|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
67.188.250.219
This post is a result of the latest “weeping and gnashing of the teeth” over on sa-cd.net a couple of weeks ago when it was revealed that many Chandos SACD’s previously thought to be originating from DSD masters were, in fact, derived from PCM masters. Although for some reason I only had a couple of Chandos SACD’s and thus did not have too much skin in the game (and, in any case, I don’t post on sa-cd.net), I had recently downloaded a couple of Chandos 24/96 multi-channel files from The Classical Shop: the Debussy orchestral works with Stéphane Denève and the RSNO, and the Saint-Saëns orchestral works with Neeme Järvi and the RSNO. I had previously downloaded a number of stereo (two-channel) 24/96 files from the same source, and I found these downloads, almost without exception, to be very impressive – particularly the two recent multi-channel downloads.
Just for fun, I use a software program called “Spek” to check out the spectrographs of the various tracks I’ve downloaded, and every one of the tracks I referred to above looked like a normal PCM derived file, as I expected, because when I checked the equivalent CD’s of the files I happened to download, they clearly stated “recorded at 24/96”. For example:
(I don’t know if you can make it out here, but it’s in the upper right quadrant.)
However, there was one exception: the free “test” download of part of a Weinberg symphony. In this file, you could see the dusting of ultra-high-frequency noise (absent on the other files) which can be a tell-tale sign of DSD. (BTW, I should also emphasize that this test file also sounded excellent to me, just like the PCM-derived files.) The interesting thing about this is that Chandos has contracted out the recording of some performances in their catalogue to third parties, and many of these third parties evidently DID record in DSD, as in the case of the Weinberg symphony. So far, it seems to be pretty straightforward: in-house Chandos recordings were PCM, and third-party Chandos recordings could be DSD or PCM.
So here’s where it gets confusing. Someone at Chandos wasn’t focusing on their SACD booklet information, because, in the upper right-hand corner of the back of their SACD’s, there’s a statement, “recorded in DSD” – even for their in-house productions. For example:
(Again, it’s hard to make out, but it’s in the upper right quadrant.)
This was what provoked all the outrage at sa-cd.net: that Chandos were “lying” to their customers about the origin of their SACD’s, i.e., they SAID they recorded in “DSD”, but they USED “PCM”. Especially offended were the “DSD only” listeners, who have a penchant for expressing their disdain for PCM. ;-)
But wait – there’s more confusion! Let’s return to The Classical Shop downloads for a minute. Before this information broke on sa-cd.net, I had assumed that all of the hi-rez multi-channel downloads EXCEPT for the Debussy and Saint-Saëns albums referred to above had been recorded in DSD – because that’s what it said on the backs of the SACD cases for the equivalent albums. But the verbiage on The Classical Shop site said (for each multi-channel album) “Recorded in 24 Bit / 96Khz”. OK, I think I can now accept that in view of the recent “revelations”. However, when you check out the Weinberg Symphony No. 3 on that same site, it ALSO says, “Recorded in 24 Bit / 96Khz”, despite what my spectrograph of the test file from that album seems to indicate!
So, as a customer, what are you supposed to do? If you say, “Who cares whether it’s PCM or DSD!”, I say, “Where’s the fun in that? Inquiring minds want to know!” So, in an effort to put the truth out before the public (by endangering my cash!), I risked all of £19.99 to download the Suk album with Belohlávek and the BBC Symphony - it took seven hours. (Remember, it's 24/96, 5.1, and one of the tracks had a problem downloading and I had to do it over.) I've been able to listen to only one track so far, but I've seen the spectrographs for all the tracks, and there is no dusting of ultra-high-frequency noise, as seen on the Weinberg test track. My verdict: it's likely to be a PCM recording. But remember that the Weinberg album (which is also available in multi-channel) seems to be a DSD original.
Strangely enough, I had actually posted on The Classical Shop site about a month ago, stating that if they were going to offer 24/96 downloads, then they should use 24/96 masters - it turns out that, for the most part, that's what I now think they were doing, but I just didn't know it when I posted there - and apparently, neither did they (at least until recently)!
So what are our lessons learned from all this? Gee, I dunno. I think with a lot of these hi-rez formats (DSD, 24/96 and 24/192 PCM), many of us are trying to correlate what we hear with the method or type of file used for the recording, but it seems to me that we're now almost at the stage where the differences are so minute that arguing so passionately about the sonic merits of PCM to the exclusion DSD (or vice versa) is the equivalent of arguing about how many angels can dance on the head of a pin.
Follow Ups:
It's all BS created by Sony/Philips marketing. DSD is PCM, just a particular format, 2822.4 kHz at 1 bit. Sony/Phillips did their best to ensure that all the equipment was incompatible as well, in the hopes of milking the industry and consumers for their "invention" which wasn't anything really new in the first place.
