Comparing DSD Downloads, SACD Sonoma Copies & PS3 SACD Copies

[Jim the Oldbie]

The real Bruce Brown could certainly help to clear up a few salient points on this interesting topic. For example, I'd be very interested to know some more about any listening tests comparing the PS3 rips vs the Sonoma & other gear; particularly, were they sighted or blind? (Are you still there, Bruce?)

-- Jim

 

[Eclectic]

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

This is a wee bit silly. One shouldn't do analog comparisons of digital data. Grill demonstrated that the PS3 reads bit-perfect images off of SACDs. If the Sonoma does not, then it is inferior. That's one possible implication of the graphs you showed us, but another is that something in the processed used to generate the graphs is broken (i.e., they don't faithfully represent the digital data read by the PS3 and/or Sonoma). The obvious engineering solution is to remove unnecessary layers from the tool chain, and do precisely the experiment that Grill did: See if the Sonoma faithfully reads the digital data from the SACD, as the PS3 does. It really is that simple.

 

[Jim the Oldbie]

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

Yes but as Kal already mentioned, these are major errors indicating some sort of problem with the ripping process. Not to imply that they aren't important or don't need to be addressed, but they also don't really have anything to do with any differences in sound quality between the various processes, as alluded to in other messages. We're sort of talking about two things at once here.

-- Jim

 

[bmoura]

Yes but as Kal already mentioned, these are major errors indicating some sort of problem with the ripping process. Not to imply that they aren't important or don't need to be addressed, but they also don't really have anything to do with any differences in sound quality between the various processes, as alluded to in other messages. We're sort of talking about two things at once here.

-- Jim

You could say there are inherent problems with the PS3 rip process.
That's one way of looking at it.

 

[Eclectic]

You could say there are inherent problems with the PS3 rip process.
That's one way of looking at it.

Umm.. no. Right now it looks like there may be major problems with the Sonoma rip process, until someone dupicates Grill's experiment on the Sonoma. If they do, and the Sonoma also does bit-perfect rips, then there are major problems with the process used to produce those graphs, independent of the ripping abilities of the Sonoma or the PS3. That was the point of my last post on this thread.

 

[Jim the Oldbie]

You could say there are inherent problems with the PS3 rip process.
That's one way of looking at it.

Brian, is this coy, passive-aggressive tone really necessary? Incivility takes many forms, you know? Besides, statements like this do nothing to advance the discussion. To characterize these isolated examples as "inherent" is simply inaccurate.

-- Jim

 

[bmoura]

Having compared a Sonoma rip and a PS3 rip of the same SACD disc, I've found that the Sonoma rip is the one that matches the SACD sonically.
So I find the tests and graphs by Bruce Brown on the What's New forum very believable.

But, others may hear it differently.
To date, I am not aware of any recording and mastering engineers who have tested both and then sidelined their Sonoma and Pyramix setups in favor of the PS3.

 

[Eclectic]

Having compared a Sonoma rip and a PS3 rip of the same SACD disc, I've found that the Sonoma rip is the one that matches the SACD sonically.
So I find the tests and graphs by Bruce Brown on the What's New forum very believable.

But, others may hear it differently.
To date, I am not aware of any recording and mastering engineers who have tested both and then sidelined their Sonoma and Pyramix setups in favor of the PS3.

With all due respect "matches sonically" is not scientific. All sorts of (unintentional) biases may come into play. That's presumably why you said "others may hear it differently." When you're dealing with digital data, there's no need to perform such crude tests. There's an easy test that's not prone to any such biases: Are you reading off the same data that was written, bit-for-bit? That's the test that grill performed on the PS3, and the answer was yes. No one has yet given us similar results from the Sonoma (perhaps because few, if any, of us have accesss to one).

 

[JonUrban]

Bruce B is the real Bruce Brown. Let's treat him with respect, OK?

 

[bmoura]

Bruce B is the real Bruce Brown. Let's treat him with respect, OK?

Bruce is one of the premiere mastering engineers in the recording field. He was responsible for the recent transfer of Analog Master Tapes to DSD Downloads for Wilson Audio (17 albums), the transfer of numerous albums from major labels to FLAC download for HD Tracks and many other projects.

