|
|
Post by geebo on Sept 29, 2014 6:48:04 GMT -5
I've always thought that at any point in time, there is but one value to measure regardless of how many frequencies are present or how many instruments are playing. There is only one waveform that is the sum of all sounds in the music. Simplistically, because I'm a simple guy, if one (or a number of instruments) change their notes faster than the sample rate then that change in note could be missed? If that's actually the case, then logically more instruments means more chances of that occurring. For example on the S&M album there are 26 violins in the SF Symphony Orchestra. If the sound mixer/engineer knows that say DSD can only handle perhaps 6 violins does he deliberately mix out the other 20? Or if he is using say PCM, can he fit in another 6 and only mix out 14 violins? Cheers Gary I guess I don't understand your logic. I would think that it would make no difference if there were a million violins playing. If something is missed in 1/44,100 of a second, we wouldn't have heard it. Not only would it have had to change between samplings, it would also have had to end before the next sample to be missed. |
|
|
|
Post by linvincible on Sept 29, 2014 7:20:43 GMT -5
there's a lot of for and against either formats above and I didn't read analytically enough to enter that discussion.
But just my opinion though : after experimenting with same albums on 96/24 and DSD I always preferred the DSD. The soundstage sounded larger and more quiet. Maybe that just lies in the architechture of my DAC and how it perform on both formats? (Mytek 192 DSD)
But I'll keep experimenting with your nice ideas above ;o)
Charles
|
|
|
|
Post by drtrey3 on Sept 29, 2014 8:18:04 GMT -5
My experience with DSD and PCM come from some sacd's that I really enjoy for the DSD and some dvd-a and downloaded files for PCM. While I have some general thoughts, it is difficult to know just what I am listening to as I typically run the sacd's through hdmi to my UMC-200, so I am listening to DSD to PCM. Sometimes I use the analog outs from my Oppo, but then I am listening to a different dac. And I typically listen to PCM files through my XDA-2. So while I think that PCM sounds a bit more incisive and DSD sounds a bit smoother and rounder, who knows what it is that gives me these impressions? Not me.
Trey
|
|
|
|
Post by moovtune on Sept 29, 2014 13:03:50 GMT -5
So far everyone seems to be talking about Stereo files. I prefer SACD because of the multichannel option. DVD-A discs with Meridian encoded PCM were a pain because you sometimes had to turn on a monitor to select what audio option you wanted or to select tracks. SACD's were more like CD's in that regard and were multichannel. Unfortunately my older Oppo DVD Universal player doesn't do DSD as an output even though my processor could deal with it, so I hear all my SACD's as converted to 88.2 PCM and it sounds great to me.
|
|
|
|
Post by solarrdadd on Sept 29, 2014 14:05:01 GMT -5
was this thread created to do anything other than have people argue (yes, argue and not the good type either) yet again about DSD vs PCM? and by emotiva staff no less.  |
|
hemster
Global Moderator 
Particle Manufacturer
...still listening... still watching
Posts: 51,952
|
Post by hemster on Sept 29, 2014 14:40:03 GMT -5
was this thread created to do anything other than have people argue (yes, argue and not the good type either) yet again about DSD vs PCM? and by emotiva staff no less. I don't know about you, but I've learned a lot from this thread (from multiple posters). If you don't find value in reading it, you could just ignore it and move to other threads.  |
|
|
|
Post by Gary Cook on Sept 29, 2014 18:50:51 GMT -5
Simplistically, because I'm a simple guy, if one (or a number of instruments) change their notes faster than the sample rate then that change in note could be missed? If that's actually the case, then logically more instruments means more chances of that occurring. For example on the S&M album there are 26 violins in the SF Symphony Orchestra. If the sound mixer/engineer knows that say DSD can only handle perhaps 6 violins does he deliberately mix out the other 20? Or if he is using say PCM, can he fit in another 6 and only mix out 14 violins? I guess I don't understand your logic. I would think that it would make no difference if there were a million violins playing. If something is missed in 1/44,100 of a second, we wouldn't have heard it. Not only would it have had to change between samplings, it would also have had to end before the next sample to be missed. Maybe I need to be more explicit, I'm not reaching conclusions I am simply asking questions. I'm not commenting on the sound quality achieved by DSD or PCM, or reaching any conclusions. I'm simply asking why in one format there is information that is simply not there in another format? It's not good sound or bad sound, it's just there in one and not there in the other. Is the sound engineering / mixer mixing it out because he knows the format that he is using can't handle that amount of information? Or is the resulting file too big, so something has to go? Or is he mixing it out just because he wants to? I don't know the answers, what I do know is that it occurs frequently, so I suspect that there must be a reason, a good one perhaps, but I don't know what it is. Cheers Gary |
