Warner DVD-A not full resolution (1 Viewer)

[Tony Casler]

On several of my Warner DVD-A releases, the music is not full resolution DVD-Audio. Disturbed's Believe is 48kH/24 bit surround, and merely 44.1 kHx/24 bit in stereo, not much better than CD. Linkin Park's Reanimation is mastered at the same rate. Metallica is in 96KHz/24 bit in both Stereo and Surround. The Eagler's Hotel California is the only one I own that uses 96/24 surround and 192/24 stereo. Has Warner isssued any statements about these discs? Is it legal for them to deceptively label the discs as Advanced Resolution DVD-A when they are not?

 

[Phil A]

Tony, there is no std. for what constitutes 'Advanced Resolution.' On some of the later Warner discs I have noted that they have listed the resolution on the back of the case (even though I hate the in-between size) which I think is a good step forward. I have a bigger gripe about 2 discs I bought that say 'Advanced Resolution' stereo when there is no stereo track. Warner as well has at least priced their discs realistically vs. other DVD-A software. If I am buying a release that has been out sometimes I check http://www.greatgig.com/quad/dvd-a-list.html for the resolution.

 

[Felix Martinez]

On several of my Warner DVD-A releases, the music is not full resolution DVD-Audio. Disturbed's Believe is 48kH/24 bit surround, and merely 44.1 kHx/24 bit in stereo, not much better than CD. Linkin Park's Reanimation is mastered at the same rate. Metallica is in 96KHz/24 bit in both Stereo and Surround. The Eagler's Hotel California is the only one I own that uses 96/24 surround and 192/24 stereo. Has Warner isssued any statements about these discs? Is it legal for them to deceptively label the discs as Advanced Resolution DVD-A when they are not?

Not sure about the recording format of the titles you mentioned, but as an example, Donald Fagen's The Nightfly - one of the benchmark audiophile recordings of the past 20+ years - was recorded at 16-bit, 50kHz back in 1982 on the venerable 32 track 3M machine. 16-bit PCM digital recording continued into the early 1990s, so as these make their way to DVD-A (or SACD), don't expect 96 or 192kHz. Or if they're upsampled, don't expect to get something from nothing.

Likewise - and of course this is hotly debated - there are analog recordings that frankly don't exceed (and some don't approach) the dynamic or frequency range of 16-bit 48kHz. But this is neither here nor there.

Bottom line is - WB is not saving any money mastering at a lower sampling rate. The decision is probably based on the source material and/or the mixing engineer's preferences.

Just use your ears. Don't just go by numbers and think that's the only determination of quality or value.

Cheers,

 

[John Kotches]

Felix,

I agree... the practice of "disguising" for lack of a better word the recording heritage is IMO a bad one.

Of course, with SACD, you have no choice -- it's a digital sausage grinder. It doesn't matter what you put into it, the output always looks the same.

Many SACD fans think this is a selling point, since you're always getting disc at "full resolution". Of course if the recording is originally a PCM recording, at whatever rate, you're getting something entirely different than full resolution.

Whether this is beneficial or not, is a matter of personal preference.

Regards,

 

[Lee Scoggins]

It doesn't matter what you put into it, the output always looks the same.

Then PCM is a digital "sausage grinder" as well, since any source is output in the same target resolution that the engineer intends.

John, that's just the mathematical nature of digital encoding, you get results at whatever resolution you choose. The good thing about SACD is you always get 2.8Mhz resolution no matter what you start with and that's a good thing. There is a standard resolution.

If the DVDA camp really wants to improve itself in this area, they should list original recording resolution on the case for every release. People should know if they are getting 48k or 96k.

 

[John Kotches]

Lee,

God forbid, someone should post anything remotely critical of DSD. I'll explain this again, in more detail since you seem to be missing my point.

If the original recording is 48K, 96K, 192K or even 44.1K fs, DVD-Audio has the capability of delivering the updated recording at its initial sampling rates. To get DSD one is required to do a transcode to the original data that represents the recording.

With DSD there is NO choice for the artist or engineer to preserve the original fs if the recording was initially captured via PCM. DVD-Audio supports fs from 44.1 to 192K in multiples of 44.1K and 48K and the choice to change fs is entirely in the hands of the artist or engineer.

I can do a table if that makes the point even clearer.

