SACD fundamentally flawed?

[WVK]

I found the following statement on "rec.audio.high-end" It was posted 12/16/2003 by a Phillips engineer. For informational purposes only>
WVK

"SACD is indeed fundamentally flawed. Using 1-bit as a conversion method can be a valid choice when the analog circuit does not have performance higher than the 1-bit signal. To use this as a data format, thus binding everyone to the noise and distortion limits, is quite another thing...

SACD is a typically Japannish invention in that it is a solution to a
nonexistent problem (decimation-interpolation), which in turn creates some
very real problems left for real engineers to solve. Some examples:
1. Splicing (editing) two DSD signals together creates a "click", even if
both represent silence.
2. Any processing (except delay results in a longer word length. Getting
back to 1-bit requires another stage of deltasigma modulation. Sony dreamt
of a new signal processing paradigm operating entirely in DSD. It was not to be - they even officially admit it now. Any quantisation mixes the signal
with quantisation noise. They can no longer be separated. This is not much
of a problem at 24 bits. At 1 bit however... well...
3. The accumulated noise from previous conversions reduces the deltasigma modulator's headroom. After 5 conversions (e.g. level control, eq, mixing, fader etc), the modulator already overloads at silence.
4. DSD is not distortion-free.
5. The signal bandwidth and the noise zone overlap. In a correctly designed
converter, the signal occupies the "clean zone" only, thus allowing the
noise to be filtered away. With DSD, the noise zone starts at 20kHz but the
signal bandwidth extends -by Sony's definition- to 100kHz. The SNR over
100kHz is only 30dB. Many amplifiers produce audible distortions when
presented with this noise (hence the switchable filter on many SACDs).

It was "invented" when someone took a CS5390 chip, wired the 1-bit test
outputs straight to a D/A converter and liked what he heard. Thus, the
standard was fixed at 1-bit/64fs which happened to be the internal operating parameters of this particular chip.

This chip is now long obsolete. Current ADCs operate at rates of 128fs and
over, at 4 bits or more.

[Paul K]

But...does it sound good to you?

[thomh]

Originally posted by WVK
I found the following statement on "rec.audio.high-end" It was posted 12/16/2003 by a Phillips engineer. For informational purposes only>
WVK

Not by any engineer. It's Bruno Putzeys, Chief Engineer at Phillips Laboratories.

_________
Thom

[thomh]

With regards to comparisons between the formats, Bruno Putzeys also wrote:


SACD vs CD: some improvement in "air" and stereo width.
DVD-A vs SACD: there are no true comparisons available. Put mildly, the
record companies are "not in a hurry" to produce a pair of discs that would
allow for a shootout. Technically, the two formats require different A/D
converters to get the most of them. Two different converters are bound to
sound different. Otherwise, ie. if digital conversion is used, the
comparison will always be skewed one way or the other, however minimally. My
position is that DVD-A is technically better by far. Either the sonic
difference will be zip, or it will be to SACD's disadvantage (specifically
when the noise drives your amplifiers nuts).




__________
Thom

[Jamie Tate]

Everybody was complaining about PCM until DSD came along and now it seems people are nostalgic for PCM.

DSD is basically PCM. PCM has been using 1 bit converters with high oversampling for years. Then the signal is, to put it in simple terms, dumbed down and made shorter to fit into the PCM world. All DSD does is keep the signal in this high oversampled state.

I don't care how many graphs and charts and nay sayers post stuff on the internet about the distortion inherent in DSD recordings. I've heard it and non of us (including some well respected golden ears) could tell the DSD signal from the live mic feed. All of us could hear the high res PCM signal.

Do you know how much processing analog tapes have? There's a lot of EQ on the way in and more on the way out. It adds audible distortion, changes the tonality of the recordings, etc, etc.

DSD is being considered the new analog without all the problems. The professional industry is eagerly anticipating this new format so we can make better sounding recordings. PCM is antiquated now. We've reached the point of diminishing returns. Let's move on and exploit DSD until we soak every ounce of fidelity from it.

End of rant.

[Jamie Tate]

SACD vs CD: some improvement in "air" and stereo width.
DVD-A vs SACD: there are no true comparisons available. Put mildly, the
record companies are "not in a hurry" to produce a pair of discs that would
allow for a shootout. Technically, the two formats require different A/D
converters to get the most of them. Two different converters are bound to
sound different. Otherwise, ie. if digital conversion is used, the
comparison will always be skewed one way or the other, however minimally. My
position is that DVD-A is technically better by far. Either the sonic
difference will be zip, or it will be to SACD's disadvantage (specifically
when the noise drives your amplifiers nuts).

Who is this guy and why are we believing all this?

I could not disagree more with this quote. It's obvious this is based on something other than a proper listening test.

This is pointless. I urge you to do some real listening test of your own... compare real DSD to high res PCM. Judge for yourself.

