The Future of High-Resolution Computer-Based Audio, Andreas Koch

As part of our Panel discussion at the NY Audio Show 2012, I Want To Take You Higher – The Present and Future of Digital Music Delivery and Playback which also featured Rob Robinson (Channel D), Larry Ho (Light Harmonic), and David Chesky (HDtracks & Chesky Records), Andreas Koch of Playback Designs was kind enough to share this presentation.

After the seminar, an attendee asked if it would available online and Andreas was kind enough yet again to let us reprint it here.

Of course these PowerPoint slides benefit from having Andreas present them with his expanded commentary. As it is, they offer attendees of the seminar a great reference/reminder and for those who did not attend they offer a handy introduction to many important topics. Enjoy!

COMMENTS
fufanuer's picture

What's wrong with TosLink? It should have listed reasons. TosLink has got to be better than a 1/8 (or 3.5 mm) analog jack, right?

Michael Lavorgna's picture

Directly after Andreas’ presentation. To summarize (from memory which has been known to be leaky), Toslink shares some of the minuses of AES/EBU and S/PDIF including limited sample rate (up to 24/192 max) and no DSD support. To these Toslink adds the further minuses of its relatively fragile composition (and plastic Toslink cable can introduce errors), and it is the only “link” type which does not allow the DAC to act as the master clock.

Essentially the point was USB is preferable since it offers asynchronous communication allowing the DAC to control the clock, and it handles any PCM resolution as well as native DSD. FYI - Playback Designs is one of the main companies leading the native DSD cause, so to speak.

Larry Ho's picture

The only good thing for Toslink is, it is galvanically isolated. So there will be no ground loop or noise transmisted from digital source to DAC. The problems are, 1) the speed limit there is 96K/24bit. 2) Toslink shares the very similar internal protocol of AES/EBU or SPDIF. The DAC needs to "re-discover" the clocks by PLL circuits and that creates more jitter than necessary. Asychronous USB transmission doesn't have this fall back becasue the master clock sits on DAC side.  3) Toslink use optical elements that is normally much slower than direct connected eletronical components. So it further increases the jitter than SPDIF or AES/EBU.

Michael Lavorgna's picture

For that clarification.

kavon yarrum's picture

Larry, TosLink is very much capable of 192Khz. I have several DACs on hand that accept 192.

Secondly, I agree that the major benefit of TosLink is it does not pass on any internal noise from the source component.

Lastly, upsampling/reclocking should eliminate any issues with jitter. My Bryston DAC has optional upsampling that is integer based...44.1 goes to176.4, 96 goes to 192 etc.

Your thoughts.

A side story: a few years ago I asked three very well known cable designer what digital connection was best. I got three separate answes...coax, optical, and AES/EBU.

Larry Ho's picture

Thanks for sharing

Let me do some calcuation. 10M Hz for fine quartz fiber, 5M~6M Hz bandwidth for plastic fiber is the theorical limit for TOSLINK. 

For SPDIF(AES/EBU is similar). We use Biphase Mark Code inside, so 96K/24Bit PCM * 2 channels will equals to roughly 4.6M Hz bandwidth here. 

So the 96K/24Bit limitation should apply to plastic fiber (the common one). And as you said, 192K/24Bit should be achievable if use finer fiber. 

---------

Another thing I want to emphasis here: No matter what physical interface, BNC/RCA/TOSLINK...... The internal protocol involves "Clocks recovery" from the source. It means the receiver (the DAC) can only get the clock signal from the source (the transports or music server)... That is NOT the best situation. The master clock should sit on DAC side to minimize the potential jitter. So that is the reason why asychronous USB 2.0 or Firewire interface should be better in this perspective.

---------

deckeda's picture

Yes, the Toshiba Link can do 24/192. My tc electronic Impact Twin audio interface only does up to 24/192 out of its Toslink jack. Its digital coax RCA jack only does up to 24/96 (why, I don't know. Ghost of Sarnoff? Nah.)

Lately, with the audible (they claim inaudible) watermarking UMG is alleged to be doing on *all* digital distribution (including lossless FLAC) I'm increasingly leary of downloads, overall. As the first slide here shows, labels are waaaay too tempted to futz with digital files in whatever way suits them.

On another front, I walked out of a thrift store today with 32 LPs @ 50 cents apiece. Nothing particularly special about that price, except that they were a treasure trove of Angel, London ffrr, Everest and early Mastersound classical titles in superb condition. As was the more pedestrian Leo Sayer, just because I "feel like dancing". As a bonus I also now have two LPs in decent shape from 1949! (not Leo Sayer, that would be more like 1976. Bell bottoms. Nevermind.)

