprietary format, (which is the perennial Sony thing).

Personally I doubt I could hear a difference between, say PCM 24/88.2 or 24/96 versus DSD64, if only because I'm pretty deaf at high frequencies, but I will not be going out of my way to acquire or accommodate DSD content.

Unless my recollection is quite wrong, (possible), Sir Terrence strongly favors the DXD format for recording over DSD "Wide" or "Pure". DXD is a PCM format at 24 bits, 352.8 kHz.

I'm not fond of DSD myself. Its one of the oldest formats, and I can't find anything, either from a technical standpoint, or from an audio one, that impresses me.

But then, I don't like these high bandwidth formats anyway. I've never done any testing where anyone has been able to pick one out from the other.

[Sir Terrence the Terrible]

Objective comparison of DSC vs. PCM has always favored PCM on account of DSD's noise level climbs quite high at high frequencies. My layman's surmise is that DSD and SACD were always mainly Sony hype and largely based on Sony's desire to create a copy-protected and proprietary format, (which is the perennial Sony thing).

Feanor, noise shaping has made the high frequency noise problem irrelevant. If I push the noise above what humans can actually hear, then the argument becomes academic. If you can't hear it, you can't hear it PERIOD!


Personally I doubt I could hear a difference between, say PCM 24/88.2 or 24/96 versus DSD64, if only because I'm pretty deaf at high frequencies, but I will not be going out of my way to acquire or accommodate DSD content.

If this is the case, then you are not really interested in high resolution audio. DSD today is quite different than DSD of yesterday. 1bit DSD of yesterday required noise shaping to keep noise out of the audible band, and us unable to hear it. DSD today uses 8 bit word samples at a very high sample rate, so the need for filtering and shaping the noise is lessened if needed at all. You now have DSD64, DSD128, DSD256, and the latest DSD512. The last three formats of DSD render the noise argument irrelevant, and the use of noise shaping unnecessary.


Unless my recollection is quite wrong, (possible), Sir Terrence strongly favors the DXD format for recording over DSD "Wide" or "Pure". DXD is a PCM format at 24 bits, 352.8 kHz.

You are right, I do favor DXD and use it anytime I record anything. However I use a 48bit 384khz sample rate(192khz doubled) algorithm rather than standard 24bit 352.4khz sample rate(more suited towards the CD sample rate). You can losslessly resample 384khz DXD to any format and sample rate.

Here is why I think the Ayre Acoustics test have a major problem.

1.
Therefore the official Sony specification (the “Scarlet Book”) specifies a third-order low-pass filter starting at 50 kHz, and the actual usable frequency response of SACD doesn’t extend much beyond 30 kHz.

Wrong and non specific. The earlier storage version of DSD had this problem, but let's face it - it was a storage medium, not a playback one. DSD64 had bandwidth to 50khz, DSD128 beyond 100khz, DSD256 and 512 beyond 200khz.


To minimize this problem in professional gear, most DSD recordings today are made at double the rate of the DSD used on SACDs. This was modulated at 64x the CD frequency so is often called DSD-64, while the professional equipment running at double this rate is often called DSD-128. While this reduces some problems it introduces others, such as doubling of the file size and of course the download times and storage space required.

The doubling of the file size argument is at best a very minor problem Storage costs are so cheap now, storage issues are a non issue. Since most DSD128 content is two channel, download issues are no worse than 24/192khz PCM. Their argument is not sound quality related, and should be discarded from their argument.


Since DSD is a one-bit format it is literally impossible to perform any signal manipulation at all — even a fade-out. So to perform recording in the modern methods where signals are mixed, EQ’d faded, reverb added, et cetera, all of the DSD signals must first be transcoded into PCM (or analog) signals, then the signal processing applied, and finally re-modulating the signal back into DSD, adding another layer of high-frequency noise.

Again, this is early DSD, much has changed since them. I have used a Pyramid workstation that had DSD post tools built in, so this comment is outdated by almost 6 years. That is how long Pyramid has been building workstations with DSD editing and mixing tools.

What makes their arguments and comments equally interesting is that no one has done a blind test(necessary when evaluating audio format) that has come to the conclusion that high resolution PCM sounds better than DSD. They do sound "different", but one cannot be conclusively judged as better over the other. None of the "test" material is DSD or PCM based, but originally recorded in analog. If you are going to make this comparison, they have to do like I did and create two recording signal paths with identical information going to the DSD encoder, and to the PCM encoder. The test signals must be digitally recorded, not encodings of older analog masters with almost no high frequency information.

Considering the beginning of the article begins with this;


I think its fair to say that Charlie Hansen, Ayre's Founder and Designer, is not exactly a fan of DSD.

When you start off with a biased opinion, then coming to a neutral fair conclusion is virtually impossible.

[melgross]

Quite frankly, I'd like to see a double blind test where anyone can tell the difference between these formats with real consistency. I haven't seen that happen yet. Almost every "test" have told listeners what they will be hearing first. When I confront the individual, and mention that they should tell people what they're listening to first, I'm told that I'm right, and that they didn't think of that.

