Is DSD mastering THAT good?
Oct 18, 2005 at 4:18 PM Thread Starter Post #1 of 10

sionghchan

500+ Head-Fier
Joined
Jun 1, 2005
Posts
565
Likes
14
Hi all,

I've just been reading here and in the sa-cd.net forums that direct recording to DSD and then mastering it on DSD is the best. However, thinking about it more, it seems to be a little counter intuitive. Even in DSD, it is still not analogue, i.e. digitisation happens even if it is at a much higher frequency. Don't you all think that if it is recorded onto pristine analogue tapes is the best?

Analogue contains the ENTIRE sound wave. So, if you keep the entire sound wave, should there be better digitasation process in the future, we can then redigitise these analogue copies rather than using an approximation (which what DSD is) which has lost some data...and if in the future, the digitisation process has even higher resolution, we will have to find a way to upsample DSD data...

confused.gif


Am I missing something here? Please educate me here!

[MODs: I'm not sure if this is the right forum to ask this question...if it isn't, please feel free to move it.]

Thanks!
SH
 
Oct 18, 2005 at 4:33 PM Post #2 of 10
I am not the person to instruct or teach you the in's and out's of DSD. However, my ears have been for me the greatest test to pass. I am a strong believer in the merits of Direct Stream Digital. And that for both when it is applied to SACD or just Redbook. Telarc recordings can be found with DSD in both SACD and non SACD or Redbook. Far be it from me to knit-pick it's internal structure. If it works, it works, and I will let it go at that and just enjoy it being there. And to thies ears it WORKS WELL.
icon10.gif
 
Oct 19, 2005 at 6:56 AM Post #3 of 10
What if the noise floor and resolution on the best analogue equipment are worse than on the best DSD gear? You can't get infinitly precise, and even though analogue is analogue, you don't get every last detail, some of the information is just too fine to be recorded on current gear.

And if DSD can provide lower noise, distortion, and finer detail than analogue, then wouldn't it be better?
 
Oct 19, 2005 at 8:38 AM Post #4 of 10
Your question is really about DSD recording, not DSD mastering (the final step in making a CD).

The truth is, there is no perfect recording medium. When you record something on analog tape, even though it's still analog, there are limits to how accurately that recording reflects the original signal (noise floor, frequency response, impulse response, etc.). High bitrate PCM (192kHz, 24-bit) is generally superior to analog recording on virtually any measurable scale, but quality analog recording is not that bad either, and an argument could conceivably be made that it "sounds better".

DSD is more slippery, because it is inherently inferior to high bitrate PCM, and it's not clear at all that DSD is better than analog tape recording. When DSD first was proposed by Sony and Philips, there was an important paper published by Lipschitz and Vanderkooy that argued that DSD was, mathematically speaking, a step back from analog tape. The researchers actually argued that it would be unfortunate if DSD became the archival method of choice for old recordings. There was a slew of followup academic papers, some on each side of the issue, but the bottom line has always been that DSD is something of a compromise. In the real world, it's probably roughly on par with analog tape, at least for recording.

Another serious objection with DSD is that it has to undergo format conversion in order to be edited in any meaningful way. Even Sony's DSD editing consoles convert to multibit internally for most operations. Despite the marketing spin, there is no pure DSD editing path except for the simplest edits. This is another reason why DSD is not clearly better than analog tape, since both have to undergo a format conversion for realistic editing.
 
Oct 19, 2005 at 8:56 AM Post #5 of 10
The "direct DSD" SACDs are the worst sounding ones in my collection! As Wodgy mentioned, even "direct DSD" is not really direct. Objectively, analog tape and high-bitrate PCM sould be the best. Subjectively, they are the best to my ears. The best SACD I own (to my ears) is a AAD of an old classical analogue tape. Sounds great.

Since my first SACDs were all Telarc, therefore all direct DSD, I blamed SACD for the veiled sound I heard. As it turned out, the wooly and veiled sound I was hearing was a result of the so-called 'direct DSD' conversion.

Bottom line, do not use DSD as a recording format. High-bit PCM and analog, are, in reality, more direct and sound better. Basically, wait for the absolute last minute to convert to DSD.

EDIT: Newer "direct DSD" recordings sound better than the first ones, however, I still prefer hearing a Redbook CD recorded/edited in PCM than an
SACD recorded/edited in DSD. Seriously.