Time will tell.

[-=Rudy=-]

Jamie! My hero! :love:

I'll be serious now, and honest: SACD is not fatiguing to listen to like CD is. DVD-A? I don't own enough to draw a conclusion. Once I navigated DVD-A's stupid menu and found the 2-channel mode, the Fleetwood Mac "Rumours" disc didn't sound so bad. I do like the higher-rate PCM, and prefer it to the CD. The Steely Dan and Donald Fagen discs just sound overly sterile and clinical to me, and I don't enjoy them as much as I used to. (Give me "Aja" or anything earlier.) But I think that's the way they were mixed and engineered...can't really blame the format. But with "Nightfly" recorded at 48khz/24bit, it still has what I might call a slight bit of "hardness".

Believe what you want about the formats though. Anyone can find a test that will "break" DSD and pass PCM with flying colors, and vice versa. The tests mean nothing IMHO...what DOES matter to me is what comes out of my speakers. And I like what I'm hearing. :) If it's flawed at some ultrasonic level that I can't hear and my electronics don't see it either, well...*shrug*...I'm not losing sleep over it.

[thomh]

Originally posted by yesman


Who is this guy and why are we believing all this?

I could not disagree more with this quote. It's obvious this is based on something other than a proper listening test.

This is pointless. I urge you to do some real listening test of your own... compare real DSD to high res PCM. Judge for yourself.

Time will tell.

As I said, he is Chief Engineer of Philips Digital Systems Labs

Here is another quote from him taken from rec.audio.hi-end


As far as I know, all comparisons between DSD and PCM have been flawed in one way or another. Even in players where the same DAC chip handles both formats, the converter is run at higher speed using a different modulation scheme. Comparisons between both formats, for the better or the worse, get severely skewed.

I've recently built a converter that converts PCM and DSD using the same signal path. That is, the PCM first gets converted to a signal per DSD specs and this signal is then converted to analog. In this way the only difference between the two modes is that in PCM mode, the audio content is band-limited to 20kHz (plus possibly some minor aliases, see "halfband") and the quantisation noise floor in the audio band is at the 16 bit level, whereas the DSD -3dB bandwidth is approximately 70kHz and the quantisation noise floor in the audio band is at approximately 20 bit level. The outband noise is pretty equal in both cases (ie shooting up fearsomely after 20kHz).

It affords a good comparison between low and hi rez, as the difference lies precisely in bandwidth and noise floor.

Well, I daresay I'm a bit underwhelmed by the difference. OK, it is quite noticeable and if I get the choice, I prefer the high-res playback, but to proclaim DSD or 192/24 "da bomb" is IMHO an extreme case of hyperbole.

Apart from this, I'd like to attract attention to the fact that the majority of SACDs actually contain no more than the original digital master (48kHz or even 44.1kHz) converted to DSD.



_________
Thom

[Parkertown]

Originally posted by WVK
I found the following statement on "rec.audio.high-end" It was posted 12/16/2003 by a Phillips engineer. For informational purposes only>
WVK


1. Splicing (editing) two DSD signals together creates a "click", even if
both represent silence.
2. Any processing (except delay results in a longer word length. Getting
back to 1-bit requires another stage of deltasigma modulation. Sony dreamt
of a new signal processing paradigm operating entirely in DSD. It was not to be - they even officially admit it now. Any quantisation mixes the signal
with quantisation noise. They can no longer be separated. This is not much
of a problem at 24 bits. At 1 bit however... well...
3. The accumulated noise from previous conversions reduces the deltasigma modulator's headroom. After 5 conversions (e.g. level control, eq, mixing, fader etc), the modulator already overloads at silence.


Is it just me, or do these points seem like he's complaining about the fact that he can't process, eq, level control, etc the master tape to HIS liking in DSD? He sounds like one of those guys who really likes to alter the sound.

[poweragemk]

Philips is one of the partners in the SACD format, aren't they? Why is an employee of theirs sabotaging their project? Anyway, isn't this guy an electronics engineer? Aren't we making an assumption, perhaps false, that he knows/gives a crap about music?

Aside from that, in the last quote block, the whole "aside from that, most SACDs are made from the PCM master" part (aside from utilizing an argumentative strategy that I will dub 'left turn,' designed to minimize argument and distract by moving along to another point of truth before letting anyone else speak) underscores the fact that SACD has not been properly exploited by the record companies. They're generally not using CD or DVD-A's full capability, either. What's new? Thank god for Audio Fidelity and labels like them!

Just MHO as usual...:D

[Jamie Tate]

Originally posted by poweragemk
Philips is one of the partners in the SACD format, aren't they? Why is an employee of theirs sabotaging their project?

What? Everybody hates the company they work for at one time or another.