But I wonder. I believe they did think of that, and dismissed it.

[Feanor]

Feanor, noise shaping has made the high frequency noise problem irrelevant. If I push the noise above what humans can actually hear, then the argument becomes academic. If you can't hear it, you can't hear it PERIOD!


Personally I doubt I could hear a difference between, say PCM 24/88.2 or 24/96 versus DSD64, if only because I'm pretty deaf at high frequencies, but I will not be going out of my way to acquire or accommodate DSD content.

If this is the case, then you are not really interested in high resolution audio. DSD today is quite different than DSD of yesterday. 1bit DSD of yesterday required noise shaping to keep noise out of the audible band, and us unable to hear it. DSD today uses 8 bit word samples at a very high sample rate, so the need for filtering and shaping the noise is lessened if needed at all. You now have DSD64, DSD128, DSD256, and the latest DSD512. The last three formats of DSD render the noise argument irrelevant, and the use of noise shaping unnecessary.
.

Well, Sir T, maybe it's just me and I'll concede I'm dear above 10kHz, but I don't think I could hear a difference. But please don't construe that I'm not interested in whether there is a difference, nor that I won't want to hear the difference if there were one.

Most of the music I listen, i.e. versions of classical works, just aren't available in hi-rez. Further, I've always found the recording and mastering have more to do with the overall quality of the recording than the distribution medium, (excluding 8-track, cassette, and MP3 of course). Consequently when I hear of some new medium, I say, What?! Please, not another one -- just do the current ones right!

[Sir Terrence the Terrible]

Quite frankly, I'd like to see a double blind test where anyone can tell the difference between these formats with real consistency. I haven't seen that happen yet. Almost every "test" have told listeners what they will be hearing first. When I confront the individual, and mention that they should tell people what they're listening to first, I'm told that I'm right, and that they didn't think of that.

But I wonder. I believe they did think of that, and dismissed it.

I am going to offer a counterpoint assessment of your comment. Many audio engineers like myself have always done double blind(level matched) comparisons of each audio format we have invested in, but don't want to get muddied up in a comparison argument. We(yes me) don't really publish, because that is not our end goal. The amount of money we invest in audio technology is a lot higher than what the average consumer invest in their system, so our scrutiny of these formats is a lot higher than the consumer.

Those of us who have done these comparisons(that would meet AES standards)realize that there are too many variables in these test to not be very specific about how we set them up. I have personally switched between a live feed, a DXD feed, a DSD512 feed, and a PCM 24/192khz feed, and I could easily detect a difference(not better or worse but difference) between each in my mixing studio. My goal was not to shape public opinion, but to evaluate which format to invest my money in. In the end I was able to purchase software that did DXD, all forms of DSD, and redbook(16/44.1khz) PCM up to 32/192khz. I don't like limitations and like being able to provide a wide variety of delivery formats to my clients.

Listening test destined for AES scrutiny are very difficult to set up, and not cheap to do. The parameters of the test must be very tightly defined, and without variables that can be exploited into doubts. Not very many folks will conduct these kinds of test for that reason.

[Sir Terrence the Terrible]

.

Well, Sir T, maybe it's just me and I'll concede I'm dear above 10kHz, but I don't think I could hear a difference. But please don't construe that I'm not interested in whether there is a difference, nor that I won't want to hear the difference if there were one.

Most of the music I listen, i.e. versions of classical works, just aren't available in hi-rez. Further, I've always found the recording and mastering have more to do with the overall quality of the recording than the distribution medium, (excluding 8-track, cassette, and MP3 of course). Consequently when I hear of some new medium, I say, What?! Please, not another one -- just do the current ones right!

Bill, I don't care how well you mix and master, the delivery format will always be apart of what we judge to be quality.

I can mix and master to perfection(or some reasonable facsimile thereof), but if I only have Dolby Digital at 448kbps as a delivery system(far from neutral and uncolored), then my mix will always be defined by that level of low resolution, coloration, lack of body, impact, and spatial resolution.

Remember, a delivery system is a chain defined by the entire path to your ears. That goes from the microphones, through the mixing board, through the encoding process(defining resolution) through your A chain(sources) to your speakers. If anything in that chain degrades the sound(you won't know this without a reference), then it will change what you should ultimately hear.

[Resident Loser]

Hmmmm....You're 'bout the only name I recognize...Pax Vobiscum
jimHJJ (Ain't you lucky)

[Sir Terrence the Terrible]

Bill,
Out of my vast collection of SACD, and multichannel music files(also available in two channel as well), most all of them are classical music.

The main genre of music that has benefitted from downloads has been classical music. It is everywhere, you just have to look.

[E-Stat]

Interesting article and be sure to read the comments on the page this link takes you to.
At least the conversation of late has progressed to examining the relative merits of various hi-rez formats.

There are still some, however, like Monty who just don't get it. :)

<object height="315" width="560">


<embed src="//www.youtube.com/v/d7kJdFGH-WI?hl=en_US&version=3" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" height="315" width="560"></object>