High Resolution Audio Comparison (1 Viewer)

[Lee Scoggins]

another btw - lee, i think you've done a commendable job here...especially handling john's emails

Thanks Ted.
Keith,
It sure was nice of Mike Broadman to take a swing at the jazz content of the two formats!

 

[KeithH]

Lee said:

at the jazz content of the two formats!

NP: John Coltrane Blue Train DAD

 

[Lee Scoggins]

http://www.audioasylum.com/forums/hi...es/109365.html
The above story helps create additional momentum and capacity for Super Audio given Viva Magnetics stature in CD manufacture.
This may also create some additional grease for more major titles to be released.

 

[KeithH]

Lee, the Audio Asylum link said:

Earlier today, Viva Magnetics, one of the world's largest CD production firms announced that they plan to start production of Hybrid SACDs at their plants in Canada and Hong Kong in January 2003.
Annual production will reach a capacity of 10 Million Hybrid SACDs per year with the initial 3 production lines being installed.

Gee, I hope Sony is paying attention.

 

[Terry St]

Terry, there is no full paper but you might try these links for more on DSD:
http://www.daisy-laser.com/tech3.htm
The above link will also explain why Sony left our certain things on the DSD diagram. The bottom line is that no perfect brickwall filter exists and leaving it out by doing one bit encoding is very useful for improved sonics.

Tried that and the following links too:
http://www.opticaldisc-systems.com/2...Feb/SACD66.htm
http://www.stereophile.com/showarchives.cgi?180:7
http://ken-gilbert.com/techstuff/dig...mat_primer.htm
http://www.stereophile.com/showarchives.cgi?374
http://www.simaudio.com/pdf/Upsampling.pdf
http://www.opus1.com/~violist/help/nyquist.html
I'm still pretty new to this topic, but let me try to answer my own questions in case anyone else was interested.
True PCM ADC/DAC's have been extinct for over a decade!
I was formerly under the false impression that modern ADC/DAC's convert directly to and from a PCM stream. Namely, I thought a 44.1 KHz PCM stream was produced by measuring the amplitude of the waveform 44.1K times per second. This hasn't actually been the case since the 80's!
Pretty much all modern ADC/DAC's are delta sigma ADC/DAC's. A 1-bit delta sigma ADC takes an analogue signal and compares a snapshot of how it is to how it was one sample interval previously. It then outputs a 1 in the amplitude was higher or a 0 otherwise. This is effectively a negative feedback loop. While these samples are grossly imprecise compared even to a 16-bit sample, if your sample rate is high enough you can still represent the same waveform. It's sort of like making a giant poster out of a few thousand polaroids.
You might be asking yourself, "Gee... Isn't this a lot more complicated? Negative Feedback? I thought simple was good! Why the heck did they do this?" The answer is that manufacturing a device that can perform grossly inaccurate measurements several million times per second is a lot easier than manufacturing a device that can perform highly accurate measurements less than a hundred thousand times per second. Delta Sigma ADC/DAC's dominate today because they are probably every bit as good as an equivalent PCM ADC/DAC but are a ton easier and cheaper to build.
What are the Decimators, Interpolators, and Delta Sigma Modulators in the PCM stream?
So how do we get to PCM from a sigma delta data stream? That is what the Decimator does. It takes a large number of 1-bit samples and converts them to a smaller number of 16-bit, 24-bit, etc. words. The interpolator and Delta Sigma Modulator do the opposite.
Wherefore art thou SACD?
A SACD stores the delta sigma stream directly off of a Delta Sigma ADC. In an ideal world where the music goes straight from the singer's lips, through a microphone, to the ADC and then to the master CD you have indeed cut out some digital processing. Likewise, playing the SACD back on your home system is simpler. You just feed that Delta Sigma stream to a DAC and off it goes. No interpolation or modulation required!
The Nyquist theorem tells us that we cannot accurately represent a frequency that is more than one half the frequency of the sample rate. Ergo, the requirement of filtering frequencies above 22.05 KHz. for 44.1 KHz. PCM CD's. This filter has to be steep enough to meet this requirement without affecting lower frequencies which may be audible.
PCM streams have a much slower sample rate than DSD. 44.1 KHz in particular comes perilously close to filtering out audible frequencies and in many cases probably does due to the difficulties of producing "brick wall" filters.
The DVD-A Camp
In theory, 24-bit 96 KHz. PCM does actually contain more data than 1-bit 2.8224 MHz. DSD. PCM is a more compact data form and easier to losslessly compress. (Both DVD-A and SACD employ lossless compression algorithms) If all else were equal, DVD-A's would be the better format just for this.
As for the problems with the 22.05 KHz brick wall filter, if you up the sample rate you can place the filter higher. The further it is from audible frequency ranges the less steep the filter has to be. A 96 KHz sample rate allows for frequencies of up to 48 KHz, which is pretty close to the 50 KHz filter that must be placed on the output of SACD systems as a result of their use of noise shaping. The filter argument seems to be pretty much obsolete when considering a 96 KHz DVD-A. (Although lower sample rates are supported and would still have this problem)
PCM proponents claim interpolation and decimation can be comletely transparent in a good design. These are digital operations and do not take place in an analogue stream where every component you add to the path degrades the signal. This claim requires proof I have not yet seen, but likewise, SACD proponents have not proven that the opposite is true. More study here is cleary required.
One of the stereophile articles above briefly suggests that a DSD stream might be easier to watermark than a PCM stream, and that this is a driving force behind the push for DSD in SACD's. It's an interesting suggestion, but one which needs more support. Again, I remain unconvinced.
Putting it all together.
I feel a bit better informed now, but have a lot more research to do before I can reach any good conclusions. At present I remain on the fence. Neither format appears clearly superior. You can knock the good ol' red book standard all you want, but CD's are still where the music is.