[thomh]

Originally posted by Parkertown


Is it just me, or do these points seem like he's complaining about the fact that he can't process, eq, level control, etc the master tape to HIS liking in DSD? He sounds like one of those guys who really likes to alter the sound.

He is not a *mastering* engineer.

________
Thom

[Grant]

Originally posted by Parkertown


Is it just me, or do these points seem like he's complaining about the fact that he can't process, eq, level control, etc the master tape to HIS liking in DSD? He sounds like one of those guys who really likes to alter the sound.

I think you hit it on the head. :sigh:

[poweragemk]

Originally posted by yesman


What? Everybody hates the company they work for at one time or another.

Yeah, but so publicly? In many companies, this is grounds for dismissal...guess Philips is a bit more liberal than my employer, heh.

[WVK]

Originally posted by yesman


What? Everybody hates the company they work for at one time or another.

From 12/17/2003:

>would you care to elaborate on the implications of a
>> high-ranking Philips engineer taking this stance?

It's no longer a secret. I can do this.

(Before it was out in the open, I used to be quite vocal on this issue as
well though. The result would be my colleagues going "Shhhhh!", .followed by a burst of laughter from everyone.)

From quite early on, Philips' position on SACD has been to emphasize on the multichannel capability (6 full-bandwidth channels available on a
full-length album), not so much on the DSD scheme.



WVK

[thomh]

Originally posted by yesman


What? Everybody hates the company they work for at one time or another.

He has, of course, been asked the same thing on rec.audio.high-end. Here are his comments:



> Would you care to elaborate on the implications of a
> high-ranking Philips engineer taking this stance?

It's no longer a secret. I can do this.

(
Before it was out in the open, I used to be quite vocal on this issue as
well though. The result would be my colleagues going "Shhhhh!", followed by
a burst of laughter from everyone.
)

From quite early on, Philips' position on SACD has been to emphasize on the
multichannel capability (6 full-bandwidth channels available on a
full-length album), not so much on the DSD scheme.




__________
Thom

[Richard Feirstein]

DSD is not one bit! Don't care who says otherwise. It is Delta Sigma Modulation.

The fact is that nothing is perfect. But the SACD and DVD-A audio formats have pushed every other step in production to new levels. (Much like the introduction of Dolby-A, many years ago).

Richard.

[Grant]

Originally posted by poweragemk


Yeah, but so publicly? In many companies, this is grounds for dismissal...guess Philips is a bit more liberal than my employer, heh.

We don't know anything about this guy. I don't think he could really say this stuff publically in his own name without being fired. This could be some bozo making this up, he could be a disgruntled worker, or someone impersonating the chief engineer, or someone who works with Meridian. Who knows?

[Michael St. Clair]

PCM is no dinosaur as long as DACs and ADCs are improving. It is not right to say that DSD is better than PCM (or vice versa), because you aren't listening to formats, you are listening to current hardware.

Many controlled 'tests' comparing the two were made years ago on early hardware. (I'm talking listening tests, blind or otherwise)

[fjhuerta]

If SACD is fundamentally flawed, I can't wait to listen to it done right. Because SACD is absolutely fantastic as it is, IMHO.

That's what counts, right? :)

[poweragemk]

It's all in the mastering, after all...I bet I could come up with a format made of pressed cow dung and a good mastering engineer would master fantastically on it. (they do build houses from that stuff, ya know...:D )

[thomh]

Originally posted by Grant


We don't know anything about this guy. I don't think he could really say this stuff publically in his own name without being fired. This could be some bozo making this up, he could be a disgruntled worker, or someone impersonating the chief engineer, or someone who works with Meridian. Who knows?

Ah, such sceptics. The guy is for real and the chief engineer at Phillips. Look up his name on google and check some of his AES papers.

I emailed him and extended an invitation to join the discussions here. He said he would if time allows.

_________
Thom

[GabeG]

Much of this has been discussed many, many times before. Without commenting on the sound, people need to understand:

DSD is multibit, but the delivery system of SACD is one bit. Simple as that.

[Sckott]

Whatever you do, take with what you read on Usenet with a grain of salt, always. :)

[Alan T]

Sounds like both high-rez formats have technical problems.

Nobody will talk about the effects of watermarking on DVD-A, which is not a problem on the SACD format from what I understand.

But from a practical standpoint there really isn't a universal player under a grand that solves all the problems associated with both formats, i.e. bass management, true DSD converters, compatibility.

And it seems that the record companies are not going to be putting mass amounts of software because of sluggish sales so the consumer with tastes throughout the musical spectrum will have to wait a long time for enough software to justify purchase.

And quite frankly I've been bitterly disappointed with the CD layers of the Rolling Stones, Elton John, and Bob Dylan SACDs I've heard - too much friggin' high end/treble boost.

But with the sluggish economy and people dumping their Rolling Stones CD's, I was able to replace my copy of a West German Aftermath for under $20.00.