SACD fundamentally flawed?

I was offering the article's conclusion, not a representative quote. As far as I'm concerned, the conclusion is where the author puts his money down, so to speak, and the statement that DSD was about copy protection looked iffy to me.

The part you quote equates noise-shaping in DSD with negative feedback as employed in amplifier design to reduce distortion. I'm very skeptical about this connection and the author's argument doesn't work through the argument fully in any case, so I'm left to conclude that he brings up "negative feedback" to go "boo!" more than anything else.

And why exactly are you praising the Almighty for Steve's SACD if DSD is no miracle?

That's put facetiously, but a serious question is behind it: if DSD is fundamentally flawed, why does that Zombies disc sound so wonderful?

 

Every reference to the Nyquist theorem I looked for (including my text books) say that it should be equal or greater than twice the highest component in the analogue signal... here's another link I found (the previous one said basically the same).

http://cnx.rice.edu/content/m10791/latest/

EDIT: Look for that part on the web site... I can't seem to be able to display mathematical formulas on this forum

Still, the example you are referring to has been proven conceptually wrong, so this point is moot...

 

I have a question regarding DSD technology and the ability to shape the quantization noise to rise outside of our audio limits. Is this bad for our hearing? Can long term listening to SACDs do damage to the human hearing by exposeing us to an unusual amount of high frequency sound waves? This high quantization noise can't be good for us, is it? Does anybody here know exactly what type of frequency range and sound level we are talking about? If the quantization noise was shaped to curve in at about 60-75khz at 0db de-emphasis, that could possibly be detremental to our hearing. But if the noise was shaped to exist at the 60-75khz region with -40db or more de-emphasis, that should be OK, right? Does anybody know what the quantization noise curve looks like.

 

I can say that looking at the output of my modified SACD player on an oscilloscope at the modifier's workshop (Ric Schultz - EVS) was a sobering experience. The ultrasonic crap is scary. It sounds fantastic, but I can understand how the presence of DSD noise in the ultrasonic range offends some from a technical standpoint.

I don't believe it bothers us - it may bother your dog.

 

The subject of the nyquest theorem is very eloquently discussed in the little book CYBERNETICS by Norbert Wiener. There is actually a whole chapter, Chapter 4 FEEDBACK AND OSCILLATION, that discusses this subject in great detail. Actually, this whole book is really a good read, and covers a whole of great ideas that deal with information theory and the storage and communication of information.

 

It was the first SA-CD I bought that made me think that it might be possible to make an SA-CD sound good enough, an SA-CD that didn't sound out of control. And the redbook sounds just great, too.

FFF

 

I never thought about that high frequency noise except from a "technical" point of view. Not to be negative, but your question is VERY interesting and deserves to be taken seriously!

 

FabFourFan,

Try some of the Channel Classics discs by Jared Sacks. Some of the most lifelike classical music I have ever heard from a recording.

My view on the HF noise factor is that it's just not practical in that it is never or rarely heard on a good recording/mastering. Remember, a lot of us got into Super Audio in the first place due to PCM's inability to capture a realistic cymbal sound among other things.

I like to record in 24/88.2 but I'm still aware of the limitations.

 

Yes, but "never or rarely heard" is not the point imho. The question is: is it harmful to listen to very high frequencies at loud volumes, even if one cannot hear that noise? In other words: do we run the risk of getting deaf within 10 years only by listening to cranked up SACDs? Maybe there are studies about high frequency noise effects on health?

 

This dialog is great, and there are a lot of important issues that are being discussed here, but I feel that the issue of exposure to ultrasonic noise is an issue that warrants greater consideration. The owner's manual on my Sony DVP-NS755V indicates that it has an output frequency response of 2hz to 100khz, with a rated deemphasis of -3db@50khz. The frequency response of my Phase Linear 400b power amplifier is rated to be DC to 100,000 hz, +/- 1db across the entire range. Thus, the quantization audio signal is well within the scope of audio components bandwidth, although, of course, the loudspeaker's ability to reproduce this sound is seriously limited, the noise is still present. Considering all this, is anybody aware of any medical research, or other such scientific research, that has looked into the possible effects of high frequency quantization noise as a by-product of DSD's 1 bit conversion process and how this noise may effect human hearing. Tinnitus is a big concern of mine.

 

That's the key. I'd guess that there is very little ultrasonic energy in your room. What does your dog think?

 

You know, Gardo, I am a grumpy mid-life sceptic, and when I read things like this I get worried. Close to analog excellence?

Why on earth should we strive to emulate analog? Note I said emulate and not approach as even good ol' Redbook's got a noise floor lower than analog tape. Witness the JVC XRCDs, where the tape noise from the 30ips two-track analog masters is clearly audible as a nicely smooth background hiss. And you know what a lower noise floor gives you, MORE RESOLUTION!

_________
Thom

 

Re grumpy mid-life scepticism: take a number, Thom.

Perhaps I should have said "what they believe to be the excellence of analog." With regard to the analog vs. digital debate, I confess I'm an agnostic. I enjoy and have worked with both. And I don't believe either is flawless.

I'm not sure a lower noise floor and more resolution are the same thing, myself. Steve's example of the reverb on the opening snare hits of "Fortunate Son" is pretty persuasive to me. The noise floor on the redbook layer is the same (more or less) as the noise floor on the DSD layer, but I hear more "there" there on the DSD layer. Something besides the noise floor would have to account for that.