There's no standards with downloads, even if PCM or DSD "wins." At least with the Sony debacle some years ago concerning audible watermarking of some CD titles, we could easily point out that the watermark fell outside of the RedBook standard, and hence was "off limits" morally, if not legally.

One thought about DSD, since Koch's team is understandably pro-DSD:

The file sizes and expected download times for 2.8MHz DSD are roughly in the ballpark of PCM's 24/96. He has to realize this, and that 24/96 has become a sort of defacto "format" if you will, in many ways, for years already. But if file size and download times are a DSD main "feature", that race is already lost since decent broadband adoption is happening faster than the 24/96 juggernaut can be overturned.

kavon yarrum's picture

Woah, that was a frantic post.

Yes, TosLink does 192. And more note about it...that is the way "pro" SACD players allow for DSD output.

Secondly, DSD files are at least a third or more bigger than 24/96 PCM files.  

Of course Koch and team are fighting for DSD, he helped develop it and has a stake in it.

It is rather an absurd conversation to discuss which will "win". Unless you listen to exclusively classical or some of the most esoteric music around, there will be a tiny sliver of true DSD music for downloading. Outside of some of the classic catalogs of Dylan, Moody Blues, Stones, Elton John, The Police, and a handful of others, there is not much on the rock front.

And there are almost zero NEW rock or pop DSD recordings. The recording system is way, way, way too expensive and cumbersome. Even major bands like u2 still record at 48 Khz, 24 bit. 

DSD is an absolute pipedream for any one who is not a classical music fan. The fact that Koch in his presenation says DSD may become the "dominant" download format shows he lives in an ivory tower. Last time I checked there are not even 44.1 FLAC files for download for 99% of the recorded music in history.

 

kavon yarrum's picture

Audiophiles are becoming obsessed with numbers again....

"My files are double the sample rate as yours!!!!!!!!"

The music sucks, but dammit, it sounds good!!!

Vade Forrester's picture

Is that the entire presentation? While it has lots of good info in it, I'm surprised a presentation about the future of computer-based audio didn't talk more about the future. It stopped with a slide about asynchronous USB, which is today, not the future. All it really talks about that's futuristic is DSD files. Maybe that is the future of computer-based audio-I'm sure Koch is pulling for it. It would have been interesting to hear if there's a future for higher sampling rates and greater word sizes for PCM files. Yeah, I know it's possible now, but will there be a commercial market for it? What about the I2S interface-any chance that will be standardized?

 

Vade Forrester

Michael Lavorgna's picture

Of course these PowerPoint slides benefit from having Andreas present them with his expanded commentary.

Again, the main reason I published these notes was at the request of one of the attendees.

Vade Forrester's picture

I appreciate being able to review the slides-kudos to the attendee who asked for them. It's nice to learn where a leading industry figure thinks the field is going. I wonder if the expanded commentary would have addressed any of the areas I mentioned, or if Koch's current focus on making downloadable DSD files available and playable is where he concentrated the presentation. I hope he succeeds with DSD, but I wonder if a competition between DSD and PCM would hurt overall sales and hardware development like the SACD/DVD-A type competition did.

 

BTW, nice show coverage, Michael.

 

Vade Forrester

Michael Lavorgna's picture

I hope he succeeds with DSD, but I wonder if a competition between DSD and PCM would hurt overall sales and hardware development like the SACD/DVD-A type competition did.

The most recent paper from Andreas Koch and others (see link) is titled, “DoP Open Standard, Method for transferring DSD Audio over PCM Frames”. As you’ll see, it is exactly what is says – they’re defining a method for serving DSD over PCM and they’re making this available as an Open Standard.

In terms of Andreas' expanded commentary, I opened up the floor for questions as soon as he finished, Andreas was the last presenter, and the topics quickly diverged into other areas. So while he did provide more thoughts during his presentation, there wasn't an opportunity for any focused follow up.

I had intended to get into "The future..." after an initial round of questions but we never got there. We easily could have filled an hour with any one of the people on the panel or just a Q&A session.

Sjeb's picture

I'm curious what was discussed on the panel about thelack of reliable info about the source of the masters and how they were converted. Speaking for myself, I've slowed down on buying hi res files from HDtracks an others over this issue. 

 

Any chance the panel was recorded?

Michael Lavorgna's picture

Yes, there were a number of questions relating to HD downloads and David Chesky spent a fair amount of time answering them as well as providing additional information. To paraphrase from memory, David pointed out that HDtracks is a digital music distributor so they are not responsible for, nor are they able to dictate the mastering/re-mastering process. At best they can suggest 'best practices' to the record companies they work with. Wether or not they listen is another story.