 

[John Kotches]

Terry,

You're going to get a lot of pushback on this one:

True PCM ADC/DAC's have been extinct for over a decade!

The Burr-Brown division of TI still manufactures ladder type PCM DACs such as the very highly regarded PCM1738. This DAC interfaces with a DSD DAC, which provides a neat two-chip BB solution for PCM and SACD.

Most of their D/As are Delta/Sigma, because (as you've mentioned) they're cheaper to implement, but they have a handful of ladder PCM DACs in the catalog as well.

Regards,

 

[Lee Scoggins]

Terry and John,
I am glad to see we are debating the technical specifics.
In reference to the amount of data, I feel that the higher sampling rate is very important to capturing transients of the music better based on extensive auditions of both format's reference works.
You can push the brickwall filter up the food chain more but I still hear limitations on high frequency events like reproducing cymbals and items above 10khz. I feel DSD just seems more effortless at that point.
Nevertheless, DVDA is a very good format and will likely see even further refinement that may overcome these issues or at least make them inaudible.
DSD is still very, very young and we are likely to see more dithering added to further lower any ultrasonic noise issues (which I still claim you cannot hear). Derk Reefman has demonstrated how dither can lower the noise floor to -180db. That's seem good enough to me for now.

In addition, the common signal processing tools are readily available with PCM.

True, but this is changing rapidly. There are pure DSD workstations in use now and Sadie is on a road show promoting the ease of use and accuracy of its new DSD product line.

 

[Walt O]

I think its worth noting that the technical disadvantages that DSD has are not going to last forever. We can only do 8 channel recording now, with 16 on the way. It's only going to get better.

Someone earlier mention that Aerosmith's Just Push Play album on SACD is only marginally better than the CD version. While this is true, keep in mind that JPP was recorded on a Pro Tools Digital Workstation in PCM, it is not an analog or DSD source. However, most of the stuff on their Oh yeah Greatest Hits SACD is from an analog source, and sounds INSANELY good! I am a big fan of Aerosmith, and I thought I was hearing these songs for the first time again. Stuff from the 1970's has no business sounding this good, especially when most releases today don't. Now that the Stones are out on SACD, maybe we can get all the Aerosmith stuff out on the format?

As a reader of Widescreen Review, I think they should have a DVD Audio v. SACD test. I thought the catchphrase of WSR was "The best that it can be." Gary championed DTS over DD for years, and SACD is a MUCH bigger jump over DTS than DTS was over DD at the time. IT SOUNDS REAL! I must confess my ignorance over DVD Audio as I have not heard it. I would however like to read a detailed comparison in WSR soon.

Walt O

 

[John Kotches]

Lee,

You said this:

Look at the Sony site and examine the 10khz waveform. PCM is not easily able to create a similar form.

I agree completely. 16 bit/44.1 kHz isn't capable of capturing a 10k waveform as well as DSD. That's what the fs was for the diagram as depicted. Of course it's easy to overlook this fact when you're looking through the world with DSD colored glasses.

Why wouldn't Sony compare to the best available PCM technology which is 24bit/96k and 24bit/192K?

BTW, that 10k wave @ 24bit/96K has similar numbers of bits (2.3m vs. 2.8m) to describe the wave over a period of time as DSD, and PCM @ 24/192K (4.6m) carries nearly twice the number of bits.



Regards,

 

[mike_decock]

BTW, that 10k wave @ 24bit/96K has similar numbers of bits (2.3m vs. 2.8m) to describe the wave over a period of time as DSD, and PCM @ 24/192K (4.6m) carries nearly twice the number of bits.

To me, this weighs heavily in favor of PCM. The more data you capture, the more accurately you can recreate the original wave.


-Mike...

 

[KeithH]

Walt said:

Someone earlier mention that Aerosmith's Just Push Play album on SACD is only marginally better than the CD version. While this is true, keep in mind that JPP was recorded on a Pro Tools Digital Workstation in PCM, it is not an analog or DSD source. However, most of the stuff on their Oh yeah Greatest Hits SACD is from an analog source, and sounds INSANELY good! I am a big fan of Aerosmith, and I thought I was hearing these songs for the first time again. Stuff from the 1970's has no business sounding this good, especially when most releases today don't. Now that the Stones are out on SACD, maybe we can get all the Aerosmith stuff out on the format?

I made the point about Just Push Play, and I understand that this is a crappy PCM recording. This is a case where the master is the limiting factor in sound quality of the disc. You could probably have the best mixing engineer in the world work with Just Push Play, and the SACD would sound lousy.

One concern I have about SACD or DVD-Audio making it with the masses is that a lot of mainstream recordings, the ones the masses are buying up, are crappy. The high-resolution formats may offer little or no sonic advantage over the CD. That, of course, is when the backers of SACD and DVD-Audio should especially market surround-sound capability, and in the case of DVD-Audio, video capability, very hard. Actually, they should do that regardless of the sound quality of the recording, in my opinion. I still contend that neither format will make it with the masses on sound-quality grounds. The masses love MP3, so they obviously don't care about sound quality.

 

[John Kotches]

Keith said this:

I made the point about Just Push Play, and I understand that this is a crappy PCM recording. This is a case where the master is the limiting factor in sound quality of the disc. You could probably have the best mixing engineer in the world work with Just Push Play, and the SACD would sound lousy.

The master is always going to be the limiting factor on a recording. Crappy masters yield crappy delivered recordings to the consumer, regardless of initial capture method (Analog, PCM or DSD).

You can look at a number of recordings captured as PCM and resampled to DSD that sound extremely good. So it isn't that it's a crappy PCM recording, it's that it's a crappy recording period.

Regards,

 

[John Kotches]

mike said:

Although I may not fully understand DSD, this doesn't make sense. If an individual sample contains no meaningful information about the wave, then NO individual sample contains meaningful information about the wave. If this were true, you could never get a sound wave out of the binary data encoded in DSD.

Remember the wizard of Oz, when he's discovered? "Pay no attention to the man behind the curtain!".....

In this case, behind the curtain is the accumulator and the filter. It's the combination of the samples, the accumulator and the filter which derive meaninful information. Without these, you'd be listening to a frequency modulated square wave, which would sound worse than even the crappy Aerosmith recording being referenced in the posts above.

Regards,

 

[Lee Scoggins]

As a reader of Widescreen Review, I think they should have a DVD Audio v. SACD test. I thought the catchphrase of WSR was "The best that it can be." Gary championed DTS over DD for years, and SACD is a MUCH bigger jump over DTS than DTS was over DD at the time. IT SOUNDS REAL! I must confess my ignorance over DVD Audio as I have not heard it. I would however like to read a detailed comparison in WSR soon.

I agree Walt, but I don't understand why John will not raise the issue with his editors.
I've tried to be as friendly about this as possible.

 

[Lee Scoggins]

One concern I have about SACD or DVD-Audio making it with the masses is that a lot of mainstream recordings, the ones the masses are buying up, are crappy. The high-resolution formats may offer little or no sonic advantage over the CD.

Good insight Keith. There are two ways to look at this:

1. The enhanced realism of high rez may in fact force engineers to be more careful in the studio which is good for us music fans.

2. The enhaced resolution also may permit more digital enhancer technologies which will allow engineers to further cover up mistakes and create less natural sound.

The best medicine is to support quality labels that produce music you like. I don't like all the content on audiophile labels, for instance, but I like most of the music on Chesky and Rereference and the new Telarc recordings.

 

[John Geelan]

You can look at a number of recordings captured as PCM and resampled to DSD that sound extremely good. So it isn't that it's a crappy PCM recording, it's that it's a crappy recording period.

One reason why Aerosmith's recent cd's have sounded like crap even though the music is pretty good is the fact that they master the music to sound good in a car/boombox.

I read a interview about their recording process and they would run out to a car (not sure what kind or what system in the car) and see if it sounded good in that environment.

They could give a rat's ass about how it sounds in an audiophile environment (the world we inhabit).

So consumers like us here at HTF will always find fault with Musicians who record like this to the master tape.

It can never sound good on SACD or DVDA.