His Puget Sound Studios company will have a room at the upcoming Rocky Mountain Audio Festival (Room 8000) where you can listen to some of his work over top of the line headphones, headphone amps and digital converters, including the Abyss Headphones and the new Lampizator Golden Gate DAC. Should be a show highlight. Having been to the PSS room at past editions of RMAF, if you're attending the show, I'd recommend a visit and listening session!

http://pugetsoundstudios.com/
https://www.audiofest.net/wp-conten...udes/room_exhibitor_info.php?customer_id=1196

 

[grill]

Imho a scientific comparison of SACD rips would/should be as follows:

1) Proving the ripping method is bit perfectly reproducible. I've done it for the PS3 sacd iso ripping at least for a disc. That disc is still available for an affordable price at ebay.de (not my auction) for those who want to do/repeat the test (Bruce?).

2) Direct digital comparisons of the extracted dsd audio files. I don't mean wave form or CRC checks but a binary check using eg. a hex editor (sorry I'm not an expert on this IT field). For this binary check silent regions should be avoided/ignored.

Comparing various DSD masters with SACD is another story, ofc.

 

[Eclectic]

Bruce is one of the premiere mastering engineers in the recording field. He was responsible for the recent transfer of Analog Master Tapes to DSD Downloads for Wilson Audio (17 albums), the transfer of numerous albums from major labels to FLAC download for HD Tracks and many other projects.
http://pugetsoundstudios.com/

Marvelous! In that case, he should have no trouble replicating grill's experiment on the Sonoma, and telling us whether it too is capable of doing bit-perfect reads of SACD ISOs. Using the (unencrypted) Guano Apes - Don't Give Me Names sacd seems like a good idea. How about it, Bruce? I'll bet grill will loan you his copy if you don't have access to one.

 

[Bruce B]

Imho a scientific comparison of SACD rips would/should be as follows:

1) Proving the ripping method is bit perfectly reproducible. I've done it for the PS3 sacd iso ripping at least for a disc. That disc is still available for an affordable price at ebay.de (not my auction) for those who want to do/repeat the test (Bruce?).

And several of us have done it for SACD's where we've taken the Edit Master and compared it to the extracted data from the corresponding SACD. They were bit for bit accurate and nulled to Infinity.

 

[Eclectic]

Imho a scientific comparison of SACD rips would/should be as follows:

1) Proving the ripping method is bit perfectly reproducible. I've done it for the PS3 sacd iso ripping at least for a disc. That disc is still available for an affordable price at ebay.de (not my auction) for those who want to do/repeat the test (Bruce?).

2) Direct digital comparisons of the extracted dsd audio files. I don't mean wave form or CRC checks but a binary check using eg. a hex editor (sorry I'm not an expert on this IT field). For this binary check silent regions should be avoided/ignored.

Comparing various DSD masters with SACD is another story, ofc.

Sorry, I didn't see your reply before I wrote mine. I am a software engineer / computer scientist with over 30 years' experience, and you got this exactly right. There is one small refinement: because the data is so big, it's a pain to check every byte, and not really necessary. If the MD5 hashes match, that's essentially proof that the images are the same--the odds of an md5 hash collision are truly infinitessimal.

 

[grill]

And several of us have done it for SACD's where we've taken the Edit Master and compared it to the extracted data from the corresponding SACD. They were bit for bit accurate and nulled to Infinity.

You mean from Sonoma? If so then that's good. Then Step 2 (direct digital comparison) should be performed.

Sorry, I didn't see your reply before I wrote mine. I am a software engineer / computer scientist with over 30 years' experience, and you got this exactly right. There is one small refinement: because the data is so big, it's a pain to check every byte, and not really necessary. If the MD5 hashes match, that's essentially proof that the images are the same--the odds of an md5 hash collision are truly infinitessimal.

Afaik Mr Wicked's sacd_extract doesn't perfectly handle silence during iso -> dsf audio track extractions. That's why I think one should be cautious when comparing dsd audio tracks with using a CRC method.

 

[Eclectic]

Afaik Mr Wicked's sacd_extract doesn't perfectly handle silence during iso -> dsf audio track extractions. That's why I think one should be cautious when dsd audio tracks are being compared using a CRC method.

