Saturday 6 April 2013

MUSINGS: On SACD & DSD audio...

SONY multi-page spread (with hybrid SACD 'RS500' sampler!).
Rolling Stones magazine, December 2003.

Okay, now that I have posted a number of measurements over the last months, permit me a moment to play "Speakers Corner" and talk a little about the state of affairs around DSD now in early 2013...

Firstly, remember that DSD isn't anything new. It's basically a digital format based on Delta-Sigma modulation which has been used in consumer digital audio since the late-80's and early 90's - remember those old Sony 1-bit and Panasonic MASH players from around that era? It was said that as the CD patents were set to expire, Sony and Philips decided to create a new disk format for the 21st century - hence the birth of SACD around 2000 using the "1-bit" encoding method instead of multibit-PCM carried on DVD-like disks capable of higher data capacity (multi-bit ended up being DVD-A of course). Sound quality was said to be better because of things like "100kHz frequency response", better noise floor and in time, multichannel. Of course in creating this new standard, copy protection was job one and they did a great job. I remember in 2004 visiting China and seeing pirated DVD-A's which were easily "cracked" soon after release, but not so SACD's.

It wasn't until 2011 when digital "ripping" of SACD's became a reality with the PS3 techniques easily found on the Internet. This finally allowed hobbyists to have a good look at the digital data "under the hood". What was seen in many cases was unflattering... The controversy around Norah Jone's "Come Away With Me" back in 2004 having been just 16/44 converted to DSD for the SACD release was really the tip of the iceberg. In fact, many albums were given the same treatment.  These "SACD's" were nothing more than just 16/44 PCM transcoded with no possible sonic improvement whatsoever - in fact, since the conversion process is not perfect, it's probably preferable to just buy the CD and enjoy technically 'better' data and save the cost differential (unless the SACD has a multichannel mix which could represent some extra value).

In terms of my personal experience with SACD, I bought my first SACD around 2001 and over the years have heard a number of high end models including the first Sony SCD-1, 555ES, 777ES, couple of Marantz, a Denon or two. I remember spending a whole evening back in late 2001 with the just-released Telarc 1812 Overture on hybrid SACD going back and forth between the stereo SACD vs. CD layers on the Sony SCD-1 (and later SCD-555ES) with a good pair of headphones late into the night. Despite the hype, my impression was that any differences were at best subtle.

As we all know, commercially, SACD was a failure. Sony ended their push into SACD disks around 2006 and likewise, most SACD hardware available in their product line these days exist as part of Blu-Ray players (how many actually still do "pure" DSD is unclear). [I was reminded by Geoff on Audio Asylum that the audio-only Sony SCD-XA5400ES is still available!] Newer versions of the PlayStation 3 likewise have lost the ability to play SACD.

On the software side, it's good to still see regular releases of SACD usually in the classical and jazz genres with some audiophile MoFi, Audio Fidelity goodies every once awhile.

In the last year, it's interesting to note the increase in DSD-capable USB DAC's (witness Benchmark's DAC2, Chord, Mytek, EMM, dCS, Esoteric, MSB, Meitner, etc... See List). It seems to be part of the feature-set du jour; perhaps much like the upsampling feature with various filter settings which started a few years ago. The fact is that many 24-bit DAC chips themselves have had the capability for DSD decoding for years now, but it only appears recently that manufacturers are adding this to the driver set since the DoP standard - and in doing so ticking off another feature on the product brochure.

But does this make any sense? Lets examine a few factors...

1. Is there any music available in DSD? The answer is of course YES. Thousands of SACD titles exist encoded as DSD64 in fact. But at present there does not exist a way to download this stuff legally. Many SACD rips are already out in the wild but unless you own a copy of the SACD, it would not be legal to obtain such copies... This even is contentious depending on the copyright legislation in your country since to get at the data, copy protection mechanisms have been circumvented which may not be legal.