Hopefully their older recordings like Toys In The Attic and Get Your Wings will benefit in SACD because they were recorded the old way, nice warm analog!

 

[Justin Lane]

That, of course, is when the backers of SACD and DVD-Audio should especially market surround-sound capability

This is why I feel DVD-A will ultimately be the winner with the masses. Most people do not have high end stereo systems to truly take advantage of DVD-A and SACD, so a Hi-res stereo mix that costs more then the regular CD will be of no interest to them. DVD already is associated with DD and DTS surround, giving DVD-A an instant association advantage in the multi-channel realm even if all they listen to ultimatley is the low res DD and DTS mixes on a DVD-A. Including a multi-channel mix on almost all DVD-A discs is keeping DVD-A a step ahead of SACD for wide spread acceptance (possibly the reason Sony is beginning to re-release stereo SACD in multi-channel?).

There has been alot of talk about the number of releases between SACD and DVD-A but I wonder what the true sales numbers have been. Because of the obscurity of many SACD titles, I would not be surprised if DVD-A, with fewer total releases, has actually sold more units of software because you can actually buy titles at mainstream retailers at decent prices. Case in point are the CCR discs which should have been a very big release for SACD, but instead you are almost limited to buying them online because of their high retail prices ($25), and the fact that you will never find these in a Best Buy or Circuit City. Neil Young Harvest on the other hand can widely be found at Best Buy, Circuit City and even Border's (which near me had them out in the front of the store on one of their Holiday picks displays).

J

 

[KeithH]

Justin,
I don't know what the actual sales figures are for DVD-Audio discs and SACDs, but this has been a hot topic of debate over on the Hi-Rez Highway of Audio Asylum. I think few, if anyone, there knows what is going on, but there have been a lot of posts on the subject, so you may want to do a search. Anyway, I wonder about the sales figures too, though I don't obsess over them like many on Audio Asylum do. They talk about it like their life depends on it. Those folks are always looking for some incidental nugget to give them security that their format is succeeding and the other one is failing.
One thing about the sales is for sure. All Sony SACDs and the few Universal single-layer SACDs are being purchased by people with SACD players. No one knows what hybrids SACDs and DVD-Audio discs are being used for, though I suspect that most DVD-Audio discs are being used for playback of DVD-Video material. DVD-Video players still greatly outnumber DVD-Audio players. It wouldn't surprise me if the majority of hybrid SACDs are being used by people with SACD players, however. The exception is the Stones discs, which are probably being purchased primarily as CDs. With the exception of the Stones discs, hybrid discs are sold primarily through specialty retailers, and with standard CDs of these titles more readily available, I would think that most people buying hybrids are doing so for the SACD layers. This is all speculation on my part, of course.
Lee,
Thanks. In the end, I buy the music I like. I prefer that it sound good too, but "you can't always get what you want." As an example, I recently bought the new Boston Corporate America CD. I didn't expect it to offer audiophile-quality sound, but I love Boston. The disc doesn't sound very good (there is low-level buzzing during parts of my favorite song), but I like the music overall. On the flipside of this discussion, I'm not too keen on buying music just because it sounds good.
John Kotches,
Absolutely. The master is the limiting factor. What is a shame is when a good master exists and then the mixing engineer messes it up. Anyway, there are a lot of poor master tapes out there, and as the old cliche goes, you can't polish a turd.
John Geelan said:

One reason why Aerosmith's recent cd's have sounded like crap even though the music is pretty good is the fact that they master the music to sound good in a car/boombox.

That is so pathetic. I guess that means overblown bass. It's a shame that there are recording and mixing engineers who seem to have no standards and who take no pride in their work.

 

[John Kotches]

Lee,

I'm not missing the point at all.

The sample is a delta, I've known that for a long time. What meaningful information does that carry about the wave though, only that it goes up or down. There's not even a valid slope associated with it, unless one considers the slope to be +infinity for a 1 and -infinity for a 0.

The fact that you have to derive waveform information with the addition of an accumulator/comparator (on the input side) further verifies that individual samples don't contain meaningful information. If they did, you'd not require an accumulator. After all, it is the summation (Sigma) of the change (Delta) samples that carries the wave information, not the individual samples.

I think it hugely ironic that if SADIE is a pure DSD solution (ie, that it maintains 1-bit 64fs throughout all operations) that it didn't come from either Sony or Philips. None of the publically available documentation at the SADIE website has any information with respect to the internal processing of the DSD bitstream.

Suffice it to say I don't find DSD to be a simpler process, just different.

Regards,

 

[Lee Scoggins]

Suffice it to say I don't find DSD to be a simpler process, just different.

Bunk. It is definitely a simpler recording chain process. There are fewer circuits to pass through in both DSD encryption and decryption.