It may be that listening to a recording in the "original" format, whatever that means, will sound better. But not always. It depends on many factors. For example, I have a 192/24 Mahler 6 that was originally tracked at 96/24, and these tracks played back and mixed on an analog desk and then captured at multiple digital formats. The recording engineer justified the higher sampling rate because it sounded better. This was a case of two digital generations at 96/24 and 192/24 sounding better than two digital generations at 96/24 and 96/24. Either way the recording sounds good, which is really the only point.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
If you are, I agree that that's a very fine sounding recording.BTW, I think one of the sub-threads on sa-cd.net connected with Chandos "revelations" (the main thread was titled, "Stéphane Denève conducts Debussy") was a heated discussion as to whether PCM actually starts out as DSD, for reasons I couldn't quite follow - I assume it had something to do with PCM being processed 1 bit at a time before being "decimated" into 16- or 24-bit samples. I believe tailspn was arguing in favor of this, so perhaps he can elaborate here. In fact, any light shed on this contention would be appreciated.
Edits: 07/07/12
Yes, that's the recording.
Most "DSD" starts out in multi-bit format and this has been so for at least the past decade. There are some exceptions, e.g. the Grimm ADC. Similarly, most DSD DACs convert the input DSD to a multi-bit format, often at a higher master clock rate. That's the case with the SABRE chip in my Mytek Stereo192-DSD DAC.
There is some discussion in forums about PCM being processed one sample at a time, but that's really not correct, as all PCM formats require some form of low pass filter to smooth the output waveform and since these filters retain memory decoding these formats requires processing multiple samples "at once".
DSD is converted into PCM by low pass filtering. This will produce, for example 2822.4/24 PCM. Then, for example, 15 out of every 16 of these samples will simply be discarded, which results in 176.4/24. Of course, a more efficient implementation would not bother to calculate the samples that are about to be thrown out and some hardware implementations of the filter might not do a multiply by +1 or -1, rather just adding or subtracting the coefficient. (It would also be possible to process multiple bits of input with a single table lookup if desired, as would be familiar to any old-time program hackers.)
I am presently listening to my multi-bit recordings by upsampling them in my computer and sending DSD128 to my DAC. This seems to sound slightly better than playing them in their native format or upsampling to 176.4/24 or 192/24. I play my DSD tracks in the native 2822.4/1 format, but this will probably change if I get software that can do room EQ at 2822.4 as this will produce multi-bits and upsampling to DSD128 will preserve this resolution. (I am waiting for the needed software and hoping that it will run on my CPU.)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
But what did it sound like??
I STILL haven't listened to it all, but my first impression is that what I've heard (some of both works) is very good - a bit more distantly recorded than the Debussy and Saint-Saëns hi-rez multi-channel files I had previously downloaded from Chandos. I actually prefer the newer recordings (which are pretty spectacular) to this Suk recording - although part of this impression may have to do with the nature of Suk's orchestration vs. that of Debussy and Saint-Saëns, both of whom (despite Debussy's "impressionism") tend to choose from among the clearer colors in their respective tonal palettes. I'll try to post a follow-up once I listen to the whole thing.
Studies have been done that show when you're told the price of a bottle of wine, the more expensive it is, the more blood and oxygen is sent to a portion of the brain that activates pleasure. Thus, the taste of the wine is actually determined by pre-conceived perceptions and beliefs about it, and this is happening on a chemical level in your brain.
http://news.cnet.com/8301-13580_3-9849949-39.html
The whole DSD vs 24/96 PCM is the same thing. On paper DSD is superior. There's no debating this. People claim they can hear the difference, and wax poetics about the sound of DSD. But no one was done a true controlled experiment where a performance was simultaneously recorded in both DSD and 24/96 PCM with as much as the same equipment (such as microphones), and then tested to see if people could distinguish between the two.
Thronhill I think you are missing the point. The point here isn't which format is better, or even that one is better than the other...it's which format is the original copy in..then let me have that one. I would agree that 24/192 and DSD can possibly be indistinguishable (I own the Massimo Liszt piano SACD/DVD-A where both were recorded natively)...then that simply means that several thousand (2k+ out of 7k SACD's are native DSD-recorded, and nearly a thousand more analog-to-DSD direct) SACD's are now available to be heard in very very high resolution. Don't take that away from me by converting them!! Can you find me 2000+ 24/192 native recordings? If so, great! I'm all ears.
Edits: 07/05/12
Chris,
Thanks...this is good stuff. I am one of those DSD fans, rip hundreds of SACDs, and own a couple DSD-capable DACS (Meitner, Mytek, demoing an eXD one). But your detective work, and that of others, is quite relevant for all of us. Now that DSD recordings and DSD playback gear is becoming available, and usually coupled with great PCM playback as well.. it is important to get to the bottom of the ultimate goal....that is:
"I want to hear the recording in the closest format/generation to that of its master tape. If it's DSD, then let me hear it in native DSD. if it's PCM, then let me hear it in native PCM. If it's analog, then transfer it as cleanly as possible to one of the formats and let me know....(My pref is analog-to-DSD like the Analogue production folks, but whatever)". This means please do not convert from PCM to DSD and back again, then try to sell me it is native. If my DAC happens to do that internally, so be it...at least I know...but then its even more important that i don't have it on the front end.
Thanks again....
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: