HDCD - High Definition Compatible Digital

[Disclord]

Ssully responded to some stuff I wrote about Pacific Microsonics/Microsoft's HDCD process in the "Tomlinson Holman" thread and rather than derail it with further HDCD discussion, I thought I'd make my reply as a new topic. If anyone has anything to add or just questions about HDCD, please jump in - so many topics that seem ripe for lots of in-depth discussion seem to die out too quickly.

With all due respect to the folks at Pacific Microsonics, such claims for HDCD have never been verified by proper psychoacoustics research, AFAIK -- not to mention the controversy around the idea that we need/hear more bandwidth than 20kHz. Moreover, AIUI, reconstruction filtering was never fully implemented in the consumer chips, and there was only ever one filter available to us (see posts by Charles Hansen of Ayre here) . And too, modern audio can easily be recorded, produced, and delivered at 48, 88, 96, or 192 kHz, which is plenty of bandwidth even if 44kHz seems restrictive, don't you think?

AIUI, anything that goes through an HDCD ADC gets an 'HDCD' bit set -- whether any of the 'audio enhancing' mastering options are used or not. So there are HDCDs out there that are only 'HDCD' because they used Keith Johnsons' (admittedly fine) hardware, without utilizing the (supposed) benefits of filter switching , peak extension...

Reading the link you provided really opened my eyes - I mean, I always knew that HDCD was over-hyped - but I had no idea how little its actually used on discs that light the HDCD light! In the beginning the Reference Recordings/Pacific Microsonics people were very good at filling Stereophile and The Absolute Sound with pages of sonic superlatives but without revealing ANYTHING about what HDCD really did. At least they were keeping silent, unlike Sony who, around the same time and with full knowledge of what they were doing, seriously misrepresented Super Bit Mapping to Stereophile and The Absolute Sound - Sony knew both magazines would eat up the 'sounds better' and 'gives more than 16-bit resolution' claims and had no staff with the technical knowledge and could understand what was really taking place.

What Pacific Microsonics didn't realize was that while a patent application is a protected document in America, at the time in Germany they were all laid open for the general public - and the info got out. HDCD was, at its most "basic" level, a noise shaping and noise reduction system controlled by a pre-computed hidden bit for decoding instead of the signal itself. HDCD also had the option for some really bizzare signal synthesis that created totally artificial waveforms and placed them where supersonic frequencies might be. Although the waveform synthesis part of HDCD was present in each patent for HDCD, it was never implemented and the HDCD chips didn't even have the capability to do it.

Do you remember the mini scandal when Audio Magazine did some in-depth tests of HDCD VS non-encoded discs on identical recordings? The conclusion was that HDCD had been seriously misrepresented and instead of improving accuracy, actually degraded the original master tapes - the non-processed CD was a more accurate replica of the master. Keith Johnson and Pflash Pflaumer (SP?) both wrote letters to the editor and ended up saying different things, contradicting each others claims! It was a huge mess for Pacific Microsonics, and kind of a shame because regardless of the benefits of the HDCD process, the Model One and Model Two processors are incredible sounding units - the DTS LaserDisc of EVITA was encoded from 6-track mag Dolby SR master to 20-bit 44.1 kHz digital with three of the Model Two processors (no HDCD 'encoding' was used and DTS didn't have a method to convey the HDCD bit to the HDCD decoder at that time) - the Dolby Digital master for Evita, used on both the Criterion AC-3 LaserDisc and the Disney DVD, was taken from the same exact 6-track SR master tape (and played back on the same aligned dubber) but for some reason, unknown 20-bit converters were used for the A/D conversion. I say "unknown" because I've never been able to find out what converters were used - even Dolby couldn't tell me, but they were NOT the Pacific Microsonics. Masters for both were stored on the Tascam DA-88 format with the bit-splitting adapter to allow 20 bit resolution on the 16-bit machines. DTS did the encoding for Evita and Pioneer did the encoding for the Criterion LD (and Disney used the same AC-3 encoding) That difference in converters is audible - the DTS LD is head and shoulders above the AC-3 discs - and I don't believe its due to the AC-3 algorithm because the Titanic AC-3 LD sounds EXACTLY the same as the DTS LD, both of which used the same digital master tape.

I still think the digital filter switching would be useful (if it had been fully implemented - I never knew it hadn't!) - it would be useful for the DTS Core signal of a 96/24 encoded recording - to give some benefit to listeners that don't have a DTS decoder that can decode the 96kHz encoded signal.

One surprising thing that I've always noticed is the number of stand-alone CD/DVD players that don't implement the Peak Extend feature of HDCD. So far, all receivers and stand-alone processors I've seen use the "full" HDCD process (whatever that means now), but many players, like my Panasonic DVD-S97 have a prominent notice that the HDCD Peak Extend is not available on that unit. So I wonder if that means the low-level expand is also ignored? If it is, then just WHAT is the HDCD process "doing" in those units???

One "good" thing about HDCD is at least they were not using incorrect decoding in their reconstruction filters, like Wadia Digital - and last time I bothered to look, Wadia was still using their incorrect "Frenchcurve" filtering which had worse low-level distortion than a 50 dollar Walmart portable CD player - Wadia gave about 10-11 bits of resolution and they charged over $2000 for the privilege of listening to that distortion too! But Stereophile and The Absolute Sound LOVE it! Pioneer even copied it - and got sued by Wadia - Pioneer called it Legato Link - now the Legato Link in current Pioneer products bears no resemblance to the original marketed version.

As a side-note, I don't understand the current love of oversampling and then omitting the output filter which allows all kinds of supersonic garbage to pollute the output of a normally sampled (44.1/48-kHz) CD or DVD's signal and intermodulate downstream. Now THAT will change the sound!

 

[Almen]

I can only agree with most of what you say. I, too, got a real eye-opener when discussing this issue here (in a long thread not specifically about HDCD), where ssully also straightened some things out.

Many strange acronyms has occurred in the name of "High Resolution Digital Sound"...

 

[jdmack]

I haven't read the other thread, so this may have already been mentioned, but mastering engineer Steve Hoffman recently posted a thread on his forum entitled "I've decided not to run my masterings through an HDCD converter."

http://www.stevehoffman.tv/forums/showthread.php?t=246642

J. D.

 

[Disclord]

I can only agree with most of what you say. I, too, got a real eye-opener when discussing this issue here (in a long thread not specifically about HDCD), where ssully also straightened some things out.

Many strange acronyms as occurred in the name of "High Resolution Digital Sound"...

Almen, thanks for pointing out that thread - I had not seen it before. Yet more eye-opening information for me. But, that's good - I'd rather have correct information about something than be 'right' - in other words, I'm not looking just to have my preconceptions and biases reinforced.

I have an HDCD sampler (well, had an HDCD sampler - it got scratched and wasn't recoverable) from Reference Recordings and its taken from Keith Johnson's superb analog recordings - you can clearly hear the hiss of the non-noise-reduced analog master tape, which implies much less than 16-bit resolution in the master right there. Also, Johnson's recording/microphone methods seem to be very, very 'wet', awash in reverb, so it's hard to tell just WHAT is being heard - and without a non-encoded duplicate from the same master, made at the same time with the same converters, at the same levels, all bets are off in saying which is superior. The Reference Recordings LaserDisc had just such a comparison for the CX Noise Reduction System where it was turned on and then off in the middle of a song - it used the auto CX switching that the non-tube-laser players employed so the listener was never aware of exactly when the system turned on and off - only a card on the screen saying it was on or off - the precise frame-number moment was never revealed. That should have been done with HDCD on the later Video Essentials LD that Reference Recordings produced - and they could have done it on the DVD release too.