|
|
|
Post by geebo on Sept 29, 2014 19:53:36 GMT -5
I guess I don't understand your logic. I would think that it would make no difference if there were a million violins playing. If something is missed in 1/44,100 of a second, we wouldn't have heard it. Not only would it have had to change between samplings, it would also have had to end before the next sample to be missed. Maybe I need to be more explicit, I'm not reaching conclusions I am simply asking questions. I'm not commenting on the sound quality achieved by DSD or PCM, or reaching any conclusions. I'm simply asking why in one format there is information that is simply not there in another format? It's not good sound or bad sound, it's just there in one and not there in the other. Is the sound engineering / mixer mixing it out because he knows the format that he is using can't handle that amount of information? Or is the resulting file too big, so something has to go? Or is he mixing it out just because he wants to? I don't know the answers, what I do know is that it occurs frequently, so I suspect that there must be a reason, a good one perhaps, but I don't know what it is. Cheers Gary I think if there is more information it may be because of a higher sampling rate and or higher bit rate. More samples means more information. Higher resolution in those samples means more information. The real question is when does it become audible. Now I also believe CDs are mastered differently than say an SACD. The CDs very often are recorded at very high levels for "competition" and cause clipping that can be anywhere from mild to severe. Clipping causes loss of information. SACDs on the other hand, are aimed at a different market and more care is usually taken in the production process. That can lead to cleaner sound with more detail. But CDs are capable of much better sound than we usually get. |
|
|
|
Post by Priapulus on Sept 29, 2014 21:01:00 GMT -5
> Simplistically, because I'm a simple guy, if one (or a number of instruments) change their notes faster than the sample rate then that change in note could be missed? If that's actually the case, then logically more instruments means more chances of that occurring. For example on the S&M album there are 26 violins in the SF Symphony Orchestra. If the sound mixer/engineer knows that say DSD can only handle perhaps 6 violins does he deliberately mix out the other 20? Or if he is using say PCM, can he fit in another 6 and only mix out 14 violins?
Can your ears (or anybody's) hear the 26 individual violins? No, you hear one signal (actually 2 signals; two ears) that is the aggregate of all the violins, and the brain processes the signal so you recognize some/all of the violins. In the same way, whether there is one or a thousand violins, there is still only one aggregate waveform (however complex) presented to the microphone, which the adc has no problem encoding.
|
|
|
|
Post by Gary Cook on Sept 29, 2014 22:34:49 GMT -5
Can your ears (or anybody's) hear the 26 individual violins? No, you hear one signal (actually 2 signals; two ears) that is the aggregate of all the violins, and the brain processes the signal so you recognize some/all of the violins. In the same way, whether there is one or a thousand violins, there is still only one aggregate waveform (however complex) presented to the microphone, which the adc has no problem encoding. I agree I'm sure that I can't hear every one of the 26 violins individually, but I sure as hell notice when some are missing. I can't identify how many, but I can hear that a proportion are simply not there. I've previously put that down to the sound engineer / mixer's decision, "I want this, I don't want that". Why he/she made that decision is what interests me now. Is it a technical limitation of the equipment he/she is using, a PCM or DSD coding limitation or is it simply personal preference (ie; "there's too many violins"). Cheers Gary |
|
|
|
Post by garbulky on Sept 29, 2014 22:57:24 GMT -5
gary: I think I recall remembering what kind of finite time was captured vs the different types of sample rates. For instance between 16 bit and 24 bit the actual resolution of the data "ladder" was increased. Of course the ladder is not what is reproduced. But You take the data to recreate the waveform of the signal.
