Ripping HDCD audio

dark mass, perhaps helpful info: take a look at post #128 earlier in this thread (HDCD flag will be set if the Pacific Microsonics box was used for any functions, even if only digital-to-digital conversion, or dither); and follow the link in post #128 to an explanation presented by John Marks wrt HDCD flag and indication.

hth
chuck

 

I just placed an order for "Nimrod". I'm looking forward to it and to diagnosing the differences between a straight CD rip and ripping via my HDCD Oppo BDP-93 coax out/USBPre 2/Vegas Pro (at 24/44.1k) chain.

Based on your 2012 list, you have a good selection of HDCDs, and no doubt much more HDCD targeted diagnostics than I have to this point. I did notice at that time you were not in possession of Joni Mitchell's "For The Roses" HDCD (Asylum 5057-2). If the album meets your listening criteria, and you don't yet have it, you might give it a shot. No, I am not expecting an HDCD report on the album from you...but perhaps it's an album you'd like, even if only for its technical merits.

 

Thank you. I did take a look at your recommended link following your post in #176. I now gather that when I said the HDCD flag was set "in error" on my mentioned Patty Hurst Shifter CD, I may have been the one in error, due to the now recognized possibility that there just might have been Pacific Microsonics equipment in the mastering set. Based on my technical comparisons, I still tend to believe that such equipment may not have been particularly used to improve album dynamics in this case. (Not to mention that pretty much the entire CD is dynamically squashed like a bug.)

Perhaps it's more proper to say the HDCD flag was set as a side-effect of its mastering.

 

I didn't have any of the Joni Mitchell HDCD titles back then. I have a few now. Working on getting others. But it's slow. There's only one of me to listen to what I get. I'll put "For the Roses" on the wish list.

Both "Blue" (2006 Rhino) and "Clouds" (Reprise) have peak extend
HDCD: peak extend: yes, transient filter: none, gain: none
The DR numbers are slightly higher (by about 1) for the decoded tracks. So the peak extend is being used but not excessively.

 

If I understand your findings, the Oppo is indeed outputing a digital HDCD-decoded stream out of coax. My subjective hearing tests tell me the same thing, but like you, the results vary with which CD I use.

I did correspond with somebody at Oppo (for my 103) who said that the digital HDCD-decoded stream only comes out of HDMI, not the coax or the optical. I don't think that is correct, but I can't prove it with any kind of measurement that I can make.

 

One way to test would be to use the Green Day "Nimrod" CD. Make a CD-R copy of "Nimrod", but apply dither to the audio before burning it to the CD-R. The dither will remove the HDCD magic bits so it won't be recognized has having HDCD encoding. It will play back as a regular CD. Compare the sound of the pressed CD to the CD-R version played on the Oppo over the coax. If they sound the same then the Oppo isn't decoding what is being sent out on the coax. If they sound different then the Oppo is decoding what is being sent out on the coax.

Or, instead of listening you could record the coax output and then compare the DR measurements to see if they're the same or different.

 

I think the best way to settle things with convincing certainty would be to follow Ham Sandwich's diagnostic suggestions.

I now have a small amount of new to me insights which tend to contradict your Oppo correspondent's information, but if I were in your place I'd still want to personally diagnose things. That spirit, after all, is what caused me to actually try recording hopefully decoded HDCD from the coax out of my Oppo 93.

The first thing is that I just completed a comparison of HDCD references in online PDFs of the Opp0 93 and Oppo 103 user manuals. Where HDCD is concerned, the two manuals are identical to each other (if a little less clear cut than I'd like).

Next, this page itemizes differences between the 93 and the 103: http://watershade.net/wmcclain/BDP-103-faq.html#how-is-the-bdp-103-different-from-the-bdp-93 - not too far down the page you'll see it's stated that for HDCD the 93 and 103 are identical. I'm personally convinced that decoded HDCD is provided by the 93's coax out--that being a true view on my part, the inference is that the 103 coax out is also capable of providing decoded HDCD.

Finally, there's this Oppo Knowledge Base article concerning the proper circumstances to enable HDCD decoding: http://www.oppodigital.com/KnowledgeBase.aspx?KBID=44 - lifting text, there is this (pretty much a restatement of user manual information):

"1. Go to the Setup Menu and then Audio Format Setup.

2. Select HDCD Decoding and set it to On.

If you are using analog outputs for audio, you will want to ensure that HDCD Decoding is set to On. If you are using any of the digital audio outputs (HDMI, optical, or coaxial) whether or not you enable HDCD Decoding will depend on whether your receiver is capable of performing HDCD decoding on its own.

If your receiver cannot decode HDCD audio, set HDCD Decoding to On in the player. If your receiver can decode HDCD audio, set HDCD Decoding to Off in the player. In some cases, having HDCD Decoding enabled in both the player and receiver may result in occasional popping or crackling sounds due to HDCD being decoded twice."

Actually, that's a bit different from a user manual in that, unlike the manuals, it groups HDMI, optical, and coax together for HDCD purposes. You'll also see in the article that the article refers to the BDP-105D, BDP-103D, BDP-105, BDP-103, BDP-95, BDP-93, BDP-83SE, BDP-83, and BDP-80.

If "HDCD Decoding" set to "On" could lead to HDCD being decoded twice when an Oppo is connected via coax to a receiver capable of decoding HDCD, logic dictates that decoded HDCD is being transferred through the coax.

But me? I'd still want to personally verify.

 

Good research, Darkmass. I see the nuance in the difference between the manuals and the article you quoted above. If I was the engineer designing the player, I would make all digital streams the same coming out of the player. Again, it doesn't have to be like that, but it seems a more elegant and logical design.

I hope to have some time this Saturday to do some serious listening. I am going to use "Classic Yes", "Yesshows" and "The Trident Sessions" (Mahavishnu Orchestra) as my test discs I remember these as being ones where the decoded vs. undecoded difference was pretty strong.

Now, not to be contrarian, my friend, but tell me if my interpretation of the above is wrong. In the statement :

If your receiver cannot decode HDCD audio, set HDCD Decoding to On in the player. If your receiver can decode HDCD audio, set HDCD Decoding to Off in the player. In some cases, having HDCD Decoding enabled in both the player and receiver may result in occasional popping or crackling sounds due to HDCD being decoded twice."

If the player did decode and output digitally a decoded digital stream, wouldn't this stream be not recognizable as having originated from a HDCD source since the decoding already happened and whatever sonic benefits (if any) the decoding produced be masked or integrated (I am grasping for the right word here) into the resulting digital stream?
In other words, the HDCD capable receiver would only see a digital stream that it could recognize and decode as HDCD only if it was the undecoded digital stream (i.e. the bit perfect data) read from the CD?

 

I think your "contrarian" perception is a good one. And really, both you and I are trying to somehow properly decode Oppo's somewhat problematic attempt at conveying precise information. Our give-and-take is a good thing, and helpful in my view.

I just did a little online looking for the file location of HDCD flags, and the little I now know is that they are in-stream and "typically repeated ten times a second" (from the final post on this page: http://www.hydrogenaud.io/forums/index.php?showtopic=30999&st=125). But I suspect Ham Sandwich could provide information both more complete and more nuanced.

Perhaps what it comes down to is this: does an Oppo clear out these flags in the process of decoding HDCD? I would hope it would clear them out, leaving a result that would be, say, a stream of pure 24/44.1k musical information...which would pass through any downstream HDCD decoder with no further attempts at signal modification. However, if in the process of HDCD decoding an Oppo left the HDCD flags intact within the musical stream, then Oppo decoded HDCD could well be "re-HDCD decoded" by a downstream HDCD decoding receiver. From Oppo's wording, this second possibility seems like it might be the case.

 

I agree with you that from Oppo's wording, the inference is that the possibility exists of a second decoding.

But I think that is wrong. I think that the HDCD flags do not survive into the decoded 20 or 24 bit stream. It has no reason to. As you say, the decoded stream should be pure musical information ready to be turned into analog and all the HDCD menu of possible features have already been applied and used to create the decoded stream. I think the flags only exist in the undecoded 16/44.1k data.

I agree that Oppo's various explanations can't all be correct. It is really a small detail and we shouldn't rag on them for it. I am just happy somebody still supports HDCD. It really is the reason I have had two players from them, rather than a cheaper alternative. I have a couple of cheap (<$100) Sony BR/SACD players that are pretty good, but Oppo is in another league.

 

The wording by Oppo's warning about potential glitches if the HDCD gets decoded twice is ambiguous, and doesn't make sense. Maybe it actually is possible for some HDCD codes in certain situations to survive the first decoding and end up triggering a second decoding by an HDCD DAC that gets sent that digital signal. Maybe. Seems unlikely. But I don't know how the encoding and decoding works at a low enough level to confirm or deny that the double decoding could happen, and if it would cause problems.

In general though, any processing that alters the original HDCD encoded audio data will obliterate the HDCD magic bit codes. If those magic bits hiding in the dither are altered then audio data won't be detected as having HDCD.

 

I tried a new test and went down in flames.

The thought was that maybe I could use Vegas Pro to record Oppo 93 coax output (via the chain I've previously mentioned) first with the Oppo providing HDCD decoding of a "For the Roses" segment, followed by second recording with the Oppo HDCD processing turned off. Each recording was to 24/44.1k wav, and I was hoping Windows Media Player would signal HDCD for one or both recordings. Unfortunately, neither recording lit the WMP HDCD indication.

I followed that by a normal "For the Roses" track rip to 16/44.1k wav--first using Vegas Pro, then using Sound Forge Pro. Neither of these resulting wav files lit the WMP HDCD indication. I followed that by using WMP to rip one of the "Roses" tracks (it ripped to 16/44.1k)...and using WMP to play back a WMP rip also did not light the HDCD indication--even though WMP has a level of HDCD awareness. I verified that the WMP rip also did not provide any HDCD expansion within its rip.

Without HDCD flag maintaining recording software, I can't provide any knowledge about HDCD flags in Oppo output streams. However, my feeling is also that the Oppo information is incorrect and there is no potential of double HDCD decoding. Between Oppo hardware engineering and documentation there was some technical misunderstanding...which will be preserved forever as long as Oppo HDCD hardware is not changed.

 

from an older SHF thread:

2/ re decoded HDCD via S/PDIF: iirc, the decoded HDCD signal is truncated to 16bits before transmission via S/PDIF; the only digital output thru which you can access the full 20-bit HDCD decoded signal is HDMI (20-bit HDCD packed in 24-bit HDMI audio words).

This is not an oppo issue, but rather a system design issue of the Mediatek decoder IC. I do not know definitavely, but am informed and believe that it is an unfortunate byproduct of older restrictions placed upon Mediatek (and other IC decoder manufacturers) by licensing agreements regarding transmission of either/both DVD-A and/or DVD audio; these require that audio be limited in resolution for digital transmission over S/PDIF links. It seems that in the Mediatek system architecture (and other manufacturers' decoder ICs as well) this truncation is applied to all decoded formats before transport via S/PDIF

 

So we should all move to HDMI capable receiver/amps to get the full digital information? I think I had heard something like what you are saying, Emmodad, if so, have I been listening to 2/3 the music I thought I was for years now?

Actually, what you say, if correct, explains a lot of what we have been discussing. Yet when reading the specs for my receivers, they say that the opt/coax inputs can handle up to 24 bits. But if I understand you correctly, these inputs were never going to get anything but 16 bits out of a digital link with a player. Am I understanding your point correctly?

 

Does the 16 bit limitation apply to coax too, or just S/PDIF?

 

Coax is (usually) S/PDIF.

 

You are right. I usually associate the term S/PDIF with optical fiber, but you are right it could be RCA coax too.

 

Isn't the question, "will any hardware HDCD decoding solution provide a digital bitstream with resolution higher than 16 bits?"? Perhaps the HDMI vs. S/PDIF question is a red herring.

I'm thinking the answer is no, and many of us are used to using "hdcd.exe", hence, the expectation of a 20 or 24 bit output.

 

If it decodes HDCD correctly, some discs should output more than 16bit in ideal conditions.

The HDCD softwares certainly do, I have verified it with Foobar.

 

I just captured Bluebird from Buffalo Springfield Again out of the Oppo 103, first with HDCD decoding turned off and then with it turned on. It's definitely doing the decoding (although there honestly isn't a huge difference with that disc/track), but I don't know how to answer the question about whether the low order bits are used. I can send the files to someone else if they wish to analyze them.

 

I can put them through a bit-meter and tell you what resolution they really are.

 

Here are the raw 24-bit files, complete with the very beginning of Hung Upside Down.

Bluebird [undecoded]
Bluebird [decoded]

FYI, captured with a Tascam DR-680. I haven't yet checked to see if the undecoded file is identical to a rip right from the CD. Hopefully it is.

 

Okay, I'll take a look tomorrow and let you know what I found.

 

Not sure what you mean by ideal conditions, but I think that where this discussion is headed is that while internally a hardware player will have a resultant 20 or 24 bit depth from the HDCD decoding process (and that is what goes into the player's DAC), the digital output gets truncated to 16 bits (as per Emmodad's comment, which does ring a bell with me and is worthy of further research). We just won't know for sure until somebody can measure the bit depth coming out of a player.

 

I'm not sure if the decoded stream comes out of the stock coaxial output from an Oppo 103. But I can definitely confirm that the decoded HDCD output comes through my Vanity HD card which is mounted inside of my Oppo 103. I'll have to do a test to compare the stock digital output against the HDs output.