 

Agree. And according to all respected research, adult human hearing falls off somewhere between 16 - 20kHz.

Reproduced how? Most popular large capsule studio microphones rolloff under 20kHz, don't they?

But for the sake of argument, let us examine a 17kHz waveform which falls nicely in our range of human hearing. A digital reproduction of this will have <0.01% distortion while the analog reproduction will have >10% distortion. And being that both systems are band limited to below 30kHz, the original 17kHz waveform is by definition a pure sine wave.

By considering the first three components as the minimum necessary to reproduce a recognizable square wave, you are asking your speakers to reproduce out to 119kHz with little or no phase shift. The first overtone is at 51kHz. Got some speakers that will reproduce that with little or no phase shift?

How are you going to "sub-conciously sense" something when it is not even coming through correctly or at all?

That is the second time that is mentioned in this thread.

In a broken system, maybe. Otherwise, prove it. There are some rare occasions, like when doing a lot of post processing and NOT AD conversion, where FIR pre-ringing may be heard. But by the time you sample at 60kHz you will NOT hear it. 88.2kHz is plenty margin. So relax.

Tweaker also mentioned that impulse response somehow improves by going out to 192kHz or DSD? This is simply NOT true! The impulse response width IS the time domain representation of the Nyquist bandwidth.

And what study do you base this humble opinion on?

And which format do you propose we use for that. DSD? Well, let's see now, the noise begins to rise rapidly at 22kHz, so if you are one of those freaks of nature and can hear 30kHz or even higher, the noise will simply bury the signal. At least DVD-V/A has got a smooth and even noise floor up there.

Remember if the ear can't hear it then the brain cannot do anything with it.

_________
Thom

 

To all in the "debate" portion of today's program, please read this thread here before continuing.

I dunno, mate. My tummy sure feels my subwoofer.

 

A while back I read an article and some scientists had found that even though our ears cannot hear above 20khz that the surrounding bone mass was capable of detecting frequencies up to around 50khz. I sure wish I would have bookmarked it so I could post a link. I'll try searching for it and post if/when I find it.

 

Yes, I rememberr this article as well. I'm not sure why, but sometimes my emotions, or is it my physical body, pick up certain thimgs in music (bad sound mainly) that my ears aren't detecting right away, like heavy mid-range compression. I refer to it as audio cued hyper-tension.

 

No need to. You are probably referring to work from the Kanagawa Institute. Reportedly they had representatives at AES shows laughing in the aisles. Nevermind that it was funded and (apparently) conducted by Pioneer among others who at the time was introducing a 96kHz DAT, it was proved to ignore large amounts of contradictory data.

At the moment I am not aware of any studies which has gotten past published in a peer-reviewed journal. You have to remember that if such effects really existed, there would be no shortage of papers on it as it is important information to many medical and scientific fields.

___________
Thom

 

I've got a question, I've got a question!!

I was waiting for someone else to bring this up but after 13 pages and two bleary eyeballs, I gave up.

1) According to what I've read here, particularly the extensive descriptions from Mr. Putzeys, if DSD and 192/24 are so close in transparancy, resolution, etc. why do we need DSD at all then???

And 192/24 PCM doesn't seem to have all these problems with ultrasonic noise that DSD has that must be dealt with.

2) The "sound" of DSD: I was wondering about all that carefully designed noise-shaping that DSD needs to sound right--can this noise-shaping be manipulated enough to cause DSD to sound like certain other formats?

Thanks.

[T]

 

I think we need DSD because it is a very efficient, cost-effective means of delivering analog-like sound quality to the end listener. A well-designed $400 SACD player can compete with hi-res PCM consumer gear costing 10 to 20 times as much. (fire retardant suit on - even though the Gort team keeps the heat to a minimum here.)

 

Yeah, you're probably right there.

Well, I enjoy both. But when it comes to fidelity to the source, digital has the upper hand.

Dynamic range (lower noise floor) and resolution are completely unbreakable. This is borne out by more than a half century of work on the subject.

Let us say we have a bandwidth of 20kHz with a dynamic range of 80dB, so assuming perfect conversion, a 96kHz 24bit sampling cannot capture any more information about that signal than a 44.1 kHz bit system because there isn't any more information in the signal to capture.

And again, in a sampled system, the sampling theorem MUST hold. So unless we want to prove the sampling theorem wrong, we must look elsewhere.

__________
Thom

 

I think that's fair enough.

Thom - a personal question. You may have answered this in the previous 13 pages of the thread (ahem) but, do you hear a difference between 2-channel SACD and CD?

jeff

 

Also - how does one interpret the following oscilloscope graphs of a 10khz square wave, as seen as the input, 44.1khz, and DSD? Are "we" saying that it doesn't matter if everything is represented as a sine wave in the upper audible frequencies?

http://www.smr-home-theatre.org/surround2002/technology/page_07.shtml

input:


44.1 CD:


DSD/SACD:

 

Hey - I finaly understand something in this thread! Thanx, HZ!!!

 

Actually, that trace is used by the DVD-A camp to find fault with DSD. The blurry haze along the line is ostensibly the UHF noise in the signal. It's hard to correleate images like these with sound quality, as interesting as they may be. There's also this set published by a DSD workstation maker in support of the format:

http://www.merging.com/2002/html/pyradsd.htm