|
|
KeithL
Administrator
Posts: 10,274
|
Post by KeithL on Oct 12, 2014 14:24:55 GMT -5
Exactly..... People tend to forget that you only have two ears, and each ear can only hear one "signal". You can "hear" 26 violins because your brain can pick out the hints that tell it that there are 26 violins - just like you can recognize 26 different voices at a party. (Although, even though you might recognize that many different people if they talk individually, at some point, when enough of them are talking, it often becomes "a crowd over there".) Different violins may be pitched slightly differently, or they make produce sound that has different percentages of different harmonics, or their player may not always play the same note at precisely the same time or intensity. Also, since those violins are not in the exact same place, the time delay between the sound from each of them and your two ears will be different; both obvious stuff like the violin on the left being closer to your left ear, and less obvious stuff, like the fact that the reflections of the left violin from the left wall arrive sooner at your left ear than the reflections of the right violin from the left wall - because the left violin is closer to the wall. It can be interesting to try and figure out exactly how much of that detail is "real" and how much you're imagining. (Do you really hear "26 separate violins" - so you could tell me how far apart violins #8 and #9 are - or do you really hear "a violin section, ten feet wide, with a bunch of violins spread across it"?) Amazingly, all this detailed information is contained is a single pair of signals (for stereo), and each signal only has TWO characteristics... voltage and time. In other words, that line you see on an oscilloscope, showing the relationship of voltage vs time, REALLY IS ALL THERE IS. If you reproduce that squiggly line PERFECTLY, then it WILL be identical to the original. And, just like you can recognize someone's voice, even on a poor phone connection, your brain manages to extract all those details and figure out pretty much what's going on. All the other stuff you think you hear is, quite literally, in your head. It's sort of like talking about a great painting, and going on and on about "feeling" and "light" and "depth" and "color relationships".... when, in reality, if you describe precisely what color each speck of the surface of the canvas is (and maybe the surface texture of the paint in some cases), you've described everything that's there in the information. A "glowing halo" is really nothing more than some yellow paint next to some dark paint.... and is perfectly defined by that information, and "the way Rembrandt handles light" is entirely "encoded" in the colors on that surface. Those terms are actually a form of shorthand we've developed to talk about the result of huge numbers of colored points arranged in specific ways. And so, if you have perfectly duplicated the color, and the surface texture, then your duplicate IS perfect. (And, if it does look different, there isn't some "magical missing thing", it's simply that you didn't get those two things perfect after all, or you miscalculated on how much error could be allowed before it would be noticed.) If I were to take a specific sound, make both a PCM and a DSD recording of it, and determine by subtracting one from the other that they really were identical (the numbers will be different, but we can compare the analog result when we convert them back), then BY DEFINITION they WOULD sound exactly the same. If in fact you find that they don't sound the same, then there are ONLY two possible explanations: 1) You didn't measure carefully enough, and so you missed a slight difference (or you guessed wrong about how tiny a difference would be "inaudible"). 2) You're imagining it. There IS no third option, there is no third dimension, there is nothing else. What many people don't understand is that terms like "phase" are simply ways of talking about that same signal - in a special way that allows you to talk about certain aspects of it while ignoring others. If I reproduce a given signal, such that, at every instant in time, the voltage of the reproduction is exactly the same as that of the original, then they ARE THE SAME. The phase will be the same, the THD is the same, the harmonic relationship is the same, the color is the same, the sound stage is the same, the "emotion" is the same, the transient response is the same... and you can continue that list as long as you like. I don't need to talk about individual distortions because, if the two copies are really the same, then the distortion is zero. In fact, even if someone "invents" or "discovers" some new distortion next year, IT will be the same too. There are an infinite number of ways two things can be different, and an awful lot of ways you can describe those differences, but there is only ONE " SAME". Once you understand how this stuff really works, it all starts to make sense. The world's cheapest cassette recorder can record a billion violins. However, if it turns out that the way your brain can tell that it's hearing two violins six inches apart is by noticing a 