I'd love to hear the 'real' HDCD process - the one in the patents, with the supersonic waveform synthesis, reconstruction filter switching, etc... Keith Johnson is a brilliant engineer - heck, he helped found Gauss Electrophysics that made high-speed magnetic tape duplication possible and in partnership with David Paul Gregg they filed the first patents for what would become the optical Laser Videodisc. Of course their patents were completely unworkable and more of an 'idea' for a flat disc that spins, is played back via a light of some sort and plays a movie or audio. It took MCA Universal Studios and a small band of engineers, with no reference to the original Johnson/Gregg patents, only 3 years to go from 'idea' to a replicated disc and showing for the press in September 1972. Add a few more years of R & D and bring Philips on as a partner to manufacture the players and we get the LaserDisc from MCA DiscoVision in December, 1978. And, of course, the Philips CD used all of the LaserDisc's specs, such as pit geometry, disc thickness (LD's were 2 single sided discs bonded together), etc... and Sony's error-correction know-how - and their (thankful) insistence on 16-bit encoding instead of Philips preferred 14-bit (Philips thought an accurate 16-bit DA converter was too expensive to make).

Gee, can you tell that the early history and development of the LaserDisc is one of my obsessions?

Anyway, back to HDCD - apparently, Microsoft wanted it to use the buried data channel for copyright and licensing control. I've been told that some of their old Windows Media HD downloads had the buried HDCD bit in the video, used to convey playback licenses to the Windows Media Player - but that might just be rumor - I 'heard it on the net' and you know how reliable that can be!

One technology - and I'm not sure if it's used much anymore on CD's - that really did 'work' and improved the sound quality from standard Red Book CD releases was Apogee's UV22. While it was just basically a sophisticated form of dither, it really did lower the noise in the sensitive midrange area, giving the subjective 'performance' of 18-bit PCM. For a while I was seeing the UV22 logo on quite a few releases - it was even used on a Lumivision LaserDisc ("The Dream Is Alive" reissue in the gate-fold jacket) and the "Terminator 2: Special Edition" CAV Box Set - and I 'think' I remember reading, perhaps in Widescreen Review, that the "Star Wars: Definitive Collection" box set used UV22 as well. The "Star Wars" set was the first to employ Analog Devices asynchronous sample rate converter (to re-sample the 48k D1 tapes sound to LD's 44.1k CD standard PCM) - but I can't say that I've seen it listed for years - maybe it's still being used but just not marked? It's a shame it wasn't used on more LD's, especially films like "The Abyss" with its awesome dynamic range. NOTE: Looking at the Apogee website just now has answered my question about UV22 being used nowadays - it appears that UV22 is still widely used and has been updated and improved to optimize a version of it for low bitrate sources such as MP3 and AAC encoded audio. And many, many studios employ it routinely as part of the mastering process - of course, that's all according to Apogee.

Isn't high-tech audio fun???

 

[Almen]

Yeah, the fact that oversampling, noise shaping, and intelligent dither can be used to improve the 44.1 kHz/16 bits CD standard is cool. I am still grappling with all this, and slowly but surely getting a grip on what makes a difference and not.

There's so much to learn...

 

[Disclord]

First off, here's a link to an excerpt from the magazine The Audio Critic and their comments about HDCD when they reviewed the EAD DSP-7000 outboard D/A Converter. It was their first review of a component with HDCD built in and the writer, Peter Azcel and the magazines E.E. David Rich, made some general comments about HDCD. I've put them together onto a single page, instead of scanning the entire review. If anyone wants to read the review, all back issues of The Audio Critic are now available as PDF's for free at: http://www.theaudiocritic.com (click "back issues" at the top of the page) current magazine is available for free under the link "Web 'zine" - here's a direct link to it:http://www.theaudiocritic.com/plog/ A few years ago The Audio Critic went web-based and became free. Peter Azcel was, apparently, getting too old to continue on with a full print publication and having a hard time finding quality writers who didn't believe in the nonsense that Stereophile and The Absolute Sound promote.

