Page 1 of 3

SACD "needle drops"

Posted: Mon Mar 07, 2005 7:24 am
by thomh
From this SHtv thread "needle drop" SACD

grant wrote:I recently did a couple of transfers of SACDs to PCM through the analog connections. The result was as bit of a letdown as any other redbook CD in comparison. I decided that there wasn't really a reason to do this, unless the commercial CD counterpart was really bad.

I did one transfer at 44.1/16-bit, and one at 48k/32-bit with my best dithering settings. PCM destroyed any advantage thab DSD had.

I have yet to try burning a DVD-A from vilyl needle drops.


Grant, could you tell me what equipment you used for the transfers?

Were the A/B tests level matched and blind?

What type of music did you record?

And what characteristic of DSD was "destroyed" by PCM?

Thanks.

Posted: Mon Mar 07, 2005 10:52 am
by krabapple
FWIW , I've done the same thing and guess what: they sound fine. (Pioneer DV-45a, all bass management etc turned off, front L/R channel out --> M-Audio 2496 analog in at predetermined below-clipping level --> digitized at 44kHz/32-bit -->normalized to 0dB --> dithered and downconverted to 44/16 with Audition). But of course there's no easy way to do an ABX with an SACD -->PCM transfer versus the original SACD. So I really can't say how different they sound. And neither can Grant, I bet.

For Street Fighting Man I *can* say that analog transfer of the CD layer, vs analog transfer of the SACD layer, reveals only *slight* superiority in dynamic range in the SACD layer versus CD (on the order of 0.2 dB) when the file are all normalized to 0 dB. Moreover, a digital rip of the CD layer and an analog transfer of the CD layer measured essentially the same after normalization, demonstrating that for CD at least, analog transfer need do no harm to levels and DR.

Posted: Mon Mar 07, 2005 11:06 am
by lukpac
Can you hear any differences?

Posted: Mon Mar 07, 2005 11:16 am
by krabapple
I haven't ABXed the transfers to each other, though that is easily doable. The numbers and spectral views left me confident enough that any audible difference between the files would be slight (there's much more 20-22 kHz content in the SACD transfers, of course, but it's all low-level noise, and my hearing simply doesn't extend even to 18 K these days). But to compare the 'native' SACD of anything to a PCM transfer of it, I'd have to level-match and synchronize my SACD player to the foobar2K playback of the converted file, and get someone to do the switching for me. Not gonna happen.

Posted: Mon Mar 07, 2005 11:29 am
by lukpac
krabapple wrote:But to compare the 'native' SACD of anything to a PCM transfer of it, I'd have to level-match and synchronize my SACD player to the foobar2K playback of the converted file, and get someone to do the switching for me. Not gonna happen.


Not even in the name of science?

I mean, you can't say "they won't sound different" without actually listening, can you?

Posted: Mon Mar 07, 2005 3:14 pm
by krabapple
First, I did not say they won't sound different. I said the PCM transfers I did sound fine. I also said Grant can't know how different the SACD really sounds compared to his transfer (using his usual half-assed 'I take 1/2 from science and 1/2 from belief' method)

Second, listening in the usual sense (sighted comparison) doesn't necessarily get you the right answer, for the usual reasons -- i.e., the ones that every objectivist has branded under his hair next to '666' when he sells his soul to the horned god Science.

Third, one can have good reason to say 'they *probably* won't sound different, without actually listening.

Fourth, if that probability is high (based on good reasoning), one can even predict that 'they won't sound different', without actually listening.

I mean, an elephant not only *probably* won't fly -- I predict it *won't* fly. Unless the horned god says it can.

However, I would be happy to perform an ABX of the actual SAC-DEE versus a PCM of the SAC-DEE, if someone else wants to set it up properly. That accords with the credo of the other religion I practice, which deifies the great lazy bastard Sloth.

I may do an ABX of the SFM transfers , though -- I'm now curious to see if I can tell a 0.2 dB DR difference with music that badly recorded.

Posted: Mon Mar 07, 2005 3:38 pm
by lukpac
krabapple wrote:First, I did not say they won't sound different. I said the PCM transfers I did sound fine. I also said Grant can't know how different the SACD really sounds compared to his transfer (using his usual half-assed 'I take 1/2 from science and 1/2 from belief' method)


Well, Grant *seems* to be saying he doesn't think his PCM transfer sounds as good as the SACD original, not that it sounds "bad" or worse than the native PCM version.

Second, listening in the usual sense (sighted comparison) doesn't necessarily get you the right answer, for the usual reasons -- i.e., the ones that every objectivist has branded under his hair next to '666' when he sells his soul to the horned god Science.


Doesn't necessarily give you the right answer, but doesn't necessarily give you the wrong answer either.

Third, one can have good reason to say 'they *probably* won't sound different, without actually listening.


What reason is that? What do we have to directly measure the SACD data the way we do with the PCM data?

Fourth, if that probability is high (based on good reasoning), one can even predict that 'they won't sound different', without actually listening.


I agree, although that's also assuming there are no unknowns.

I mean, an elephant not only *probably* won't fly -- I predict it *won't* fly. Unless the horned god says it can.


Or unless it's holding on to a blimp.

Isn't "they sound fine" no better than "I can hear a difference!"?