0.5 millisecond difference in the times when it hears them, and the cassette "blurs time" so that you can't clearly hear the delays accurately enough to "pick out" one that small, then you'll probably hear what sounds sort of like one violin, as loud as two, and more or less six inches wide, rather than two distinct violins six inches apart. Now, of course, no reproduction is absolutely perfect... and DSD and PCM are different enough formats that they may have slightly different types of errors. This opens up the possibility that the errors produced by one are less annoying than the errors produced by the other (which would mean that it "sounded better"). However, the reality is that the errors produced by DSD, and by 24/96 PCM, are quite similar, and both are tiny compared to the errors that are present in other places in the reproduction chain. In other words, while it's unlikely that you would hear a significant difference with most recordings, it's even less likely that either one would be specifically better. This leads to another point, however. The two formats differ in such a way that you CANNOT make a "bit perfect" conversion between them. Whenever you convert in either direction, the conversion process results in tiny differences, which may indeed be audible. This means that, if you start with a DSD master, and convert it to PCM, the PCM copy will sound different; and, if you start with a PCM master, and convert it to DSD, that DSD copy will sound different. So they may quite possibly sound a tiny bit different... What you need to be careful of is to avoid falling into the psychological "different must be better" trap. For purposes of this discussion, however, the point is that a 24/94 PCM recording, and a DSD recording, have approximately the same ability to resolve details, so they are about equally able to "store a recording of 26 violins in such a way that the details your brain needs to pick out the individual violins remain intact". The differences between individual recorders, and between individual DACs and converters, are far greater than the differences in the "inherent capabilities" of the two formats. This strongly suggests that neither format is really superior as a storage format - which gives the win to PCM on the grounds of "convenience". Think of the difference between metric and English measure; feet aren't inherently more or less accurate than meters, but you do have to convert between them, and, if you go back and forth, there will be rounding errors and such.... but does that mean either is "better"? No. Either can produce equally accurate results. (Some people would suggest that metric is better because the math is easier; if you follow that logic, then PCM is better because it's easier to edit  ) > Simplistically, because I'm a simple guy, if one (or a number of instruments) change their notes faster than the sample rate then that change in note could be missed? If that's actually the case, then logically more instruments means more chances of that occurring. For example on the S&M album there are 26 violins in the SF Symphony Orchestra. If the sound mixer/engineer knows that say DSD can only handle perhaps 6 violins does he deliberately mix out the other 20? Or if he is using say PCM, can he fit in another 6 and only mix out 14 violins? Can your ears (or anybody's) hear the 26 individual violins? No, you hear one signal (actually 2 signals; two ears) that is the aggregate of all the violins, and the brain processes the signal so you recognize some/all of the violins. In the same way, whether there is one or a thousand violins, there is still only one aggregate waveform (however complex) presented to the microphone, which the adc has no problem encoding. |
|
|
|
Post by bigboydaddy on Oct 12, 2014 18:31:26 GMT -5
Thanks to all I enjoyed everyone’s post! A great read. 
|
|
|
|
Post by jdsewell on Nov 11, 2014 19:06:35 GMT -5
The history from what I remember when they first started pushing SACD which was back in 1999, the argument was it would be more representative of a line signal resembling a LP. This is because the high bit density compared to CDs at the time would create a very smooth signal curve. I like to think of like when you start to learn about integration in calculus, taking the area under a f(x) with an infinite amount of squares makes a more accurate representation of the area under a f(x). The same can be thought of here, the bit stream from the DSD creates a more smooth curve "f(x)" which would be representative of signal while listening to an LP but with the advantage of more accurate source information. Though the argument could be said the same for now 24bit/192kHz source because of the greater bit depth allows for a better line signal "f(x)." Though I will say that the SACDs I do own have sonics which are outstanding but the sonics are only as good as the original recording and most SACDs are recorded off the original master tapes.
|
|
|
|
Post by moko on Jan 23, 2015 0:04:26 GMT -5
i think we should not make over generalization that "sounds good/pleasantly sounding" = not accurate/more coloration.
there is a real possibility that "sounds good" = more natural.
