Ripping SACDs

Makes sense, and very helpful - thanks!

 

You're right, darkmass - I'd completely forgotten that the KanexPro actually has an optical output as well as an S/PDIF coax. I really should have put that in the guide as it's very important for Mac owners. The 3.5mm mini jack input on the Mac definitely does take optical input - I've used one in the past. And it's easy to find optical cables that go from a standard jack to a mini jack - often, optical cables with standard jacks are supplied with mini jack adapters so you can use them in either configuration.

So, tmtomh, you shouldn't have any problems at with that setup. I'm not sure that Macs will support a sampling rate of up to 192 kHz on their optical input so that could be a problem if using an SACD player with an output of 176.4 kHz. However, there'd be no problems with an Oppo's 88.2 kHz output because I'm sure the Mac goes at least as high as 96 kHz.

 

This post is just for fun - and it's about de-embedding CDs, not SACDs, so it's not even strictly on topic!

Just suppose you have a CD with pre-emphasis such as the early Harvest blackface Dark Side Of The Moon. You can't rip it on your PC because your rip would have the pre-emphasis and it would sound terrible (though you could try to decode it in software). Since most CD players decode the pre-emphasis, a lot of people deal with the problem by playing the CD on their CD player and ripping the analogue output from the player.

But here's the question: what if you played the CD on your SACD player and ripped it through an HDMI de-embedder? Would your rip have the pre-emphasis or not?

Place your bets now! I'm going to try it today and will tell you the result tomorrow.

 

De-emphasis was originally an analog process. Now it seems to be done digitally for the few players that support it. Theoretically the player could pass through a de-emphasized version of the audio. My guess though is that a modern player would not bother. At most it might add a flag saying that the digital signal needs de-emphasis knowing full well that whatever receives the signal would almost certainly ignore the de-emphasis flag.

 

The problem is that the Pre on some of these discs are located in the SUB Q channel, quite a few newer players will not detect it, Oppo included, If you have a pre cd located in the TOC you may have better luck, would be neat if we could come up with a list of players that would add it via hdmi, i'm thinking the list will be small however

 

This would be very interesting!

But the most fundamental issue here is that since you're already ripping the disc to a computer, why not rip it properly, with full error correction and AccurateRip checking, and then de-emphasize later via software as you say? When you use an HDMI de-embedder, you're getting a live stream, including any on the fly error correction the disc play might have to apply if there are defects on the CD.

 

The pre-emphasis flag is transported in the SPDIF signal, so that an external DAC knows to engage its de-emphasis circuit. However, if you record the SPDIF signal, the recording software will have to ensure that the emphasis flag is preserved in your WAV file. I have no idea whether that is the case.

One would need a copy of the HDMI specification to answer definitively, but playing around on Google I am inclined to say that the emphasis flag is NOT transported over HDMI.

So, for your test(s) you would need:

1. a CD known to have pre-emphasis;
2. rip that to your PC;
3. record the SPDIF output of your player;
4. record the HDMI de-embed;
5. compare the files.

You can also do a listening test (Sony 5400 playing CD vs. the PC files).

One thing I can tell you, your Sony 5400 will very accurately decode pre-emphasis!

 

You could try playing the SACDs in something that will play the pure and raw DSDs and hook them up into a Mic Input on a computer, record it using Audacity, and export as anything from 24/48 FLAC or 24/96 FLAC or WAV. You could even go for 24/192 WAV (it will be very big and take forever to load in Audacity.)

 

.....the mic input on a compu.....oh dear god

 

I'm not convinced you've thought this through, still less tried it. How do you plug an HDMI cable into a microphone input? And the microphone input is analogue, not digital.

 

Here are the results of my pre-emphasis test. Many thanks indeed to those who took an interest and played along.

Well, it's good news and bad news. The de-embedding process DOES remove the pre-emphasis. But the resulting EQ is not the same as the EQ that comes out of the analogue outs!

There's always a problems, isn't there?

Here's my chain. Disc played is Dark Side of the Moon by Pink Floyd. It's the so-called Holy Grail version - sounds terrible to me, far too Dark (!), but that's another story. It's the Harvest black face CDP 7 46001 2 CP35-3017 28A5. I had a moment of horror when I found an obvious dropout in my de-embed at 6:04 on Money and then discovered it's on the disc itself. Strange that this doesn't get talked about but maybe it's only the 28A5 that has it.

The CD went on my Sony SCD-XA5400ES SACD player. I switched on the HDMI and took the HDMI output to my de-embeddder (I used the KanexPro and the Cyp with identical results) and then by S/PDIF to my Lynx Hilo DAC. The output from the Lynx Hilo went via USB to my Windows 10 PC where I recorded it into Adobe Audition CS6. I compared EQs using Voxengo CurveEQ.