Anyway, to further our discussion about HDCD, here's the link to my PDF cut-n-paste of The Audio Critic's thoughts about HDCD: http://issuu.com/Disclord/docs/tas_ead_hdcd_comments?mode=embed&viewMode=presentation&layout=http%3A%2F%2Fskin.issuu.com%2Fv%2Flight%2Flayout.xml&showFlipBtn=true

I've replied to Almen just below.

Yeah, the fact that oversampling, noise shaping, and intelligent dither can be used to improve the 44.1 kHz/16 bits CD standard is cool. I am still grappling with all this, and slowly but surely getting a grip on what makes a difference and not.

There's so much to learn...

Yes, there is a lot to learn - and so much of it goes sailing right over my head - but I've found that if I read, and re-read, and ask lots of questions online (and email specific technologies inventors, if they are still alive and online) then read everything some more, I eventually begin to understand it. I might not be able to "do the math" of something like advanced matrix algebra and vector spaces, but I can comprehend what's going on and what the circuit is doing. And Wikipedia and other sites have been a Godsend in being able to quickly explain something that might be hanging me up when reading a technical paper, etc... So I have lots of fun learning about it all - and I have even more fun discussing technology and debating its pros/cons - and meeting others who know much more than I do and learning from them! As well as helping others who are new to, say, quadraphonics, and helping them negotiate the minefield of formats, decoders, etc...

The hard part is finding people who actually want to (and have fun) discussing such things... Luckily, QQ is an exception.

BTW, the site where I've posted my Audio Critic HDCD PDF also has all kinds of tech papers and stuff that I've posted - and will keep posting more of. The URL of my 'main' page is: http://issuu.com/disclord and if you join the site (which is free with no strings attached) you can subscribe to my uploads and will be notified whenever I post a new technical document, brochure, manual, AD, or whatever. And it's all organized by subject, like CBS SQ, Sansui QS, Circle Surround, Equipment Reviews, etc... I think the folders for each subject show up when you subscribe to my profile. I don't want anyone to think I am in some kind of competition with OldQuadGuy and his wonderful Quadraphonic.info site, because that's not the case at all. Me posting on ISSUU is just a way of getting more quad/surround/information out there for everyone to read and have available.

 

[Almen]

Your site seems like a cool idea! I have some articles I could share.

I tried to sign up, but there was some sort of hangup. I'll try again later.

 

[Disclord]

Your site seems like a cool idea! I have some articles I could share.

I tried to sign up, but there was some sort of hangup. I'll try again later.

www.issuu.com is a neat website - especially since they give 15 gigs of storage for free. Be sure to sign up as an individual, not a business - that ensures its free. And they don't have any "catches" like all kinds of SPAM email or forcing you to read/watch an advertisement before being allowed to view a document. I'm glad I found a place to make all the info I've collected available. I've got probably 150 more papers to upload and many to scan and make PDF's of.

 

[jefe1]

Thanks for posting your articles online, Disclord. I think its great that you're willing to share your knowledge, and putting in your own sweat to make it happen. I think everyone benfits from having alternate locations to find information that they can benefit from. I think your site is complementary to old quad guys. His site is filling withmore technically based materials like service manuals and user guides for quad equipment while your documents cover more movie oriented surround themes and technology.

I notice that Neil Young started using hdcd on his music. I think only my oppo 970 detects it. But I've always been unsure what it brings to the table regarding improved sound. I just trusted Neil Youngs integrity thinking that he felt his music benefited from it.

 

[mandrix]

Interesting. I decided to dig into whatever I could learn about hdcd some time back but it soon became apparent it was a case of the more I knew, the less.
Thanks for the discussion.

 

[Disclord]

Interesting. I decided to dig into whatever I could learn about hdcd some time back but it soon became apparent it was a case of the more I knew, the less.
Thanks for the discussion.

Was it the more you knew, the less you wanted to know, or the more you knew, the less you did know?