in the digital era people are get used to sound of cd players and dacs. most of them think that this is more accurate because no anomalies of analog medium and good measurements. however, that good measurements are only in frequency domain. in the time domain, they just suck. if you look at dac measurements in stereophile, you can see that the FIRST thing they measure is impulse response of 44.1 pcm. impulse response IS in time domain. this is the impulse response of high end dac (PS Audio PerfectWave) :  as you can see there are pre-ringing and post-ringing. and that my friends, is what we perceived as UNNATURALNESS. depent on the person, but some people are highly sensitive of ringing that causing listening fatique, especially in 3-4 khz area where the ears are most sensitive (http://en.wikipedia.org/wiki/Equal-loudness_contour) this ringing is the side effect of oversampling and digital filter. in the NOS (Non OverSampling) cd players and dacs you won't find that kind of measurements. so why bother with oversampling and digital filter ? this is the noise floor measurements 44.1, 96, 192 in a NOS dac (DDDAC 1794) :  now, no need to tell which one is 44.1 and which one is 192  . we can see now why those oversampling and digital filter applied : to make it less grainy ! but with analog filtering that problem can be solved. so how this related to DSD ? 1 bit 2.8224 MHz format (DSD64) leaves no room of digital filtering because only 1 bit is available.  now you should understand why NOS dacs gaining popularity and some people still LOVE the analog sound |
|
KeithL
Administrator
Posts: 10,274
|
Post by KeithL on Jan 23, 2015 12:52:41 GMT -5
Interesting "thesis" - but you're missing a few things.... First, part of the pre-ringing and post-ringing you see on a DAC's output is simply because the signal is band limited. You CANNOT "record" a perfect square wave, or a perfect impulse, at ANY sample rate. The closest you can get is a signal that has been band-limited to the Nyquist frequency, which is half of the sample rate.... which means that the signal either MUST be pre-filtered, or it must not contain any content above that frequency to begin with. Any signal energy above that frequency is transformed into noise and distortion as a "side effect" of the conversion process. In order to be able to record a signal without band-limiting it, you would have to use an infinite sample rate, or make very sure there's no content or even background noise above that frequency. Even if it were possible for some magical DAC to reproduce a digital audio signal with no ringing, it wouldn't matter, BECAUSE SUCH A SIGNAL CANNOT EXIST TO BEGIN WITH. (If you look at the time scale on your illustration, you'll see that the ringing is at frequencies you can't hear anyway, although it does affect the overall shape of the impulse.) Second, because of the way digital sampling works, ALL DACS must include an output reconstruction filter. If you THINK you have an NOS DAC that doesn't have a filter, it's simply because it uses an "implied filter" (the unpredictable and inconsistent frequency response limitations of the rest of your system, including the speakers, are serving as your filter). The process of sampling audio into digital, and then converting it back to analog, produces "extra junk" related to the sampling and conversion processes; either you filter out that extra energy, or it becomes noise and/or distortion; there is no third option. The reason for oversampling is that it is virtually impossible to design a filter that works properly at a 44k sample rate, blocking everything above 22 kHz, yet NOT causing horrible frequency response and phase issues below 20 kHz. Oversampling allows a simpler filter to be used, at a higher frequency, which makes it easier - and, in fact, possible - to design around most of these limitations. (This is why most NOS DACs have horrible frequency response and distortion figures - because they have bad filters.) Note that EXCESSIVE ringing can indeed be audible... but that's simply a matter of good design. Third, the idea that DSD avoids filtering is simply a silly myth. If you actually look at the noise spectrum on a DSD signal, you will see that it includes a huge amount of noise in the ultrasonic region (we're talking a S/N of less than 20 dB, with almost all that being really nasty ultrasonic noise). For this reason, DSD requires even more aggressive filtering than a PCM signal.... If you fail to do this, then you're going to be passing massive amounts of ultrasonic noise to your equipment, which will not only risk incinerating your tweeters, but will also probably result in all sorts of distortion as your equipment fails to handle it.... At best, your other equipment WON'T pass it, which just means that it's acting as that filter than you imagined you avoided. DSD DACs do indeed require digital filtering... lots of it... (I have no idea whatsoever what you meant by "1 bit not allowing for digital filtering..."). A regular "single-rate" SACD or DSD recording is approximately equivalent to a 24/96 PCM recording, so it's no great shock that it sounds (and measures) a bit better than a CD, which is recorded at 16/44. In fact, once you get past the fact that many people don't seem able to hear the difference, it's a pretty good argument for why "CDs aren't quite good enough, so we should be using 24/96". And "double-rate DSD" is even better... but then so is 24/192 PCM. To avoid argument, you might want to check out this White Paper... (Weiss is a maker of very high end studio SACD and PCM mastering equipment, including Saracon, which is considered by many people to be the best DSD/PCM conversion software in existence. You might find their opinions and conclusions... interesting.) www.weiss.ch/assets/content/41/white-paper-on-DSD.pdfNOTE: I don't even want to suggest that SACDs don't sound good; they sound just fine. However, it's most likely that most of the differences you hear are because a given disc was simply well mastered, possibly in part because DSD doesn't allow a lot of the post processing that is commonly used with PCM content, and not because of some inherent superiority of the format itself. i think we should not make over generalization that "sounds good/pleasantly sounding" = not accurate/more coloration.