David also said that as sites like HDtracks sell more music, they gain more influence. The largest issues come about when dealing with larger record companies because HDtracks is then dealing with potentially dozens+ of record producers each having their own ideas about the recording process. And the more popular/successful the band/producer, the more likely they will do things their way even if that means squashing the life out a recording with dynamic compression.

Finally, at least for now, David mentioned that HDtracks uses a number of outside facilities to test what they're given to make sure it is what its supposed to be. Again, all this is from memory and I was focused as much on my moderating duties as I was on listening.

Unfortunately the seminars were not recorded. Perhaps I'll ask David if he'd be interested in a Q&A since this is such an important topic.

Terry Bleakley's picture

As there is presently no way of confirming the history of a high res file, I too am leary of buying them.  I have visited the HDTracks site.  Everything is always presented as state-of-the-art, with no disclaimer as to the history of the files.  I would be surprised if all of them were from true hi res sources, especially many of the older recordings.  Who knows what was done in the studio and in the subsequent process leading up to the final masters.  I would like to hear more from David Cheskey about how they verify the integrity of the source material they use for their files.

Michael Lavorgna's picture

When available.

Here's one example from their website for the new Bob Marley Legend release:

Mastering was completed by Kevin Reeves at Sterling Sound NYC, using the original U.S. and U.K, album masters from the Island Records vault. The masters were played on a modified Studer A820 with Wolke Butterfly heads and converted to digital at *192khz/24bit resolution using the DCS 904 converter and Sterling’s proprietary mastering systems. As always, the most direct signal path was maintained throughout the mastering process.

And for The Rolling Stones ABKCO releases:

"Many years of research went into locating the original mono and stereo analog tapes that would be used in ABKCO's Rolling Stones Remastered Series. That research revealed a treasure trove of first generation tapes - true stereo masters from The Stones' 1964 Chess Studios sessions including the unedited version of "2120 South Michigan Avenue," Beggar's Banquet at its correct speed and Let It Bleed with splicing that indicates that the original intention was to leave little spacing between each cut.

For the analog to digital transfers, vintage reel-to-reel tape machines were utilized - a modified Ampex 351 with original tube electronics (full track mono and two track stereo) and an Ampex ATR-102 modified with Aria Discrete Class-A Electronics (full track mono and two track stereo). A Sonoma DSD digital audio workstation was the chosen high resolution format and Meitner Design ADC8 and DAC8 MKlV converters were used for the conversion process. Cables used were the cryogenically frozen type supplied to us by Gus Skinas of Super Audio Center. Gus also provided much guidance to Jody Klein, Steve Rosenthal and myself for our first time use of DSD technology. For this HD Tracks release, the Bob Ludwig mastered DSD files were converted to both 176.4kHz and 88.2kHz high resolution PCM with Weiss Saracon conversion software."

Sjeb's picture

Yes, I saw the Bob Marley info and was happy to see it, but that is the exception rather than the rule. I understand they are only the distributor, but that's not a basis for lack of information. Maybe when their source refuses to reveal sources and process, HDtracks should say so?  That may be bad for sales in the short term, but they should consider the long term goal of building a loyal customer base

After buying about a dozen albums from them last year, right now I'm very hesitant. I don't mean to pick on Chesky, they're just the best known site.

Michael Lavorgna's picture

We're asking a distributor to help shape and define a market, i.e. create standards, as well as police every album to make sure it meets these yet-to-be-defined standards. That's asking a lot, imo. I'm not saying its unreasonable, but this a new business model that is  being defined as we go.

From my perspective the only way this can happen is if the distributor(s) has the leverage to do so and the only leverage that counts is sales. The impression I got from David Chesky is we're getting there but we ain't there yet (I'd stress that's my take on things).

reeltoreels's picture

Exactly, who's to know? Even if you did a frequency analysis it doesn't mean anything. Really, unless you can A/B the source vs the HDtracks release there is no way to truly know. Don't believe the hype. I have heard a master recording (tape master clones) or two or 3 that have been used for the source material in digital download by a smaller less known site and they are markedly different in many ways - I am not going to state exactly how they are different. It may or may not be what you think, but I think that for me, personally there is enough room to manipulate the sound and frequencies and so on that you really either need to compare the 2 side by side. Also, I know from being a tape-trader for 25 years that lineage of a particular recording is pretty much a given to include when you trade tapes, right? Why this sort of "lineage" is not disclosed by these companies selling such downloads, to me, takes away from the autenticity of it all. However, I let my ears be the judge in the end. If I like how it sounds, then great. But I know that it does gnaw at most audiophiles not knowing all the details!

X