DSD recordings, IMO, are horrible.
 
Oct 19, 2005 at 1:22 PM Post #6 of 10
IMO, there's much more to our perception of recording quality than the medium in which it's recorded (quality of microphones, recording consoles, mixing consoles, mastering consoles, and all the varying skill levels of all the engineers involved in making the recording all along the way).

One would need to have two identical recording set-ups to capture the performance through the same mics at the same time, one DSD, one analogue, use the same people to mix and master them, and then compare those end products to see if one was "superior" to the other. But even that is flawed due the different equipment needed for each medium. Then, you'd need two identical playback systems to play back the results of each, but that's kind of impossible too, because you'd have to play back the analogue master on a tape machine, and the DSD master on a SACD player.

I think it's incredibly complicated, and basing a judgement of one recording technology based on what you hear from 2 or 3 SACDs or analogue tapes is flawed, especially when you don't have the exact same performance/recording in the other format to compare against head-to-head to see if there really *are* audible differences...
 
Oct 19, 2005 at 2:01 PM Post #7 of 10
I have to agree here. DSD should NOT be used for recording or mastering. While there are proponents of doing everything in analog, I believe that recording and mastering should be done in very high bitrate PCM: 176.4kHz, 24-bit for CD intent.

Analog tapes will deteriorate over time with humidity, temperature, and the like. Digital does not. This is a moot point though as there is no longer a single manufacturer of analog master tape in existance.

DSD is not a mastering format. The main tennents of Digital Signal Processing dictate that it is actually impossible to do any mastering work on a DSD stream. So things like changing gain, filtering, mixing, etc are all impossibe with a native DSD signal. It must first be converted to PCM, then mastered, and then reconverted back down to PCM.

Another comparison between ananlog and digital production. In a digital system filters and gains can be implemented with 100% precision. With no inherient loss in audio quality. That's not to say it can't be messed up, but it is much much easier and precise to process in the digital domain than running the raw signal through a bunch of capacitors, resistors, inductors, opamps, tubes, etc with varying tolerances, age, quality, etc. In digital, once the signals pass the A/D converters you're completely free to do what you will with the music will no fears of degredation.

I won't even get into the whole ADC/DAC architecture side of the debate. Maybe Glassman will chime in on that.

If you're next questions is, "Then why do we have DSD anyway?" The answer is that nobody really knows. The best excuse I've heard is that Sony's patent ran out on CD technology and they didn't want to lose that cash flow. Yes, I know SACD can sound very good, but I'm a PCM fanboy, can't you tell
tongue.gif
 
Oct 19, 2005 at 2:21 PM Post #8 of 10
Quote:

The main tennents of Digital Signal Processing dictate that it is actually impossible to do any mastering work on a DSD stream. So things like changing gain, filtering, mixing, etc are all impossibe with a native DSD signal.


From what I've read, this has not been true for some time, this "issue" has been resolved a while back.
 
Oct 19, 2005 at 6:53 PM Post #9 of 10
Quote:

Originally Posted by markl
From what I've read, this has not been true for some time, this "issue" has been resolved a while back.


No, it never was resolved. Sony's DSD editing consoles convert to multibit internally, as jefemeister pointed out. "Pure DSD" from recording to disc has always been a marketing construction, not a technical reality. Any nontrivial edit to DSD requires format conversion.

Quote:

Originally Posted by jefemeister
If you're next questions is, "Then why do we have DSD anyway?" The answer is that nobody really knows. The best excuse I've heard is that Sony's patent ran out on CD technology and they didn't want to lose that cash flow.


If you think back to when SACD was conceived, around 1993-94, the format made a bit of sense at the time. One bit sigma-delta modulators were all the rage on both the ADC and DAC side. There was some logic to storing the data after recording or for delivery to the consumer in this format. Of course, technology soon moved on, and one bit modulators would be improved upon and fall out of fashion, but at the time, I suppose DSD made a bit of sense. There was never a rational engineering argument for a "pure DSD" signal path for mastering, but I'm sure when the marketing people got involved, they felt the format wouldn't sell unless such a thing could be pitched to the consumer.
 
Oct 20, 2005 at 12:29 AM Post #10 of 10
Redbook is good enough if you use a good enough player.

Anyways, is this thread on the right forum?

Cheers!
580smile.gif
 

Users who are viewing this thread

Back
Top