2. Is the DSD music of technically good quality? What I'm getting at here is the concept of provenance of the music - where did it come from? There are of course small music studios like AIX or 2L where the music is recorded to the highest standards, but the vast majority of SACD/DSD music exists as analogue conversions of decades-old music, or if it's a modern digital recording, almost universally processed through the PCM route. Indeed, after more than a decade since the SACD, processing techniques beyond basic splits, fades, volume adjustments remain PCM-based. There really is no reason this will ever change.

As I noted above with Norah Jones as an example, there are MANY recordings on SACD which look like they're from a 44kHz PCM-sampled source. Here are just a few that I've come across over the years that look suspicious - MFSL Dead Can Dance remasters, Sony's Brubeck "Time Out", Baby Face "The Day", Uriah Heep "Magic Night", Ryan Adams "Rock N Roll", Joe Satriani "Engines of Creation", Albert King "I'll Play The Blues for You", Yo-Yo Ma "Soul Of The Tango", Blue Oyster Club "Agent of Fortune", "The Phantom Of The Opera" soundtrack... As far as I am aware, there isn't any database out there to keep track of this (nor do/should most people care at this point given the marginalization of the format). The point is, SACD's are "polluted" by recordings of "inferior" quality, many of whom look like they originated as transcoded CD. More discussion here.

3. Does DSD sound better? This is where we can get really contentious. Technically, as you can see from my Oppo BDP-105 measurements, DSD64 doesn't seem to offer anything that 24/88 PCM cannot already achieve. Over the years, Sony and others have advertised it as offering extended frequency response to "100kHz" (check out the table in this brochure), flat square wave response, or "perfect" impulse response free of ringing. Clearly the "100kHz" frequency response is misleading given the noise shaping going on in the ultrasonic range. You can read about the square waves in more detail here at Craigman Digital. As for impulse response, sure, it's good when you sample MILLIONS of times a second, but has anyone shown that the the human auditory system is capable of perceiving this in well conducted studies by correlating perfection of impulse response with superior quality? (Feel free to drop a reference!)

Also, there are instances where the SACD layer and CD layer of the same hybrid disk have different masterings, thus making it impossible to compare quality of the underlying technology (an example is Joe Satriani's "Engines Of Creation" - CD layer dynamically compressed badly at DR7 vs. stereo SACD layer DR11.)

As much as "audiophiles" claim otherwise, there is still no scientific research to show the benefits of hi-res audio from properly dithered 16/44 "CD quality" audio much less differentiate hi-res DSD from PCM (have a look at the Wiki for a few references on this). Remember also that we are at this point more than a decade since wide availability of SACD titles!

4. File formats: DFF (Philips) and DSF (DSF). In terms of having DSD files on your computer for these DSD DAC's, there are currently 2 common file formats out there. DFF unfortunately cannot hold metadata tags - this is very unfortunate. It makes it tough to catalog music and IMO excludes its usefulness in music libraries where one has hundreds or thousands of albums. DSF at least is better in this regard.

Yes, I know multi-terabyte HD's are cheap. But this is still no reason not to maximize storage space IMO and minimize transfer time when doing file copy or backing up drives. DST compression (lossless compression sort of like FLAC for PCM) is only available for DFF... Looks like for now you have to choose between tagging vs. compression of DSD files; and it's been like this for years so don't hold your breath.

Speaking of file size, DSD64 is about 680KB/sec (and DSD128 twice that at 1.37MB/s). Compare this with stereo 24/96 PCM at 560KB/sec (24/192 at 1.13MB/s) and you really wonder whether the storage requirements make sense for what you get in audio quality; especially when you figure in the fact that FLAC or ALAC or APE can generally reduce the bit rate further by 30-40% easily AND still retain full tagging function.

