|
|
Post by mikoz on Jan 14, 2020 13:20:43 GMT -5
I'm confused..... I didn't say that you couldn't do so. What I said was that I didn't see any reason to do so.
I have frequently encountered tracks that sounded very good - and happened to be in the DSD format. However, if they're files, I avoid all the aggravation of playing them in their original DSD format, by the simple expedient of converting them to 24/96k PCM. I have listened to both versions on equipment that can play both and I have never noticed any significant difference in the way they sound.
(However, as I said, the only reason I can see not to do so would be if you believe that the conversion will cause some sort of audible difference.)
That's not correct, you can buy DSD (and native DSD files) and play them without using a SACD player.
Keith, it's not a belief, it's a mathematical fact in many situations. A DSD128 file vs a 192k PCM can mathematically be better, meaning it has lower quanitization noise and better linearity and sounds more "analogy". How that translates into audible differences is where it gets interesting, and that's the fun part of it.
|
|
|
|
Post by garbulky on Jan 14, 2020 15:03:52 GMT -5
|
|
KeithL
Administrator 
Posts: 9,938
|
Post by KeithL on Jan 14, 2020 15:50:53 GMT -5
I wouldn't dispute that DSD works...
I also wouldn't dispute that, technically, DSD64 has better performance in some respects than the PCM 16/44k used in Red Book CDs.
However, at the level of comparing DSD64 to 24/96k PCM, or comparing DSD128 to 24/192k PCM, I see no clear advantage to DSD. Each is slightly better in some regards, and slightly worse in others, and the differences themselves are quite minor.
Besides which, being digital formats, you can always improve the performance of either by going the next step up in sample rate or bit depth. (If DSD128 is slightly better than PCM 24/192k in some detail I'm sure that PCM 32/384k is far better than DSD128 in that same regard... and so on... )
To be quite honest I personally fail to distinguish any qualitative difference between the two....
Note that I do NOT even consider whether either "sounds more analog".... which is meaningless at best.
I have no specific desire for "a format that sounds more like analog"...
My goal is for gear to have no sound at all - not "analog" and not "digital".... just "not there".
However, whenever anyone attempts to prove that DSDs sound audibly different than PCM, in a proper double-blind test, they consistently fail to do so.
All that said, PCM has three distinct advantages.....
1)
PCM is the standard.
Because PCM is the standard, most of the equipment used in recording studios is PCM, and most of the digital recordings currently in existence are PCM. I'm pretty sure that, of my 100 favorite albums, every single one that was digitally mastered was recorded in PCM.
And, as a consumer, if my favorite band records in PCM, purchasing a copy of their album mastered in DSD is simply not an option, so we've nothing to discuss. Now, in fact, a few of my favorite albums are available "on SACD" - but I seriously doubt any claims that those were "mastered in pure DSD".
(I rather suspect that they sound a bit different because the SACD "audiophile version" was simply mastered a bit differently so as to appeal to audiophiles.)
2)
One of the reasons that PCM is the standard is that PCM digital audio is relatively easy to work with - to mix, edit, and otherwise process.
Until recently it was nearly impossible to do any sort of editing in DSD; with recent innovations, it is now merely expensive and difficult, but is at least possible. And a lot of so-called "DSD recording gear" actually uses a sort of hybrid intermediate format rather than pure DSD. As a result, DSD is more difficult and more expensive to produce, and you have fewer options (fewer studios, fewer engineers, and so on).
3) At the home playback level it is a HUGE drawback that features like bass management and room correction cannot be used with DSD audio...
Virtually all commercial DSP engines, and the code that goes with them, are designed to work primarily with PCM audio. Because of this, that limitation is essentially "baked into the hardware", and so seems unlikely to change. (Changing that would require a huge investment in new hardware and software... and nobody seems poised to make that investment.)
To be quite candid here.... While I cannot speak for you....
I've spoken to many people who are quite convinced that "DSD sounds better".... Yet, when I talk to them about it, that opinion is virtually always based on second or third hand information and claims. I've spoken to very few people who have actually used a high quality converter to make copies of the same file in both formats and then compared them.
And, when I've actually done so, I've consistently failed to notice a significant difference.
Note that the conversion between DSD and PCM is not bit perfect, so converting in either direction will always result in tiny differences...
The tricky part is to determine whether they are significant differences... and whether either is actually better... or simply slightly different than the other...
I'm confused..... I didn't say that you couldn't do so. What I said was that I didn't see any reason to do so. I have frequently encountered tracks that sounded very good - and happened to be in the DSD format. However, if they're files, I avoid all the aggravation of playing them in their original DSD format, by the simple expedient of converting them to 24/96k PCM. I have listened to both versions on equipment that can play both and I have never noticed any significant difference in the way they sound.
(However, as I said, the only reason I can see not to do so would be if you believe that the conversion will cause some sort of audible difference.)
Keith, it's not a belief, it's a mathematical fact in many situations. A DSD128 file vs a 192k PCM can mathematically be better, meaning it has lower quanitization noise and better linearity and sounds more "analogy". How that translates into audible differences is where it gets interesting, and that's the fun part of it.
|
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Jan 14, 2020 15:53:44 GMT -5
That's my fault.... I created this new thread by moving Mikoz post from the RMC-1 thread.... At the same time someone copied several of my expositions from that thread and started another new thread.
(I'm betting one or the other will live or die "organically".....
|
|
|
|
Post by jvenard on Jan 8, 2023 22:23:46 GMT -5
dsd
Single bit processing
More processing to create a sign wave that is closer to an analog curve.
If it's good enough for MoFi to record analog tapes to and then press vinyl LPs........
|
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Jan 9, 2023 9:41:27 GMT -5
I think you may be missing most of the point.... which is that the reality is somewhat different than a lot of the myths and legends... (And, notably, nothing you said offers a reason for why DSD should be expected to sound better.) Yes... DSD works... and it uses single-bit processing. But there's nothing specifically better about single-bit processing. And, according to all the available data, it does not "produce a sine wave that is closer to an analog curve"... Both PCM and DSD deliver an accurate rendition of the original sine wave. And, because it used single-bit conversion, the noise shaping required by normal-rate DSD ends up creating a massive amount of ultrasonic noise... (So you end up having to apply an equally massive amount of filtering to remove that noise... and far more aggressive filtering than the filtering required by PCM.) And, incidentally, it was Sony who developed DSD as a "backup storage technology"... for a variety of reasons. (Basically because, back in the 1970's, when it was developed, it was easier to get better results doing it that way.) And, yes, some people are convinced that DSD sounds better... But very few of them take the time to actually compare the exact same content in both formats to see if they can really hear a difference - one way or the other. (You can't simply compare "the DSD and PCM files you got somewhere" or "the CD and SACD disc" because they're almost always NOT from the same exact master.) dsd Single bit processing More processing to create a sign wave that is closer to an analog curve. If it's good enough for MoFi to record analog tapes to and then press vinyl LPs........ |
|
|
|
Post by donh50 on Jan 9, 2023 21:46:00 GMT -5
Most if not all DSD goes through a PCM step in the studio as virtually all DAWs (digital audio workstations) use PCM. There is one system I cannot recall that will work with native DSD, but the vast majority of ADCs and DACs in the studio are PCM anyway... Reminds me of the "straight wire with gain" people wanting as few active devices in the chain, eschewing tone controls and all that jazz, who would be horrified to see all the transistors in the mixers and processors used to create the source material...
|
|
|
|
Post by fbczar on Jan 9, 2023 23:56:10 GMT -5
I think you may be missing most of the point.... which is that the reality is somewhat different than a lot of the myths and legends... (And, notably, nothing you said offers a reason for why DSD should be expected to sound better.) Yes... DSD works... and it uses single-bit processing. But there's nothing specifically better about single-bit processing. And, according to all the available data, it does not "produce a sine wave that is closer to an analog curve"... Both PCM and DSD deliver an accurate rendition of the original sine wave. And, because it used single-bit conversion, the noise shaping required by normal-rate DSD ends up creating a massive amount of ultrasonic noise... (So you end up having to apply an equally massive amount of filtering to remove that noise... and far more aggressive filtering than the filtering required by PCM.) And, incidentally, it was Sony who developed DSD as a "backup storage technology"... for a variety of reasons. (Basically because, back in the 1970's, when it was developed, it was easier to get better results doing it that way.) And, yes, some people are convinced that DSD sounds better... But very few of them take the time to actually compare the exact same content in both formats to see if they can really hear a difference - one way or the other. (You can't simply compare "the DSD and PCM files you got somewhere" or "the CD and SACD disc" because they're almost always NOT from the same exact master.) dsd Single bit processing More processing to create a sign wave that is closer to an analog curve. If it's good enough for MoFi to record analog tapes to and then press vinyl LPs........ Discussing whether DSD sounds better or different than PCM is pointless. Some like what they hear from DSD, others cannot hear a difference or simply do not care. However, some Emotiva customers paid for an XMC-2 based on the promise that DSD over USB would be forthcoming. It is stated in the User Manual. Why does DSD over USB matter? Some Emotiva customers have a considerable DSD collection and Emotiva promised they would be able to play that collection if they bought an XMC-2. Others have large collections of 24/96 and 24/192 PCM. When the XMC-2 was released it was stated, more than once, that Dirac Live would eventually be able to handle at least 24/96 rather than 24/48. As is Dirac downsamples 24/96 and 24/192 files. As I understand it the next generation of processors will still fail to live up to the promises made relative to DSD and Dirac. Arguing that DSD or high resolution PCM offer nothing over standard files does not change the fact promises were broken and products that could not perform as advertised were delivered. The technology exists, the will to provide it does not. |
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Jan 10, 2023 9:56:33 GMT -5
There may be a few others DAWs now that can handle actual DSD editing... But, at least until recently, the only one was something called Merging Technologies Pyramix. Most if not all DSD goes through a PCM step in the studio as virtually all DAWs (digital audio workstations) use PCM. There is one system I cannot recall that will work with native DSD, but the vast majority of ADCs and DACs in the studio are PCM anyway... Reminds me of the "straight wire with gain" people wanting as few active devices in the chain, eschewing tone controls and all that jazz, who would be horrified to see all the transistors in the mixers and processors used to create the source material... |
|
|
|
Post by PaulBe on Apr 24, 2023 8:33:05 GMT -5
I think you may be missing most of the point.... which is that the reality is somewhat different than a lot of the myths and legends... (And, notably, nothing you said offers a reason for why DSD should be expected to sound better.) Yes... DSD works... and it uses single-bit processing. But there's nothing specifically better about single-bit processing. And, according to all the available data, it does not "produce a sine wave that is closer to an analog curve"... Both PCM and DSD deliver an accurate rendition of the original sine wave. And, because it used single-bit conversion, the noise shaping required by normal-rate DSD ends up creating a massive amount of ultrasonic noise... (So you end up having to apply an equally massive amount of filtering to remove that noise... and far more aggressive filtering than the filtering required by PCM.) And, incidentally, it was Sony who developed DSD as a "backup storage technology"... for a variety of reasons. (Basically because, back in the 1970's, when it was developed, it was easier to get better results doing it that way.) And, yes, some people are convinced that DSD sounds better... But very few of them take the time to actually compare the exact same content in both formats to see if they can really hear a difference - one way or the other. (You can't simply compare "the DSD and PCM files you got somewhere" or "the CD and SACD disc" because they're almost always NOT from the same exact master.) Discussing whether DSD sounds better or different than PCM is pointless. Some like what they hear from DSD, others cannot hear a difference or simply do not care. However, some Emotiva customers paid for an XMC-2 based on the promise that DSD over USB would be forthcoming. It is stated in the User Manual. Why does DSD over USB matter? Some Emotiva customers have a considerable DSD collection and Emotiva promised they would be able to play that collection if they bought an XMC-2. Others have large collections of 24/96 and 24/192 PCM. When the XMC-2 was released it was stated, more than once, that Dirac Live would eventually be able to handle at least 24/96 rather than 24/48. As is Dirac downsamples 24/96 and 24/192 files. As I understand it the next generation of processors will still fail to live up to the promises made relative to DSD and Dirac. Arguing that DSD or high resolution PCM offer nothing over standard files does not change the fact promises were broken and products that could not perform as advertised were delivered. The technology exists, the will to provide it does not. "Discussing whether DSD sounds better or different than PCM is pointless..." That IS the whole point of this thread. My prediction is that DSD and SACD is going to die a painful death; at least for the consumer. BTW, do you have any ATMOS material that is recorded at greater than 24bit/48K? Do you have any DSD/SACD material (most of which started as PCM anyway) that is greater than 5.1? Can you do any bass management with DSD/SACD without converting it to PCM first? How does Dirac, at any resolution, handle DSD/SACD material? |
|
|
|
Post by fbczar on Apr 24, 2023 9:34:03 GMT -5
Discussing whether DSD sounds better or different than PCM is pointless. Some like what they hear from DSD, others cannot hear a difference or simply do not care. However, some Emotiva customers paid for an XMC-2 based on the promise that DSD over USB would be forthcoming. It is stated in the User Manual. Why does DSD over USB matter? Some Emotiva customers have a considerable DSD collection and Emotiva promised they would be able to play that collection if they bought an XMC-2. Others have large collections of 24/96 and 24/192 PCM. When the XMC-2 was released it was stated, more than once, that Dirac Live would eventually be able to handle at least 24/96 rather than 24/48. As is Dirac downsamples 24/96 and 24/192 files. As I understand it the next generation of processors will still fail to live up to the promises made relative to DSD and Dirac. Arguing that DSD or high resolution PCM offer nothing over standard files does not change the fact promises were broken and products that could not perform as advertised were delivered. The technology exists, the will to provide it does not. "Discussing whether DSD sounds better or different than PCM is pointless..." That IS the whole point of this thread. My prediction is that DSD and SACD is going to die a painful death; at least for the consumer. BTW, do you have any ATMOS material that is recorded at greater than 24bit/48K? Do you have any DSD/SACD material (most of which started as PCM anyway) that is greater than 5.1? Can you do any bass management with DSD/SACD without converting it to PCM first? How does Dirac, at any resolution, handle DSD/SACD material? I do not understand your response. Emotiva obviously thought DSD was valuable or they would not have promised it in the Gen. 3 processors. Now the Emotiva position defends the failure to provide DSD in their processors by saying DSD and PCM a sound the same. My position is a promise is a promise. As for Atmos 24/48 is a high resolution format. In Emotiva processors withDirac 24/48 is not up or down sampled, but higher resolution formats are down sampled. I think Atmos is great, perhaps you do not. Most of the SACD’s I have are in 5.1 format. Are you trying to say that SACD’s have no bass unless they are converted to PCM? |
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Apr 24, 2023 11:10:09 GMT -5
1. At this point in time the question the question of "whether DSD sounds better than PCM" has assumed the status of a religious debate. I am of the personal opinion, based on my personal experience, that the differences are almost entirely due to differences in how various versions of the same tracks and albums are mastered. However, since there is no exact bit-perfect conversion between DSD and PCM, whenever you convert in either direction, there is also the potential for a tiny but audible change to occur. My favorite quote on the subject appeared in a now-mysteriously-vanished white paper written by the chief engineer for Weiss, who sells Saracon, arguably still one of the two premiere studio DSD converters.... To paraphrase - what he said was: "DSD is in no way technologically superior to PCM. However, if your customers want it, it makes sense to offer them content in as many formats as you can". However the fact remains that some folks do insist that they can hear a difference... And some folks do ask for it... 2. Our XMC-1, and now our RMC-1 and XMC-2, do in fact support DSD - via HDMI. (Note that this is the ONLY way in which actual SACD disc players are allowed to output DSD content they play from SACD discs.) We had hoped to also support DSD via USB... but, while that's still officially "on the roadmap", it turned out to be more difficult than we anticipated with the hardware we used. (Our new XDA-3 DAC does in fact support all of the DSD-via-USB flavors.) 3. It is basically impossible to do ANY sort of digital processing on a DSD audio signal. There are digital audio workstations that can do digital processing on DSD audio... but they do so using some rather advanced and roundabout methods... like using a sort of "intermediate format". (Some folks would argue that this "intractability" is a virtue since it virtually ensures that true end-to-end DSD productions are somewhat similar to "direct-to-disc recordings".) (But the obverse of that is that apparently many so-called "pure DSD discs" really have been converted into PCM and back again at some point in the production chain anyway.) However, to answer the question, in the normal world it is impossible to do bass management, or room correction, or any other sort of digital processing, without first converting DSD to PCM. "Discussing whether DSD sounds better or different than PCM is pointless..." That IS the whole point of this thread. My prediction is that DSD and SACD is going to die a painful death; at least for the consumer. BTW, do you have any ATMOS material that is recorded at greater than 24bit/48K? Do you have any DSD/SACD material (most of which started as PCM anyway) that is greater than 5.1? Can you do any bass management with DSD/SACD without converting it to PCM first? How does Dirac, at any resolution, handle DSD/SACD material? I do not understand your response. Emotiva obviously thought DSD was valuable or they would not have promised it in the Gen. 3 processors. Now the Emotiva position defends the failure to provide DSD in their processors by saying DSD and PCM a sound the same. My position is a promise is a promise. As for Atmos 24/48 is a high resolution format. In Emotiva processors withDirac 24/48 is not up or down sampled, but higher resolution formats are down sampled. I think Atmos is great, perhaps you do not. Most of the SACD’s I have are in 5.1 format. Are you trying to say that SACD’s have no bass unless they are converted to PCM? |
|
geebo
Emo VIPs 
"Too bad that all the people who know how to run the country are driving taxicabs and cutting hair"
Posts: 24,188
|
Post by geebo on Apr 24, 2023 11:16:51 GMT -5
"Discussing whether DSD sounds better or different than PCM is pointless..." That IS the whole point of this thread. My prediction is that DSD and SACD is going to die a painful death; at least for the consumer. BTW, do you have any ATMOS material that is recorded at greater than 24bit/48K? Do you have any DSD/SACD material (most of which started as PCM anyway) that is greater than 5.1? Can you do any bass management with DSD/SACD without converting it to PCM first? How does Dirac, at any resolution, handle DSD/SACD material? I do not understand your response. Emotiva obviously thought DSD was valuable or they would not have promised it in the Gen. 3 processors. Now the Emotiva position defends the failure to provide DSD in their processors by saying DSD and PCM a sound the same. My position is a promise is a promise. As for Atmos 24/48 is a high resolution format. In Emotiva processors withDirac 24/48 is not up or down sampled, but higher resolution formats are down sampled. I think Atmos is great, perhaps you do not. Most of the SACD’s I have are in 5.1 format. Are you trying to say that SACD’s have no bass unless they are converted to PCM? The thread asks if DSD is better than PCM or not. It wasn't a discussion of whether it's included with this or that processor. That would be a topic for another thread. |
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Apr 24, 2023 11:36:16 GMT -5
The catch here is that you're only looking at PART of the math. An SACD or DSD64 file can be slightly better in some ways than a 16/44.1k PCM file. (So, in that sense, an SACD is arguably "capable of better fidelity than a CD"). In fact a DSD64 file is mathematically more equivalent to a 24/88k PCM file. But a 24/192k PCM file is better than both. (And I think a DSD128 file is actually more comparable to a 24/176k PCM file... but the details depend on your priorities.) But a 24/384k PCM file is very clearly superior to both of them. So, if you're concerned with fidelity, and don't mind big files, it's easier just to use 24/384k PCM. (You might be interested to know that the "DXD" format used to process most "DSD audio" is actually 352k PCM... Therefore, if you wish to minimize conversions, it would be sensible to leave it as PCM, rather than convert it back to DSD.) However you've got it backwards about quantization noise... DSD files have massively higher quantization noise than PCM files... (at least in the audio spectrum). However that quantization noise can then be "pushed up above the audible part of the spectrum"... The result of this is a good noise profile in the audible part of the spectrum and a good frequency response... But at the cost of a pathetically low S/N in the ultrasonic region... Which pretty much negates any claims of useful audio content being preserved in that ultrasonic range... (Even assuming that the "useful" content in the ultrasonic portion of the spectrum was audible...) And, if you can find a proper valid mathematical definition of "analogy" I would love to hear it. The idea that the digital waveform of DSD "looks more like analog" is merely a leftover from some really goofy early Sony advertising literature. (It is in fact true that the DSD waveform looks more like an analog signal... at least metaphorically... but it has no technical significance.) Likewise there have been claims that "the conversion process from DSD to analog can be done very simply". The catch is that, if you wish to filter out the really awful ultrasonic noise inherent in DSD, you end up requiring filters of comparable complexity to those required for use with PCM. HOWEVER, because there is no direct bit-for-bit correlation, conversion in either direction requires filtering... This means that there are several parameters which you can "fiddle with to fine tune the way the output sounds"... (So, in that sense, I guess the conversion process is in fact "a bit more like analog".) I'm confused..... I didn't say that you couldn't do so. What I said was that I didn't see any reason to do so. I have frequently encountered tracks that sounded very good - and happened to be in the DSD format. However, if they're files, I avoid all the aggravation of playing them in their original DSD format, by the simple expedient of converting them to 24/96k PCM. I have listened to both versions on equipment that can play both and I have never noticed any significant difference in the way they sound.
(However, as I said, the only reason I can see not to do so would be if you believe that the conversion will cause some sort of audible difference.)
Keith, it's not a belief, it's a mathematical fact in many situations. A DSD128 file vs a 192k PCM can mathematically be better, meaning it has lower quanitization noise and better linearity and sounds more "analogy". How that translates into audible differences is where it gets interesting, and that's the fun part of it.
|
|
KeithL
Administrator
Posts: 9,938
|
Post by KeithL on Apr 24, 2023 11:40:31 GMT -5
Good point... And, as per that point, it's always going to end up being a matter of opinion. From a technical standpoint, regardless of the limitations of a given sample rate for each, you can always "make it better by using a higher sample rate". (And, for each, you can always pick a sample rate that will meet any requirement you care to set.) I do not understand your response. Emotiva obviously thought DSD was valuable or they would not have promised it in the Gen. 3 processors. Now the Emotiva position defends the failure to provide DSD in their processors by saying DSD and PCM a sound the same. My position is a promise is a promise. As for Atmos 24/48 is a high resolution format. In Emotiva processors withDirac 24/48 is not up or down sampled, but higher resolution formats are down sampled. I think Atmos is great, perhaps you do not. Most of the SACD’s I have are in 5.1 format. Are you trying to say that SACD’s have no bass unless they are converted to PCM? The thread asks if DSD is better than PCM or not. It wasn't a discussion of whether it's included with this or that processor. That would be a topic for another thread. |
|
