Comparing DSD Downloads, SACD Sonoma Copies & PS3 SACD Copies

[bmoura]

There's an interesting discussion at the Computer Audiophile forum on the differences between DSD Downloads created from:

1. Album Edit Master Tapes (the final mix)
2. Cutting Masters (the final mix with DST compression added)
3. SACD Transport or modified SACD player plus a Sonoma DSD Workstation
4. Modified PS3 Sony Playstation

The resulting DSD music files are most accurate from the final mix/album edit master tapes (level 1) and not as accurate at level 4 - created with a modified PS3 Sony Playstation.
A similar comparison and discussion was held sometime ago with Bruce Brown, the mastering engineer who does a lot of work with High Resolution releases and HD Tracks on the What's Best forum.

An interesting topic and read.



http://www.computeraudiophile.com/f...ownloads-too-expensive-experimentation-25758/
http://www.superaudiocenter.com/images/Sonoma32.pdf

 

[Eclectic]

There's an interesting discussion at the Computer Audiophile forum on the differences between DSD Downloads created from:

1. Album Edit Master Tapes (the final mix)
2. Cutting Masters (the final mix with DST compression added)
3. SACD Transport or modified SACD player plus a Sonoma DSD Workstation
4. Modified PS3 Sony Playstation

The resulting DSD music files are most accurate from the final mix/album edit master tapes (level 1) and not as accurate at level 4 - created with a modified PS3 Sony Playstation.

I think you may be (unintentionally) propagating misinfomation. This was one person's unsubstantiated claim, and it's still being discussed (ripped apart?) on the thread in question. It doesn't make a whole lot of sense from an engineering perspective and several participants in the discussion (e.g., "mansr") have said as much. Until someone proves otherwise, I still believe that PS3 rips are bit-perfect copies of the underlying images; I've seen no evidence to the contrary.

 

[bmoura]

I think you may be (unintentionally) propagating misinfomation. This was one person's unsubstantiated claim, and it's still being discussed (ripped apart?) on the thread in question. It doesn't make a whole lot of sense from an engineering perspective and several participants in the discussion (e.g., "mansr") have said as much. Until someone proves otherwise, I still believe that PS3 rips are bit-perfect copies of the underlying images; I've seen no evidence to the contrary.

Actually there have been others in the recording industry that have tested PS3 rips of SACDs vs. Sonoma copies and the original files and came to the same conclusion.

What's new on the CA discussion is that listeners outside of the recording industry are unaware that there is a difference between PS3 and Sonoma rips, much less downloads created from the Edit Masters.
And that explains why mastering, recording and transfer engineers go to the trouble of using SACD Transports, Clocks and Sonoma Workstations for their SACD to DSD transfers vs. using a much more lower cost
PlayStation 3. The PS3 approach would also be easier if the results were equal.

One of the key comparisons that Bruce Brown made in his tests of the different approaches to creating DSD files is seeing if the different files match in size, length and null completely.
Brown's tests, mentioned on the What's Best Forum, found that the Sonoma copies do match and null - and the ones from the PS3 do not.
That's why he uses the Sonoma and DSD file approach vs. a PS3 when creating downloads for his customers.

 

[Eclectic]

Actually there have been others in the recording industry that have tested PS3 rips of SACDs vs. Sonoma copies and the original files and came to the same conclusion.

Evidence please?

One of the key comparisons that Bruce Brown made in his tests of the different approaches to creating DSD files is seeing if the different files match in size, length and null completely.
Brown's tests, mentioned on the What's Best Forum, found that the Sonoma copies do match and null - and the ones from the PS3 do not.

Again, do you have evidence in the form of a reproducible experiement? The main point that people are making in the original thread is that they do not. There are people on the thread who are willing to perform such tests. I'd like to see that happen. If there are bugs in the PS3 ripper, I'd like to see them understood. It does seem somewhat unlikely: If there were flaws, they would likely result in horrendous, easily discernible failures. DSD files do not degrade gracefully.

P.S. My interest in all of this is largely academic. As you're proably aware I try to avoid DSD wherever possible for reasons I've already discussed.

 

[Jim the Oldbie]

...One of the key comparisons that Bruce Brown made in his tests of the different approaches to creating DSD files is seeing if the different files match in size, length and null completely.
Brown's tests, mentioned on the What's Best Forum, found that the Sonoma copies do match and null - and the ones from the PS3 do not.

Unfortunately, Mr. Brown has not been forthcoming with details of these differences to any useful extent (none that I could find, at least - I'd be grateful for any links). Here, he mentions various indexing-type problems sometimes encountered with PS3 rips which will of course create differences. But then he trails off with this nebulous ellipse - "And lets not get into the A/B and null tests...." - after which he basically goes silent on the topic as far as I could see. This is inconclusive at best, and at worst it casts a dubious (and not unfamiliar) odor on his premise.

That's why he uses the Sonoma and DSD file approach vs. a PS3 when creating downloads for his customers.

With respect, Brian, an equally plausible reason is that the guy would quickly be laughed out of business if any of his high-paying customers caught him using a crappy old Playstation in his megabuck studio for anything beyond video games in the waiting room.

-- Jim

 

[bmoura]

An equally plausible reason is that the guy would quickly be laughed out of business if any of his high-paying customers caught him using a crappy old Playstation in his megabuck studio for anything beyond video games in the waiting room.

-- Jim

Brown isn't the only recording and mastering engineer that has compared the PS3 vs. a Sonoma copy of an SACD and found significant differences.

There was also a "shoot out" at an audio show where attendees could hear an album transferred from the Edit Master, SACD to Sonoma and SACD to PS3. The attendees ranked them in that order:
Best sound = Edit Master, Second Best sound = Sonoma, Last place = PS3.

 

[Eclectic]

Brown isn't the only recording and mastering engineer that has compared the PS3 vs. a Sonoma copy of an SACD and found significant differences.

There was also a "shoot out" at an audio show where attendees could hear an album transferred from the Edit Master, SACD to Sonoma and SACD to PS3. The attendees ranked them in that order:
Best sound = Edit Master, Second Best sound = Sonoma, Last place = PS3.

That's not data; that's an anecdote. It's easy enough to get two extracted images (ISOs) of the same SACD, one from a PS3 and one from a Sonoma workstation, and to compare them bit-for-bit. From an engineering standpoint, it's unlikely that they'll differ in any significant way. As I said previously, I look forward to someone doing this transparently and posting the results in detail, so that others can replicate the experiment. Until then, I'll take Mr. Brown's claims with a large grain of salt.

 

[grill]

FYI
I can add two facts to this argument:

1) PS3 sacd-ripping method creates 1:1 decrypted ISO copy of the DSD layer of the original SACD disc. You can check it out with a home-authored SACD-R disc.
2) All PS3 ISO rips of the same SACD are bit perfectly the same.

 

[bmoura]

FYI
I can add two facts to this argument:

1) PS3 sacd-ripping method creates 1:1 decrypted ISO copy of the DSD layer of the original SACD disc. You can check it out with a home-authored SACD-R disc.
2) All PS3 ISO rips of the same SACD are bit perfectly the same.

Well, the tests by Bruce Brown and others have found differently. The copies from the PS3 "sound different and do not null."

By the way, the discussion continues with Bruce Brown over on the What's New forum.
See http://www.whatsbestforum.com/showthread.php?14676-Lampizator-DSD-DAC to view the comments.

A recent highlight from Bruce Brown, message # 330:

"Just to give a quick example. I line up the beginning of both the PS3 rip and the downloaded file, at the end of the album, they can be as much as 10-15 minutes off in extreme cases. Most cases, they are off 3-5 minutes. Then, invert phase of one album or the other, and they do not null. Not to mention what I found out here: Bogus Downloads!"

http://www.whatsbestforum.com/showthread.php?14676-Lampizator-DSD-DAC&p=339405&viewfull=1#post339405

 

[Eclectic]

Grill,

Thanks so much for some actual, scientific information! We need one more thing to completeley debunk Mr. Brown's claim, which is to repeat your experiment on a commercial SACD. Ideally you'd have a known correct ISO. Could anyone QQer provide that? If not, could anyone provide the cryptographic hash (e.g., MD5) of the ISO on a commercial SACD? Identical hashes would be sufficient to prove a perfect copy.

