Hoffman gives up on HDCD

[blessingx]

I've decided not to run my masterings through an HDCD converter..

[Dusty Chalk]

I still want to borrow the box so I can decode all my HDCD discs. I still think properly decoded HDCD is superior to undecoded HDCD.

[mypasswordis]

i am 12 and what is HDCD

[Grahame]

^ Why not do it in software (hdcd.exe) via dbpoweramp or foobar? Then you won't need the box (assuming your DAC of choice can handle the resultant 20/24 bit files)

[Grahame]

i am 12 and what is HDCD

http://lmgtfy.com/?q=Hdcd&l=1

[blessingx]

Hydrogen Audio gets a little critical of Hoffman.

[Dusty Chalk]

i am 12 and what is HDCD

Ask your mother.

Ask your father.

No, seriously, I'll explain later when I have a little bit more time to compose my thoughts -- I needed to read about umpteen nate-phillion pages about it before I got it, and it's been a few years, so...

[deepak]

Hydrogen Audio gets a little critical of Hoffman.

Heh it's like a vacuum tube kicked their dog

From what I understand HDCD was mostly a marketing gimmick to sell the Pac Micro boxes to mastering studios. There is only one instance (peak extension) where it makes a difference.

[aardvark baguette]

From what I understand HDCD was mostly a marketing gimmick....

Then by golly, what the devil was it doing as a subject in the Hoffman forums?

[mypasswordis]

The extremely limited reading I've just done seems to imply there's just an extra bit for improved dynamic range, but when played through a non-HDCD player only has 15 bits? What does any of that have to do with a dip in frequency response at 8kHz? I don't know much about Steve Hoffman but it's starting to sound like he's one of those special "audio artists." Audio engineers as a whole seem somewhat dubious to me.

[deepak]

yeah, no kidding. what i find funny is the idea that a theoretical plug-in that perfectly emulates a tube is acceptable, but actually just using that tube is somehow suspicious.

There is some funny stuff there, like the guy that thinks running an audio reproduction tube preamp is like a guitar preamp which is intended to over drive and distort the signal a lot.

TBH with the finer points of HDCD I would trust Filburt over many of those HydrogenAudio guys.

[monsieurguzel]

What I don't understand is that the last 20+ releases from Audio Fidelity, which are all mastered by Kevin Gray and Steve Hoffman are all using HDCD for encoding the CDs. This was being done for years now. I don't understand how all of a sudden SH has got this revelation that HDCD process degrades the sound...shouldn't he have noticed this when first using it? I'm wondering if there is a different motive for dropping it like licensing fees or unpopularity of HDCD decoder boxes in the market.

[aardvark baguette]

What I don't understand is that the last 20+ releases from Audio Fidelity, which are all mastered by Kevin Gray and Steve Hoffman are all using HDCD for encoding the CDs. This was being done for years now. I don't understand how all of a sudden SH has got this revelation that HDCD process degrades the sound...shouldn't he have noticed this when first using it? I'm wondering if there is a different motive for dropping it like licensing fees or unpopularity of HDCD decoder boxes in the market.

Do you know how many times Pet Sounds has been released as 'audiophile' versions?

Having batted through the 10 or so classic rock staples the forum likes, he is obviously gearing up to re-release the same goddamn Deep Purple album for the millionth fucking time.

[Dusty Chalk]

HDCD is an encoding process that is mostly compatible with standard redbook format. There is still 16 bits of information, but undecoded, it's supposed to be close enough to the original signal as to be inaudibly different, but properly decoded provide an additional amount of information. This additional amount is supposed to lead to slightly better dynamic range, the amount arguably anywhere from 1 to 8 bits (I have heard 24, but even most believers don't believe it is that much).

Are you familiar with coordinate systems? Just pretend it's a different coordinate system -- the 16 bits, instead of linearly taking up the normal ...whatever it is, 2V range, takes up slightly more, and non-linearly (it's when the total values are near their maximum that they are expanded, and when they are near their minimum that they are attenuated).

Okay, here's the thing about the 15 bits -- the signal as to whether or not an audio stream is HDCD-encoded is actually in the least significant bit in the audio stream, which is played by most non-HDCD-decoding players! So their claims are somewhat self-contradictory -- they claim that you cannot hear the least significant bit (the practice is described as "non-destructive" and "compatible" with redbook), and yet the subtle changes are claimed to be an improvement.

I should also note that whether or not he is correct is irrelevant as to whether or not I want my HDCD-encoded disks properly decoded -- those are two separate issues. The only thing that it might affect is whether or not I want the disk HDCD-encoded or not, if both choices are available from the same master. Which they usually aren't (I know of exactly one such disk).

I don't understand how all of a sudden SH has got this revelation that HDCD process degrades the sound...shouldn't he have noticed this when first using it?

Yes, as stated, he only is just figuring it out.

I agree with the theory that he's just setting up to re-release the remasters without HDCD encoding. Either that, or they screwed up on the in-house monitoring, and forgot to decode it.

[monsieurguzel]

Do you know how many times Pet Sounds has been released as 'audiophile' versions?

Having batted through the 10 or so classic rock staples the forum likes, he is obviously gearing up to re-release the same goddamn Deep Purple album for the millionth fucking time.

I have been through every single edition of that CD and have payed thousands of dollars hunting down for the perfect audiphile version over the years. I personally like the one issued in in 1984 pressed in E. Germany what has serial number S39784-5084. There are actually two version of this exact CD pressing, but the one that has the print of a monkey etched on the inside CD hologram is the ultimate version in every respect.

j/k but that is pretty much what every post at SH sounds like when someone asks for advice on the 'best' edition of a certain pressing

Edited April 6, 2011 by monsieurguzel

[Dusty Chalk]

Photoshoppers need to get busy photoshopping laser-etched monkeys onto CD's.

[NightWoundsTime]

The extremely limited reading I've just done seems to imply there's just an extra bit for improved dynamic range, but when played through a non-HDCD player only has 15 bits? What does any of that have to do with a dip in frequency response at 8kHz? I don't know much about Steve Hoffman but it's starting to sound like he's one of those special "audio artists." Audio engineers as a whole seem somewhat dubious to me.

Hey I wanted to be one of those. I'm still pretty dubious though, so maybe it's a type that's attracted to the job, not the job that makes the type.

The non-linear dynamic range of HDCD is somewhat cool, but it was always a gimmick and I'm surprised anyone ever cared. Oh wait, audiophools cared, and that shouldn't surprise me.

Edited April 6, 2011 by NightWoundsTime

[mypasswordis]

Okay, here's the thing about the 15 bits -- the signal as to whether or not an audio stream is HDCD-encoded is actually in the least significant bit in the audio stream, which is played by most non-HDCD-decoding players! So their claims are somewhat self-contradictory -- they claim that you cannot hear the least significant bit (the practice is described as "non-destructive" and "compatible" with redbook), and yet the subtle changes are claimed to be an improvement.

Cool thanks for the write-up. Wikipedia seems to say a maximum of 20 bits of dynamic range, the 16th bit being the control bit, and I guess it's used to re-dither back to 16 bits for non-HDCD players, which should be better than truncation to 15 bit resolution.

Edited April 7, 2011 by mypasswordis

[Dusty Chalk]

But the thing is, non-HDCD-decoding players (I.E. most of them) don't know that, so they just play it.

EDIT: There's a lot of misinformation about HDCD -- for example, that post on the hydrogen audio forum about it being a limiter -- that's only part of it. It's supposed to be inaudible on a redbook playback machine, but it's undone -- I.E. expanded back out -- on a HDCD playback machine.

EDIT: About the suggestion to decode it in software -- I'd like to, but I don't feel comfortable that the software captures every variation of HDCD decoding that it's supposed to -- there are a lot. It's not simply a "yes/no" choice, it's a function of settings.

[mypasswordis]

There's more noise due to the dithering compared to normal 16 bit redbook, is what I think they're trying to say.