5. Economic incentive. Other than small audiophile specialty labels, selling DSD files doesn't really make sense to most companies does it? Storage requirements are higher as above, more bandwidth is needed for transmission over the Internet. You'd be selling to a vastly smaller number of installed DSD-capable DAC's (compared to every CD/DVD/Blu-Ray player, every DAC, every computer). There's no special Digital Rights Management so there's no copy protection. Physical disks are on their way out and since SACD's can now be ripped, good luck getting any of the major music audio labels interested.

Okay, time to wrap up... From personal subjective experience as well as thinking about this over the last while, I really see no reason for DSD to take off. Don't get me wrong, I actually very much enjoy my collection of SACD's but then I also enjoy good CD's, DVD-A's, Blu-Ray's, FLAC's, and high bitrate MP3's. Whether audible with real music or not, DSD64/SACD is capable of better-than-CD dynamic range in the audible spectrum which I believe is the main (only?) benefit in terms of stereo playback.

Other than as a feature to hype up the new DAC's, there really doesn't seem to be anything materially new with DSD DAC's. Certainly no "disruptive" technology has been introduced by this feature. IMO DSD64 doesn't technically measure or sound better than 24/96 (or even 24/88), the de facto standard for studio work for any new project remains PCM, and other than small audio companies targeting a specialized group of music lovers (where perhaps they can meet a reasonable financial target), it doesn't make sense for any major studio to invest capital into this "new" audio delivery format which already has failed to make a significant impact in terms of proven superior audible quality or economic viability even when it had full SACD copy protection, major marketing push (free Sony and Telarc samplers & promos back in the day!), and weak competition (errr...  DVD-A?). Remember, DSD has to compete now with hi-res PCM in a very PCM-friendly world.

In 5 years, once the "second coming of DSD" hype dies down with these new DAC's, I suspect we'll be looking back at DSD much like we see it today - a niche format, owned and loved by a loyal group of audiophiles probably consisting of mostly acoustic jazz and classical music fans (where maintaining DSD purity from recording to playback could make sense). We'll see...

Whichever way this goes, enjoy the music :-)

(Update: May 12, 2013 - just wanted to reiterate something I noted at the end of my TEAC UD-501 DAC DSD evaluation...)

Before I end off, I just want to make a general comment & plea about the state of DSD computer audio now that I've got a chance to try it out.

The DFF and DSF file formats are inadequate. Compared to FLAC, APE, or ALAC, these DSD file formats feel geriatric! Seriously, PLEASE get the file format right.

Firstly, we need good tagging features - all the more important for DSD since much of the excellent material consists of classical music where it's important to document conductor, orchestra, composers, title, year of performance and composition, etc... Please let me be able to use something universal like the excellent Mp3tag to manage all my PCM and DSD files.

Secondly, a standard DSD file format NEEDS lossless compression. DSD is extremely compressible - using DST, I regularly see compression ratios >2.5:1 losslessly, getting up to 3:1 in some tracks. This becomes even more useful for DSD128 where the space savings are very substantial. By doing this, DSD64 can be compressed to file sizes overall smaller than 24/88 encoded with FLAC with equivalent (some would say better) sound quality... I think we can all be happy with that! Lossless compression would also save file transfer times and cost of storage for the music producer, distributor, and of course consumer. Seriously, what other modern hi-resolution media format doesn't allow for at least lossless compression? Uncompressed multichannel 5.1 DSD would be a colossal waste of disk space IMO!

Over the months, I have heard DSD apologists talk about how you don't need compression or native tagging because "hard drives are cheap" and "JRiver can tag with its internal database". Sorry... That's not good enough. This is a foundational matter and will impact future generations of products, so it's important to get it done properly instead of rely on work-arounds.

Please guys, now that you've gotten together to define DoP and manufacturers to make these DAC's, lets get it done right with a standard DSD format that's fully functional, preferably "free" as in "open". Maybe some enterprising coder like Josh Coalson of FLAC fame can apply their expertise!