Regards,

 

[Felix Martinez]

If the DVDA camp really wants to improve itself in this area, they should list original recording resolution on the case for every release. People should know if they are getting 48k or 96k.

Both formats can do a better job at communicating this info!

Cheers,

 

[cwhite]

I think a better or more recent example of Felix's question is Fragile by YES. On the back, it is listed as 192/24 stereo but my player displays it as 96/24 stereo.

 

[Lee Scoggins]

Both formats can do a better job at communicating this info!

I agree with this, but at least with Super Audio you know you are always getting 2.8mhz. It would be helpful to know of the original source used, but then again the record business has always struggled with leaving out vital recording information, sometimes to lower user confusion, other times to falsely market like the SPARS code.

 

[Brian-W]

Lee, get him!

 

[Rich Malloy]

I think proper labeling is very important, but even more to the point: is there an audible difference between 24/48, 24/96, 24/192?

That is, is going with a lower resolution a judicious choice in a given circumstance, or are hi-res stereo recordings being given short-shrift on DVD-A, either because they don't care to offer the higher resolution or simply to make space for videos, menus, "extras", etc.?

I've noticed this trend, and asked a bit about it in some other threads. I must say, it's yet one more concern that's kept me from plunging headlong into DVD-A. It seems to underscore the perception that it's not a format for those who care first and foremost for best sound quality.

 

[John Kotches]

Lee,

When you're going to quote and attempt to display incredulity, it helps your case to quote the original correctly.

I said

Once in DSD, the technology allows for downconversion to any format based on a clean integer multiple of PCM. Sony and Philips made this work as well as possible. There is no case against SACD on this point...

Basic mathematics aside, you mean other than the substantial limitations imposed by the format itself. They're quite substantial, but thanks to digital technology, they're also preserved for all time.

Now let's go back to the basic mathematics.

DSD is in itself an inadequate sampling rate to do a "clean integer multiple" as you claim. I'm no mathematician, but when I learned math, integers contained no fractional portions. 2,822,400 / 96000 comes out to 29.4. 2,822,400 / 48000 comes out to 58.8.

I agree that 44.1k and 88.2k are integer multiples (64 and 32 respectively).

In SuperBitMapping Direct, a 5x oversample is required to map back to all these sampling frequencies anyway, otherwise insufficient data points exist.

Then we have to take either of 147 (96K), 160 (88.2K), 294 (48K) or 320 (44.1K) data points and map them back to the appropriate bit depths.

So for 48k and 96k (the most prevalent sampling frequencies in use for production work), we don't have clean integer math to anything but 21-bit sampling depth. It's a little nicer for 44.1k and 88.2k, since both 16 and 20-bit depths are integer multiples, but common practice is 24-bit in production work these days, which still isn't an integer multiple.

To goto the highest sampling rates available (176.4k and 192k) we're talking about an additional 2x oversampling to get enough data points for a reduction.

Once we do the additional oversampling the math is identical to 88.2k and 96k.

So in short, to "downsample" from DSDs 1-bit/64fs to 44.1-96K we need 5x the data stored, and to get from 1-bit/64fs to 176.4 and 192K we need 10x the data stored.

We haven't even begun to discuss this topic:
Are the results from an SBM --> 24/96K translation identical to native 24/96K capture.

If SBMs version of 24/96K isn't equal to that of the 24/96K encoder given identical input it isn't a suitable replacement for 24/96K storage since the results vary from native capture.

This is something that is best discussed in a different thread.

Regards,

 

[John Kotches]

Rich,

If you ask someone like Mark Waldrep, he would say that until about a year ago the 192K ADCs for capture weren't as good as the 24/96K ADCs were. As such, the choice was between a better sounding 24/96K ADC or a worse sounding 24/192K ADC.

As far as audible differences between the 3 rates you suggested, without hearing all of them in comparison to the original it's a bit tough to say, isn't it? Certainly the difference between 24/96 and 24/192K will be smaller than the difference between 24/48K and 24/96K.

Menus are negligible, if memory serves me correctly 25MB is the biggest I've seen on a DVD-Audio disc which is less than 1% of the total storage space. So that isn't a significant factor. I can't cross check this, as my PC with the DVD-ROM drive is down for the moment.