 

[bmoura]

Grill,

Thanks so much for some actual, scientific information! We need one more thing to completeley debunk Mr. Brown's claim, which is to repeat your experiment on a commercial SACD. Ideally you'd have a known correct ISO. Could anyone QQer provide that? If not, could anyone provide the cryptographic hash (e.g., MD5) of the ISO on a commercial SACD? Identical hashes would be sufficient to prove a perfect copy.

More scientific than the tests by Bruce Brown?

Anyone can pick up DSD Downloads of available SACDs and compare them to an existing SACD or PS3 rips they may have made.
A comparison worth trying in my experience.

 

[Eclectic]

More scientific than the tests by Bruce Brown?

Yes, far more scientific! This is a repeatable, falsifiable experiment. That's the backbone of the scientific method. As I said, the only thing we need to do to finish this is to show that Grill's result holds for commercial SACDs. We could do that if someone with industry connections (such as yourself) were to provide us with the cryptographic hashes of the ISOs of readily available commercial SACDs. If you could get the folks at Audio Fidelity to run the md5 utility over the ISOs of some (or even one) of their recent releases, that would suffice.

 

[grill]

Grill,

Thanks so much for some actual, scientific information! We need one more thing to completeley debunk Mr. Brown's claim, which is to repeat your experiment on a commercial SACD. Ideally you'd have a known correct ISO. Could anyone QQer provide that? If not, could anyone provide the cryptographic hash (e.g., MD5) of the ISO on a commercial SACD? Identical hashes would be sufficient to prove a perfect copy.

Here's a sort of proof:

Guano Apes - Don't Give Me Names is the only SACD I'm aware of whose audio data is not encrypted. I could do perfect (ie. listenable) ISO rips of the disc in two ways: with the usual PS3 sacd-ripping method and using ImgBurn software on a separated PS3 BD drive connected to and authenticated with my PC (see the picture below).



Both ISO files have the same MD5 CRC checksums:

723ac0f76c354d519fb70394b9248ed6 *Guano Apes - don't give me names.iso
723ac0f76c354d519fb70394b9248ed6 *DISC1.ISO

 

[Eclectic]

Good work! The ball is now squarely in the court of Bruce Brown and any others who claim that PS3's don't do perfect bit-for-bit extractions. You've demonstrated that they do for SACD-R's and for the only commercial SACD that you could check. You done it in a scientific manner: with repeatable, falsifiable experiments. It looks to me like you've laid bare another case of confirmation bias influencing non-scientific testing.

 

[ssully]

If Bruce Brown is the Puget Sounds guy, he's not someone I associate with scientific rigor.

 

[Kal Rubinson]

"I line up the beginning of both the PS3 rip and the downloaded file, at the end of the album, they can be as much as 10-15 minutes off in extreme cases. Most cases, they are off 3-5 minutes."

That is a Red Herring. Such major differences are so gross that tracks or large portions of them must be missing and indicate a defective rip to be discarded or repeated. The issue is the potential difference between successful rips.

 

[Bruce B]

If Bruce Brown is the Puget Sounds guy, he's not someone I associate with scientific rigor.

Really??

 

[Fourplay]

If Bruce Brown is the Puget Sounds guy, he's not someone I associate with scientific rigor.

Data comparisons are much easier to demonstrate and prove than character slanders. I personally have no data, and nothing substantive to add to this discussion, but I am reading it with interest. Would love it if those with the real goods could stick to the facts and avoid extraneous commentary. I like the way QQ avoids the worst pitfalls of many forums and encourages civil discussion. I apologize if these comments offend. Just sayin'.

 

[Time Lord]

Really??

Probably someone with a sense of humour

Even if it is the real Bruce Brown please don't let the thread degenerate into name calling and one upmanship, this is a great and friendly site run by friendly people, they and the members expect and deserve better.

 

[bmoura]

Data comparisons are much easier to demonstrate and prove than character slanders. I personally have no data, and nothing substantive to add to this discussion, but I am reading it with interest. Would love it if those with the real goods could stick to the facts and avoid extraneous commentary. I like the way QQ avoids the worst pitfalls of many forums and encourages civil discussion. I apologize if these comments offend. Just sayin'.

Check the comparisons of Sonoma and PS3 rips of SACDs over at the What's Best forum, Message # 191
Pretty interesting images.



http://www.whatsbestforum.com/showt...-the-right-way&p=339804&viewfull=1#post339804