Posted: Mon Mar 07, 2005 8:03 pm
by Dob
krabapple wrote:...the ones that every objectivist has branded under his hair next to '666' when he sells his soul to the horned god Science.

Ah ha!!!! I knew it! Ever since we had that debate about intelligent design...

Posted: Tue Mar 08, 2005 12:34 pm
by krabapple
lukpac wrote:
krabapple wrote:First, I did not say they won't sound different. I said the PCM transfers I did sound fine. I also said Grant can't know how different the SACD really sounds compared to his transfer (using his usual half-assed 'I take 1/2 from science and 1/2 from belief' method)


Well, Grant *seems* to be saying he doesn't think his PCM transfer sounds as good as the SACD original, not that it sounds "bad" or worse than the native PCM version.


<etc>

Let's reviuew what Grant wrote again:

I recently did a couple of transfers of SACDs to PCM through the analog connections. The result was as bit of a letdown as any other redbook CD in comparison. I decided that there wasn't really a reason to do this, unless the commercial CD counterpart was really bad.
I did one transfer at 44.1/16-bit, and one at 48k/32-bit with my best dithering settings. PCM destroyed any advantage thab DSD had.



Grant said PCM destroyed the SACDness. He also implied that *any* CD is a letdown compared to SACD. I said my transfers sounded fine. I also gave data indicating that the CD and SACD of SFM are unlikely to be very different-souding, using the *same* sort of data used to indicate that the CD layer of Dark Side of the Moon will sound different from the SACD layer, in Stereophile and elsewhere. I have *also* indicated that the more rigorous test would be a listening comparison of the SACD to the PCM conversion, which is not easy to set up. So, which one of us made a bullshit cause-and-effect claim?


On the basis of what he wrote, his conclusions, as usual, are utter crap.

Posted: Tue Mar 08, 2005 12:38 pm
by lukpac
krabapple wrote:Grant said PCM destroyed the SACDness. He also implied athat *any8 CD to SACD comparision is a 'letdown'. I said my transfers sounded fine. Which one of us made a cause-and-effect claim?

We dont' know how Grant did his transfers; he's got no data about his transfers; we're pretty sure he didn't do anything like a proper listening comparison. On the basis of what he wrote, his conclusions, as usual, are crap.


Well, isn't saying your transfers "sounded fine" without a blind test no better than saying they are a "letdown" under the same circumstances?

Posted: Tue Mar 08, 2005 12:43 pm
by krabapple
I *also* said in that first post, where I said my transfers sounded 'fine':

But of course there's no easy way to do an ABX with an SACD -->PCM transfer versus the original SACD. So I really can't say how different they sound. And neither can Grant, I bet



So yes, saying it sounded fine with this proviso, *is* rather better than what Grant asserted.
Grant said his transfers didn't sound as good as the original, *and* he asserted a specific reason why that was so (PCM). Neither claim was supported or qualified..

Btw, I revised my previous post to make my reasoning clearer.

Posted: Tue Mar 08, 2005 6:15 pm
by thomh
I agree with Krab on this one. If the numbers in the SFM xls file are correct there is no need to waste time with listening tests. Even though DSD does provide the *potential* for greater dynamic range, it is the master tape in question that sets it and I hardly think that there is >90dB DR on that noisy tape. This song was actually included in our listening tests done some time ago and golden eared audiophiles could not reliably tell one from the other.

Now, if Grant heard DSD characteristics destroyed by PCM then that is a statement that that needs to be backed up by solid evidence.

So where the heck is Grant hiding?

Posted: Tue Mar 08, 2005 6:39 pm
by lukpac
I think we all agree 16/44 *should* be able to reproduce everything just fine under ideal conditions. But who's to say the reproduction chain is always ideal? And who's to say it's all about dynamic range?

Honestly, I think Grant is full of it. Nevertheless, without a blind test of some sort, I still don't think krab is much better off.

Now, Thom, do you have any details on your listening tests? I'm quite interested in that.

Posted: Tue Mar 08, 2005 6:44 pm
by Ess Ay Cee Dee
I think LeeS is the only person that can settle this debate.

Posted: Tue Mar 08, 2005 7:26 pm
by krabapple
lukpac wrote:I think we all agree 16/44 *should* be able to reproduce everything just fine under ideal conditions. But who's to say the reproduction chain is always ideal? And who's to say it's all about dynamic range?


Not me. I've been careful to qualify what I believe my transfer data shows. And you're exactly right, maybe the reproducition chain isn't ideal..but Grant doesn't seem to have considered that possibility. Nope, it's *PCM's* fault.

Honestly, I think Grant is full of it. Nevertheless, without a blind test of some sort, I still don't think krab is much better off.


Which is *pretty much what I said*, as far as knowing how the PCM copy sounds compared to the SACD.

:?


The most inferential part of my argument is that PCM capture of SACD can tell you something about the 'native' SACD. But this is no more or less than the inference used in *every* SACD-to-CD *measurement* comparison I know of to date, the most famous of which is the one for the Pink Floyd SACD. There, the differences measured between a PCM capture of 'Money' and the CD rip, was used to 'confirm' the difference reported from sighted comparison. (A Stereophile revierer decided the CD layer sounded like crap...and sure enough 'Money' turned out to be much more compressed as CD than as SACD). Personally, I suspect that it was the bias against CD that led to the initial judgement...and Stereophile lucked out with the measurements.