Videos? THis is the MTV generation, so videos are a strong selling point. Again, without my DVD-ROM drive I can't check here. This might chew up some substantial space, but it would need to be fairly high bit rate (7-8Mb/sec average) to chew up more than a few hundred MB. While this is a more substantial fraction (maybe as much as 10% of the total space) it wouldn't be a limiting factor IMO, since the disc can easily extend out to 2 layers and ~9.0GB of available space.

IMO, there are two reasons for the small amount of 192K material:

1) Limited number of tools available for 192K work, though this will grow rapidly over the next few years.

2) Limited availability of quality ADC at 24/192K work, which again will grow rapidly over the next few years.

Regards,

 

[Philip Hamm]

I can't remember a thread that veered off topic worse than this.

 

[Lee Scoggins]

2,822,400 / 96000 comes out to 29.4. 2,822,400 / 48000 comes out to 58.8.

As I will show later, this is not the right calculation to use so this analysis is not appropriate.

 

[Lee Scoggins]

I can't remember a thread that veered off topic worse than this.

You must have a short memory.

This thread is about resolution quality specifically.

 

[Philip Hamm]

IMO it's ludicrous to suggest that cross-conversion between formats in one way benefits the end result and in another way is detrimental to it.

 

[Lee Scoggins]

From Sony Super Audio site:

Downconverting Direct Stream Digital from 1-bit/64fs to 16-bit/1fs is not theoretically difficult. Every DAT recorder and A/D converter has a circuit that does much the same thing. But we needed to downconvert DSD in such a way as to retain the maximum possible signal quality in the 16-bit world. The answer was to completely filter and noise shape the DSD signal in a single stage. Thus, interstage requantizing errors would be eliminated. Aliasing would be minimized. And ripple would be suppressed. Sony designed a super-power one-stage FIR digital filter/noise shaper with an amazing 32,639 taps. This is Sony’s real-time Super Bit Mapping Direct™ processor.

Sony knew that conversion between the formats would be critical given the established nature of PCM...

But what about the math?

The sampling is designed for conversion with simple integer multiplies and divides. As Sony nicely puts it "In all cases, the conversions in sync are performed with simpler integer multiplies and divides."

You start with Direct Stream Digital (DSD) and get 2.8224 Mhz:

To get to 44.1 khz, multiple by 1/64.
To get to 88.2 khz (one of my favorites), multiply by 1/32.

For others Sony multiplies the signal by 5 to get to 14.112 Mhz, and then to get to:

32 khz, multiply by 1/441
48 khz, multiply by 1/294
96 khz, multiply by 1/147

The idea is that Sony/Philips recognized that there were many existing PCM and analog masters AND the existing redbook standard could not be ignored, so a solution was needed to create this backward compatability in simple terms.

The process looks like this (I wish I could draw a block diagram)
1. You have a PCM or analog master that gets converted to DSD.
2. This is then stored for archival purposes.
3. This "universal" source can then be used for studio work (instrumental dubs, for instance) at higher sampling rates (even for redbook CDs), a Super Audio release, a DVD & DVDA release, and simple redbook distribution. In fact a lot of newer redbooks from Sony are from DSD mastering. Generally they sound good as redbook goes. And this will increase as the pro tools proliferate just as John correctly suggested for 192k.
4. Some other benefits of this archival format is that it can support a higher number of tracks and indices (255 vs. 99) and longer playing time (109 versus 74 minutes).

 

[Lee Scoggins]

IMO it's ludicrous to suggest that cross-conversion between formats in one way benefits the end result and in another way is detrimental to it.

Okay Philip, please present any evidence you can that shows that jitter is not measurably reduced by the PCM to DSD process.

 

[John Kotches]

Lee,

In the future, read through all the math, instead of the pieces you want to. I pointed out the exact integer ratios a bit later in the post.

In the end, SBM results in a NON-INTEGER conversion when going from SBM to any PCM sampling rate with a 24-bit sampling depth which is current practice in the recording industry.

As soon as you can get:
147, 160, 294 and 320 to divide evenly by 24 please let me know.

When last I checked, they were 6.125, 6.666, 12.25 and 13.333 respectively.

My math is not in error. Perhaps you need to adjust your verbage instead.

Also, just like Sony, you discuss theoretical improvements of SACD over redbook. No one is discussing redbook in this thread but you. This discussion is about DVD-Audio.


Regards,