Yep. Matching MD5 hash is sufficient, but not necessary. The reason I suggested it was that you've already shown that PS3 extractions are bit perfect copies of the ISO images for home-authored SACD-r's. I did not know that this was not the case for commercial SACDs (due to the silence issues that you mentioned). It really is a pain in the butt to deal with proprietary, non-standard formats such as SACD.

 

[grill]

Yep. Matching MD5 hash is sufficient, but not necessary. The reason I suggested it was that you've already shown that PS3 extractions are bit perfect copies of the ISO images for home-authored SACD-r's. I did not know that this was not the case for commercial SACDs (due to the silence issues that you mentioned). It really is a pain in the butt to deal with proprietary, non-standard formats such as SACD.

In my test the Guano Apes SACD was a factory pressed commercial disc.

...Afaik Mr Wicked's sacd_extract doesn't perfectly handle silence during iso -> dsf audio track extractions. That's why I think one should be cautious when comparing dsd audio tracks with using a CRC method.

I need to correct myself a bit on this. According to Ted B and others from CA forum Mr Wicked’s sacd_extract (any version) has a small bug that leaves DSF extraction with a small click in between tracks due to the final DSF data block being truncated. Ted and others from CA can comment more on this issue. I very rarely listen to DSF audio so I haven't deeply explored this problem yet.

 

[Eclectic]

I need to correct myself a bit on this. According to Ted B and others from CA forum Mr Wicked’s sacd_extract (any version) has a small bug that leaves DSF extraction with a small click in between tracks due to the final DSF data block being truncated. Ted and others from CA can comment more on this issue. I very rarely listen to DSF audio so I haven't deeply explored this problem yet.

Aha. As I understand it, this is a non-issue if you don't make sacd_extract do the DSF conversion. it's not a problem with sacd_extract's ability to read (rip) the raw ISO, but to convert it to DSFs. There are other fine utilities to do the conversion for you, or you can just pass the ISO to some piece of software that understands it. I don't see DSF conversion as fundamental to PS3 SACD ripping. If it can get a perfect ISO, its work is done. If the Sonoma workstation can get a perfect ISO, then it's equally good at reading (AKA ripping) SACDs. But you can't get any better than perfect.

 

[user 10264]

Aha. As I understand it, this is a non-issue if you don't make sacd_extract do the DSF conversion. it's not a problem with sacd_extract's ability to read (rip) the raw ISO, but to convert it to DSFs. There are other fine utilities to do the conversion for you, or you can just pass the ISO to some piece of software that understands it. I don't see DSF conversion as fundamental to PS3 SACD ripping. If it can get a perfect ISO, its work is done. If the Sonoma workstation can get a perfect ISO, then it's equally good at reading (AKA ripping) SACDs. But you can't get any better than perfect.

The Sonoma workstation can never get a perfect ISO simply because it never sees the disc. It can, and most likely does, achieve a perfect copy of the audio data on the disc, and that's really all we need.

 

[Eclectic]

The Sonoma workstation can never get a perfect ISO simply because it never sees the disc. It can, and most likely does, achieve a perfect copy of the audio data on the disc, and that's really all we need.

To a software engineer / computer scientist, this assertion doesn't make sense, i.e., it isn't well-formed. It's digital data. The only perfect copy is the same bits in the same order. Anything else is imperfect, and inferior (unless it's trivially possible to reconstruct the original bit stream from the other copy). So, for example, if a box must do an extra digitization because it's only capable of accepting analog input, then it cannot make a copy as good as a device that precisesly reproduces the bits on the original.

If the Sonoma workistation can't read an SACD, how does it get the input from the SACD? Again, I'm only asking out of curiosity (as I avoid SACD and DSD). To me, the whole beauty of digital music is that it frees you from the distribution format, and even the encoding. I feel confident that I'll never have to buy another copy of any of the digital music I've already bought. Of course I'll delightedly buy another copy if it's a fine multichannel mix, and I currently have only stereo :banana: I wonder if I'll live to see then day when Giles Martin mixes the entire Beatles catalog into 5.1.