Thanks in advance. ;-)

Update: August 2013
Recently, I integrated the TEAC UD-501 DAC into my main system directly to the music server PC for playback. This allows me to play DSD (and PCM of course) files natively through foobar2000 or JRiver using my Android/iPad as remote. Working well...

However, one important limitation of DSD playback related to item 2 discussed above. Because DSD data cannot easily be manipulated through DSP algorithms, I lose the ability to run the DSD audio through a convolution filter for room EQ purposes. Since I'm limited with my room treatment options, this limitation makes native DSD playback audibly inferior. One way around this is to realtime convert DSD to PCM 24/176 and then run it through the DSP (this of course means it's no longer unmolested DSD)...  Maybe in the days ahead, I'll just upgrade my room ;-).

Addition: January 2, 2014
Just came across this from Mark Waldrep. Blunt piece on DSD vs. PCM.

13 comments:

  1. Large capacity spinning hard drives are indeed very affordable, yet many audiophiles have/want large capacity SSD's and although the price of SSD hard drives has come down, the 500GB and up SSD hard drives are still rather expensive for many audiophiles.

    ReplyDelete
    Replies
    1. And why would these audiophiles want or "need" SSD for just music storage?

      Delete
  2. (SSD: Playing it faster? ;) SSD has lower searching time than HDD, and it's awesome with zero heat/noise. Please note that SSD aren't optimal as a offline backup, they must be used regularly to keep the data safe)

    I didn't buy SACD earlier because there wasn't players who was cheap enoght, and now it's cheap multiplayers who can plays everything. That's much better than the ridiculous prices they had.

    I feel that Sony developed DSD to make more money, that's okay, but not so okay for the public need, the confusement in recording (DSD) equipment, the easy access, easy to backup and listen to it in other formats. I have only three SACD, and it make me wonder if these are fake; I will not buy more SACD, I almost bought Time Out for a few weeks ago, so I was glad that I waited. Thanks!

    http://www.phasure.com/index.php?topic=1510.0;all (in 2010)
    Tells about buying 24/96khz music:
    "I've also purchased from HD Traks Sam Cooke: Portrait of a Legend 24/96 and Oscar Petersons: Night Train 24/96 and they sound terrible."

    "Unfortunately the music industry is quite sloopy. Their cost control exceeds their quality control. Which sometimes means that "somebody" at the recording company just ship the Red Book master to HD Tracks. Shame on HD Tracks for not controling EVERY title they get. The frequency spectrum and compression rate of the mastertape can easily be analyzed with software such as Audacity, which even is for free."

    Seriously, the customers don't needs fake music!

    The even sadder things is that new music have low dynamicspace, the newest CD from Dido from itunes has ~6dB dynamic space. It may sounds great on the headphones or boomboxes, but it may be flat on a high quality equipment, I hjaven't measured yet how much it clips.

    ReplyDelete
    Replies
    1. Hello Atle,
      Yes, there are many benefits to SSD - lower power, faster transfer & seeks, silent, robust without moving parts... I would happily move to a 3TB SSD in time :-)

      BUT some audiophiles *think* they sound better. Another unsubstantiated claim which appears highly unlikely (assuming your hard drive is working properly and a recent model).

      I agree that hi-res PCM can be questionable as well. HDTracks at least in the past released a number of SACD converted 24/176 albums which IMO is a total waste of space. Personally, having bought about 7 or 8 24/192 albums off HDTracks, I have not found one instance where there was much above 40kHz (assuming there's even any positive benefit to keeping these frequencies!). As results, at this point, I see no point downloading anything more than 24/96... Might as well save the money for more/other music.

      Reduced dynamic range due to compression and limiting in the industry these days is a major problem for those who desire sound quality. When a recent vocal release like Michael Buble's "To Be Loved" (released last week on April 22, 2013) scores a DR6, with parts sounding badly congested on a good system, that's a sad statement of the current production "standard".