I then switched off the HDMI and recorded the same disc from the analogue outputs of the player into the line in of the Lynx Hilo and then by the same path to the PC.

These two versions were compared with a straight rip of the disc to my PC, which obviously would still contain the pre-emphasis.

Here are the results. I've crammed them all into one chart for economy.




Everything lines up on the left, which is good news. The differences are all at the right-hand edge. The green is the PC rip with all the pre-emphasis present. The dark yellow is the analogue rip from the SACD player which reduces the top end an enormous amount. But the light yellow is the de-embedded rip from the SACD player and you can see that the de-embedded rip removes even more top end than the analogue rip. The orange dotted line compares the de-embedded rip to the analogue rip and shows you how much less top end the de-embedded rip has.

I'm going to have to think about this, particularly in view of what Black Elk says. It's interesting that my DAC actually has an "Emphasis" feature. Here's what it says in the manual:




When I was recording through the de-embedder, the Emphasis indicator was showing Unknown (which actually shows grey, not green - they have their grey and green the wrong way round in the manual.)

I'm just going to leave it there for the moment while I think about it some more.

 

Interesting, can you try one with pre in the TOC, not SUB this time around?
I just gave this a whirl, no luck with either SUB-Depeche Mode Construction Time Again (Early German) and TOC-Pink Floyd The Wall (US Columbia Fatboy)

 

Unfortunately I don't think I have any CDs with "normal" pre-emphasis. I have the first CD issue of Jethro Tull's This Was but when I put it through EAC and dBpoweramp, neither of them can see the pre-emphasis, so again, I think it must be in the subcode. It comes out very well and it's a lot better than the Dark Side but it still isn't quite right. (Chart for the whole album follows.) When you say no luck with your two CDs, do you mean you tried the de-embedding method and the pre-emphasis didn't come out right for you either?

 

Correct, the Oppo did not add the pre EQ curve via HDMI using the Kanexpro as I think with most if not all the internal DAC's it will only trigger through the analog out stage, As Oppo does not support Pre in the SUB I will probably use my Denon's analog out to properly add the curve

 

OK, let's back up the train a little. back2vinyl, to get a reference, you need to compare the spectrum of the ripped CD to the analog rip of the CD. The pre-emphasis/de-emphasis breakpoints are defined as 50 microseconds (3,183 Hz) and 15 microseconds (10,610 Hz), and the gain is +6 dB/octave for pre-emphasis, and the cut is -6 dB/octave for de-emphasis between these two points. You can see the characteristic here:

https://books.google.com/books?id=c...age&q=compact disc audio pre-emphasis&f=false

You will notice that the responses are drawn as perfectly straight lines! Good luck getting such a response in practice. Real filters cannot make such sharp transitions, and so a best approximation has to be made, usually with the break starting before the lower frequency and extending beyond the upper one. Given the different ways in which these filters can be realized (for encode and decode), it should be obvious that this is a source for distortion depending on how accurately the encode/decode filters match each others responses. This image shows a practical response relative to the ideal response:



The ratio of the two breakpoints is such that the change is almost exactly 10 dB.

Unfortunately, the review of your player that I linked to earlier in this thread does not contain measurements of the CD output playing de-emphasized audio, but Stereophile did do the measurements for the fore-runners of your player:

http://www.stereophile.com/content/sony-scd-xa777es-multichannel-sacdcd-player-measurements

http://www.stereophile.com/content/sony-scd-xa9000es-sacd-player-measurements

You will notice that the response difference is something like 0.1 dB in the high frequencies, which is unlikely to be an issue. I would not pay attention to the difference between left/right channels in the 9000's measurements, as I believe that they have simply offset the channels slightly to make the dotted line visible (otherwise the responses would overlay).

So, I would expect the 5400's response to be as good, and a comparison of the CD rip to analog dub should show you the response of the pre-emphasis/de-emphasis chain (you may want to change the scales to focus in on frequencies above 1 kHz, say, and that gives you the maximum y-axis scale for comparison).

Looking to the de-embed response, it seems that that path is giving you 'correctly' de-emphasized data. You are simply seeing the difference in de-emphasis responses (since the pre-emphasis response is fixed).

How are you getting de-emphasized data this way? My guess is that the Sony is doing digital de-emphasis, and sending this as LPCM over HDMI. Why do I say that? Because I cannot find anything about the flagging of pre-emphasis over HDMI for digital audio (not to be confused with the use of pre-emphasis on HDMI transmissions!!). If anyone has the HDMI spec., maybe they can confirm? So, if HDMI digital audio transmission does not support pre-emphasis, I assume that the Sony accounts for it before creating the data stream.

Another option is that the Lynx does the correction, but I find that harder to believe. That it correctly identifies the flag and engages its decoder when needed during listening is one thing, but I would not expect it to convert incoming data, and then pass that out (I would expect a simple pass-through).

With regard to your Lynx, it is obvious from the manual that they are expecting more AES-EBU use than SPDIF. However, the Digital In Channel Status and Digital Out Channel Status indicators could still tell you meaningful things. So, I would try connecting the SPDIF out (coax or optical) from the 5400 to the Lynx, and check the In and Out status indicators. I would expect to see:

IN

Lock = Locked
Mode = Consumer
Validity = red or gray (there is no Valid bit in the SPDIF stream)
Emphasis = 50/15 us (green)
Status = Normal (I assume this equals the Normal audio mode/non-audio mode flag in AES-EBU = normal/digital data in SPDIF)
Errors = None (ideally)
Type = PCM16
Rate = 44.1 kHz

OUT

AES = don't care

SPDIF (coax or optical depending on choice) = don't care, unchecked, checked

Since there is no Valid bit in SPDIF, I don't know why it is part of the display, unless you were going AES-EBU in, SPDIF out, and just wanted confirmation that the data is valid. With your Lynx getting a pre-emphasized signal over SPDIF from the Sony, I would expect it to pass the same through via its SPDIF outputs, hence Emphasis should be CHECKED in this case.

If the Sony is doing what I suggest, when you switch to the de-embed path, the Emphasis state on the input should be OFF (gray), and the Emphasis checkbox for the output should also be UNCHECKED.

For those interested in the emphasis flags in CD, member SamS and I get into the weeds in this thread:

http://forums.stevehoffman.tv/threads/pre-emphasis-and-oppo-blu-ray-players.244533/

 

Many thanks indeed for that, Black Elk.

I'll do that Lynx Hilo test later so I'll leave that to one side for the moment, if I may.

Now, focusing on the difference between the de-emphasised result, analogue vs. de-embedded. Why are the results different?

I understand what you say about real-life filters being unable to produce the Z shape of the idealised pre-emphasis response. I'm sure the explanation lies here but I'm just trying to work it out so I fully understand the effects. After all, it is quite a big difference and I would like to understand which end-result is closer to the original master.

We're looking at three points where pre-emphasis and/or de-emphasis are applied:

1. When pre-emphasis was originally added to the recording by EMI (or whoever).

2. In the case of the analogue rip, when the pre-emphasis was de-emphasised by the analogue section in the Sony player.

3. In the case of the de-embedded rip, when the pre-emphasis was de-emphasised by the Sony player's digital section (or conceivably, by the Lynx Hilo).

Now, what I'm trying to understand is, which of these employ a true Z shape response and which of them apply a softened, bendy approximation?

Because if the answer is that 1 and 3 employ a true Z shape, and 2 applies an approximation, we'd be saying that the de-embedded version is far closer to the original master than the analogue rip.

On the other hand, if we're saying that 1 applies an approximation, 2 applies a similar approximation and 3 applies a true Z shape, then 2 is going to be a lot closer to the original master than 3.

If we're saying that all three apply an approximation - well, that's probably what I least want to hear because then we're in a world of chaos and confusion. What we'd be saying then is that two different players, and even two different sections of the same player, can provide such different approximations of the idealised de-emphasis response that, whenever the consumer listens to a de-emphasised disc, he or she can really have no idea whether or they're listening to an accurate representation of what was on the EQd master. Is that what we're looking at here, do you think?

 

Oh no... I think I have messed up again, big time. I think all the recordings match after all. The only reason they came out looking mis-matched was because the clock synch was wrong on my Lynx Hilo again. I'm looking into this and will report back later.

 

Is the jailbroken original PS3 still the only way to digitally rip the DSD layer on an SACD? I don't get why this is still not doable with a regular DVD or Blu ray drive. Wouldn't all they have to do is update the firmware so it could read the other layer?

 

The DSD layer of an SACD is encrypted.

 

I'll await to hear more about the clocking, but you may want to create a check-list for analog and digital recordings, and keep it by your Lynx so you are reminded of which settings to change in each case.

With regard to the filters, you have to remember that pre-emphasis was being used in 1982/3, and for a brief period after that. So, I would expect the pre-emphasis filter to be analog, and so have the softened, bendy shape -- I would imagine that they would not want to screw up the phase horribly either. As Sony was (a) one of the co-developers of CD; (b) a disc manufacturer; and (c) a player manufacturer, you can be pretty sure that they would have a good idea of the real-world filter responses used by the disc replicators, and what response is needed in the player to undo the emphasis. This is why their players measure so well, and you really need test signals to know how good a device is at de-emphasis. Of course, if the replicators used different pre-emphasis designs (say Sony, Denon, ODME (Philips), etc.) then it may be possible that the results could change depending on disc manufacturer. However, I would expect them to have agreed on something cheap and simple that would be consistent across the board.

Of course, if you opt to do the de-emphasis in the digital domain you can either try to simulate the analog filter response, or you could try to more closely follow the 'Z'. Ignoring any phase issues, a perfect 'Z' response may be accurate according to the spec., but may be less accurate in practice since the encoding was done with a 'soft' filter.

Anyway, let's see what you turn up when you double-check your settings.

 

Cracking of the PS3's security did not crack SACD's security. You need special chips in the optical pick-up to read SACD, a standard DVD or Blu-Ray drive is not sufficient, and since SACD's security mechanisms are still intact, you need the shroud of the PS3 to unlock the data, and then the hack to allow you to gain access to that unlocked data.

Think of it like this: there's money in a vault which you can't open, but you know some little old lady walks out of the bank with a bag full of cash every Tuesday morning, so you lie in wait for her. You keep getting the cash, but you are no nearer learning the code to unlock the vault.

 

Well, I have good news about pre-emphasis with the de-embedding technique but first of all I have to make a big apology for my initial posts on this subject which now turn out to have been completely misleading. I must especially apologise to Black Elk who went to a lot of trouble to help. I can only say, Black Elk, that what you posted was very interesting and informative in its own right so I hope you won't see it as wasted.

What I've now found is that ripping a CD with pre-emphasis delivers exactly the same de-emphasised EQ whichever method you use - analogue ripping or the de-embedding technique. At least, it does with the Sony SCD XA5400ES, which must have matching de-emphasis filters in its analogue and digital sections. Other players may not do de-emphasis in the digital domain, as I think c-eling has reported with the Oppo.

The reason my first measurements were wrong was that, unknown to me, my Lynx Hilo DAC had secretly disengaged its Synchrolock from the digital input and set it back to its internal clock. I often have to reset the defaults on the Hilo and it turns out that whenever this is done, it resets the Synchrolock as well, even though the user himself can't access that setting as long as the USB cable is plugged in. Oh well, lesson learned - I will just have to check and double-check the Synchrolock whenever I use this DAC.

Now I'll show you the true comparisons between an analogue rip and a de-embedded rip, done on the Sony. The first is the whole of the CD Dark Side of the Moon and the second is the whole of the CD This Was by Jethro Tull. Both have pre-emphasis. As you'll see, they come out ruler-flat apart from that tiny little uptick at the top of the frequency range and that's nothing to do with this exercise - I always get that with the Sony when comparing an analogue rip with a digital rip, possibly because the Sony's internal DAC doesn't cut off the top of the frequency range quite as sharply as in the digital domain.





I'm really pleased with the results I'm getting from the de-embedding technique and it now turns out that, depending on your SACD player, it also provides an alternative method of ripping pre-emphasised CDs in the digital domain, with excellent results.

 

No worries, the hours I spent on the phone yesterday and today with AT&T and Microsoft, now that is wasted time/effort!

Glad you finally have everything nailed down. I would re-iterate: make a check-list, and run through it ahead of each recording/listening session with the Hilo. Use the display feature to ensure everything is locked/in the right mode.

Did they say anything more about the 24/192 SPDIF issue?

 

Thanks for your understanding, Black Elk. I have written myself a big note and stuck it to the Hilo!

Incidentally I did try your tests and found that even using the digital CD output from the Sony SACD player - not the HDMI output - the pre-emphasis was already de-emphasised by the Sony so no pre-emphasis ever reached the Hilo. One odd thing I did notice however is that, when using the digital CD output, the Hilo read "PCM18" instead of just "PCM" which is what I normally get. I found that a bit mystifying but I'm not going to lose any sleep over it.

Lynx do seem to be working on the defective Hilo optical input. I discovered it was temperature related - the input worked when cold but failed when the unit warmed up - and they confirmed that. Then they discovered that lifting the input jack helped, to the extent that the lock was maintained to a higher temperature before the lock was lost. Then they asked me what cable I was using and I found that a glass cable bought more time than a plastic cable. That's where we are at the moment. All very strange.

 

[QUOTE="Black Elk, post: 13111312, member: 7514]Think of it like this: there's money in a vault which you can't open, but you know some little old lady walks out of the bank with a bag full of cash every Tuesday morning, so you lie in wait for her. You keep getting the cash, but you are no nearer learning the code to unlock the vault.[/QUOTE] What a great analogy!