I'm in both camps now - while I've always agreed with the position stated in The Audio Critic that the 16-bit format isn't really in need of 'fixing' (it needed true 16-bit accuracy*, which it is getting now thanks to higher resolution recording) but the idea of using a companding system with a digital control, as HDCD does, made sense. And it could allow 16-bit CD, DAT and LD to truly have greater dynamic range, if the recording needed it (movie soundtracks from 35mm Dolby SR multi-track mag masters often have +18db or more headroom over the hard digital 'limit' of +20db above the 0-db reference level of 16-bit PCM, necessitating compression when transferred to D1 tape for LaserDisc's digital audio tracks) [Side-note: the analog FM tracks on LD were given +9db headroom like Dolby SR Optical prints and VHS/Beta Hi-Fi were given +6db like Dolby A-Type optical prints - VHS/Beta could have had more headroom but it was held back with the thought that most people probably didn't have equipment that could handle it.]

So, in my mind (and prior thinking) HDCD would be useful for allowing 16-bit digital to more faithfully handle the peaks that were actually on the master tapes - and do it compatibly, since those without HDCD would have a gently limited playback and those with HDCD would get full dynamic range. Plus, the reconstruction filter switching that I thought was always present made sense because it supposedly optimized the filter from moment to moment based on the music to give the perceived effect of much greater high frequency bandwidth. But, alas, that's not true either... I'm glad I've found it out though - I'd rather know the truth than be some deluded Dorothy tripping down the Yellow Brick Road. For some reason I can't fathom, some people LIKE to not know the truth about something - they find some 'comfort' in their misinformation. I don't get that.

* Brad Miller - Quadraphile and founder of Mobile Fidelity and Miller Nevada's High Definition Surround DTS CD releases, for those who don't know who he is - always said that 16-bit audio and 44.1kHz sampling were NOT the problem with CD's and digital in general. And his COLOSSUS digital multichannel recorder, that was custom built and invented by Lou Dorren, proved it. The standard 16-bit CD's made from his COLOSSUS recordings were stunning, with levels of sound quality that I didn't think the CD was capable of. Some of his HDS label DTS CD's were made from COLOSSUS masters and you can hear what an amazing system it was recording multichannel. Especially his recordings of Space Shuttle launches on the "Surround Sound For The Millennium" DTS CD - also on the same CD is a COLOSSUS recording of a snowfall in the forest - it's amazing - the quietness is eerie and you can 'feel' the cold - it's that realistic. Brad said the 'secret' was that all 16-bits, in both recording and playback, were accurate at all times - and even today that's not usually the case. 18-bit and higher sound 'better' when playing 16-bit recordings most of the time because, since they have more bits, they get those first 16-bits that CD's use correctly converted right the from the start. They don't always do so hot when called upon to reproduce true 18-bit and higher recordings though. The reconstruction filters on many units aren't all that hot either. On my old 2nd Generation Sony 7700 DVD player you can turn off the output reconstruction filter and enjoy the full unfiltered quantization noise - I have no idea why they would do that.

I may be mis-remembering, and I wish I still had a copy of the paper, but I think it was Professor Stanley Lipschitz who issued an AES paper explaining that many reconstruction filters on CD players, DVD players, etc... were not doing it right in one way or another, and that it was clear that the filters designers didn't fully understand sampling, quantization or reconstruction theory. (It was Stanley Lipschitz who blew the lid off the fraud of SACD - he showed how 1-bit DSD recording/playback can never, ever be correct - it will always have major errors that cannot overcome and the noise shaper will always be in overload, unlike linear PCM, which is completely perfect able; I'm not claiming that all PCM recordings are perfect! Anyway, Sony and Philips were then forced to admit they had stopped recording in true 1-bit DSD shortly after SACD's 2 channel launch in 1999, and had been using an 8-bit MASH-type system ever since. Plus, true 1-bit DSD recordings cannot be mixed or edited in the 1-bit domain - I think those are two very large reasons why Sony and Philips dropped SACD so quickly)

One thing regarding HDCD - would anyone be interested in reading Pacific Microsonics original AES paper on HDCD, as well as the tech-note explaining the "gain scaling" required in all licensed HDCD decoders? If so, let me know and I'll post it to my file site as well as some of the patents.