And... the point of software is it can be written and modified to do almost whatever you want, as opposed to dealing with set hardware. I'm gonna spend the next couple of days reading the PMD100 spec sheet and then figure out a good time to meet so we can play some of your HDCDs through the Parasound 1100 (have you heard one before)? Weekend after next?

Edit: Cool thrad http://www.diyhifi.org/forums/viewtopic.php?f=2&t=2056

Edited April 7, 2011 by mypasswordis

[Dusty Chalk]

Well, one of the things that I heard "them" say is that the code isn't just sequences of bits, it's a specially designed sequence of bits that is deliberately part of the dither, so theoretically any audible effects the sequences would otherwise have should be overshadowed by the dithering process, of which the encoding is an integral part.

I don't remember if it's a one-bit dither or not.

Re: software -- yeah, I have the source code somewhere (the guy who wrote it was nice enough to share it with me so that I could attempt to compile it on linux, but that was several hard drives ago, so I'd have to find it at this point), but reversing it on Windows should (if they implemented it correctly) just be a matter of accessing the API's.

I've heard something before (in fact, used to even own an unmodified something -- 2000? 1500? 1600? I forget which one), but have prefered the modified ones I've heard at meets? I can pull out the MSB Tech Link DAC III (w/HDCD & upsampling upgrades) for a comparo, if you'd like.

[Dreadhead]

I look forward to hearing how this comes out. I did the HDCD software decoder and even modified EAC to include it in my ripping process but I don't really bother anymore.

I could also provide you guys with the HDCD wav and the non HDCD decoded wav if you are so inclined and you can do a DBT between the two using a HDCD enabled DAC (if it also accepts 24bit input) if you are so inclined. I don't have an HDCD dac myself so no luck here.

Any interest? in seeing the size of the differences? I could do some math....

[Dusty Chalk]

That's a good idea, but I can probably come up with comparo files myself (the executable I can download again, it's the source code that's harder to come by).

What numbers were you going to run? I remember that bothering me...I think it was something like: if you ran a non-HDCD-encoded audio file through hdcd.exe, it still generated the larger file. In my head, that didn't make intuitive sense -- a 24-bit file with the bottom bits not dithered is not the same as the 16-bit file. Is it? Makes me wonder if you have the same problem with HDCD-encoded audio -- I.E. if it should only be a 20-bit file, what are you doing with the bottom 4 bits?

But then again, he said he reversed it from Windows Media Player, so it's probably just what Microsoft is doing.

[mypasswordis]

Wow this is turning out to be more involved than I thought. (Having other people) doing the math is cool, and I guess so are DBTs.

Why did you sell your Parasound?

[Dreadhead]

That's a good idea, but I can probably come up with comparo files myself (the executable I can download again, it's the source code that's harder to come by).

What numbers were you going to run? I remember that bothering me...I think it was something like: if you ran a non-HDCD-encoded audio file through hdcd.exe, it still generated the larger file. In my head, that didn't make intuitive sense -- a 24-bit file with the bottom bits not dithered is not the same as the 16-bit file. Is it? Makes me wonder if you have the same problem with HDCD-encoded audio -- I.E. if it should only be a 20-bit file, what are you doing with the bottom 4 bits?

But then again, he said he reversed it from Windows Media Player, so it's probably just what Microsoft is doing.

If I remember correctly HDCD pads to 24bit regardless of there being HDCD or not and hence you end up with a larger file. (I'm not to sure about this one, nearly 50/50).

The most HDCD compliant way is to rip the disc and then use a WAV output plugin in WMP because then we'll know for sure it has all the HDCD features. I don't know if I care enough though.

As far as math goes I'd take the two files into some Matlab equivalent (name is escaping me, or if I do it at work then in Matlab), normalize based on the max value and do a difference and then see how much different there really is between the two in dB. My guess is not much but around some of the peaks with peak extension then probably.

Sounds to me that using the peak extension is a bad idea but the dynamic range suppression sounds pretty good for lowering the dither floor but at 16 bits that floor is really far down there.

Edited April 7, 2011 by Dreadhead