      When folks like Neil Young talk about "saving music" and sound quality, IMO he should just have a look at how badly the stuff is produced rather than harp about the horrors of MP3 or idealize 24/192 as if it's some kind of salvation.

      Delete
  3. Disagree here a little bit. Some of the best sounding recordings I own are DSD. Try Mahlers 1st, San Francisco Symph with Tilson Thomas. Recorded in DSD and probably the best sounding recording of an orhcestra I've heard. Similar quality from DSD recordings from Channel Classics.

    I also prefer well done DSD transcriptions from analogue. To me it just sounds more "natural" and more like the real thing.

    That said, I don't doubt that it is a niche medium, and if you don't listen to classical, it probably doesn't have much of a future. All the big labels have commited to PCM, so they won't be recording anything in DSD.

    With 4X sampling files, I don't think the only issue is if there are high frequencies above 40k. Many listeners hear improved sense of space (recording venue) and placement of instruments in space with the very hi-res recordings. This isn't due to hi-res frequencies per se, but better (more easily audible) transients, decays, and harmonics.

    ReplyDelete
    Replies
    1. Thanks for the comment and suggestion to try out the Mahler 1. Have heard some good things about it.

      Now that I have the Teac DSD DAC, I'll have more opportunity to listen to this material in the days ahead beyond the output from my SACD player.

      Delete
  4. I encode dff files to dst and store them this way. It saves disc space and lets me convert dsd to pcm 24/192 "on the fly".

    ReplyDelete
  5. Great article. Although, IMHO, we cannot just put our hearing in the 20Hz-20kHz box. Humans are not computers/robots. It is much more complicated than this. And we don't hear with our ears only. The whole body is involved.

    You were looking for some scientific research/evidence. How about this study

    http://jn.physiology.org/content/83/6/3548

    Enjoy the music.

    ReplyDelete
  6. I'm a bit late to the party here, but I thought I would just add that Stanley Lipshitz and John Vanderkooy wrote a great paper about why DSD is actually inferior to CD standard PCM encoding (it's because DSD is a 1 bit system and therefore has tremendous amounts of quantization noise/error):
    http://sjeng.org/ftp/SACD.pdf

    Also, the Michael Bublé album is mastered by Ted Jensen, who is often compressing albums to death, although it might sometimes be according to the artists' wishes (e.g. his mastering of Norah Jones' "Come away with me" measures from DR9 to DR14). Actually, most of Michael Bublé's albums seem to be heavily compressed, so maybe Bublé just likes it like that. However, four of Bublé's albums have been mastered by Vlado Meller, who's famous for compressing practically everything to death - and then he blames the artists and record companies for requesting it (which might be true in some cases). But his mastering of Red Hot Chili Peppers' "Stadium Arcadium" was very distorted, whereas the vinyl mastering made later by Steve Hoffmann was very fine indeed (Ian Shepherd has made a comparison video on Youtube). If you look up other Vlado Meller mastered albums on the DR database it's clear that he overcompresses practically everything.

    ReplyDelete
    Replies
    1. To rely on this Lipshitz's and Vanderkooy's old 2001 paper to objectively assess DSD compared to CD standard is very much like to rely on the Ptolemy cosmology to know about the physics of the Universe. Especially since this paper is not free of biases towards the proponents of the DVD-A at that time, as much as comparable AES Convention papers from that same time frame that were in favor of DSD were biased towards DSD. The truth is known since the very inception of DSD - properly implemented DSD or PCM-based system can be both audibly transparent.

      Delete
  7. The main issue I have seen between the SACD/CD debate and if it can be heard is complicated by so many DSD tracks on SACD being upscaled 44.1kHz PCM tracks. So no, you won't hear any 'difference'. But when you hear a really good DSD source, or even a 192 or 96 kHz PCM source transcoded to DSD, then yes you can hear the difference. But as you have shown, so many Pop and Rock albums are basically upsampled garbage. Any wonder why the SACD format didn't win? It was compromised from the start.

    ReplyDelete