there is a real possibility that "sounds good" = more natural.
in the digital era people are get used to sound of cd players and dacs. most of them think that this is more accurate because no anomalies of analog medium and good measurements. however, that good measurements are only in frequency domain. in the time domain, they just suck. if you look at dac measurements in stereophile, you can see that the FIRST thing they measure is impulse response of 44.1 pcm. impulse response IS in time domain. this is the impulse response of high end dac (PS Audio PerfectWave) : as you can see there are pre-ringing and post-ringing. and that my friends, is what we perceived as UNNATURALNESS. depent on the person, but some people are highly sensitive of ringing that causing listening fatique, especially in 3-4 khz area where the ears are most sensitive (http://en.wikipedia.org/wiki/Equal-loudness_contour) this ringing is the side effect of oversampling and digital filter. in the NOS (Non OverSampling) cd players and dacs you won't find that kind of measurements. so why bother with oversampling and digital filter ? this is the noise floor measurements 44.1, 96, 192 in a NOS dac (DDDAC 1794) : now, no need to tell which one is 44.1 and which one is 192 . we can see now why those oversampling and digital filter applied : to make it less grainy ! but with analog filtering that problem can be solved. so how this related to DSD ? 1 bit 2.8224 MHz format (DSD64) leaves no room of digital filtering because only 1 bit is available. now you should understand why NOS dacs gaining popularity and some people still LOVE the analog sound |
|
|
|
Post by vcautokid on Jan 23, 2015 13:20:36 GMT -5
Want to make your own. Get one of these.http://tascam.com/product/da-3000/
|
|
|
|
Post by flamingeye on Jan 23, 2015 13:30:35 GMT -5
I have some very good DSD / SACD examples and some not so good. And I also have some redbook CDs that are as good as anything else I've heard. So how much of the good sound on our SACDs is due the the technology and how much is due to the care, mastering and production? CDs are often compressed and recorded at such high levels causing peaks to clip badly just to to make it louder than other CDs. SACDs are generally not compressed so heavily and are recorded at a lower levels for more headroom for peaks. So is it the vehicle or the driver? and there is the problem CD's are heavily compressed and sa-cd are not . grant it some CD's are quit good but it's getting fare and few wile most sa-cd's are not, but as you say it's in the hands of the mastering that makes the most of what will sound good, great or fantastic . I buy the sa-cd's because they usually are better sounding then the CD. until the loudness war stops sa-sd will find it's market among the audiophiles |
|
|
|
Post by moko on Jan 24, 2015 1:21:36 GMT -5
hello keith, first, i'm not sure about "the ringing is caused by the signal is band limited". the audio note CD-4.1x cd player which doesn't employ over sampling and digital filtering looks good on time domain :  even a cheap, $60 nos dac MUSE Mini TDA1543x4 does look good on time domain : archimago.blogspot.com/2013/02/measurements-muse-mini-tda1543x4-nos.htmlhere is another effect of different digital filters to impulse response : www.stereophile.com/content/dcs-vivaldi-digital-playback-system-measurements(note that beside ringing, digital filter also affecting phase linearity like analog filter) as for square waves which is in the frequency domain, i think most will agree that nos dac doesn't measure good on frequency domain, but they DO look good on square waves : www.dddac.com/dddac1794_test_specs.htmli'm not sure either about "the ringing is at frequencies you won't hear". the impulse response from ps audio dac is about 0.1 ms and the ringing is about 0.4 ms each way (before and after). 0.1 ms = 0.0001 s. with f = 1/T = 1/0.0001 = 10000 hz = 10 khz. second, agree that all dacs use some filtering, wether it's digital or analog filtering. with EVERY kind of filter, we want the good thing to pass and the bad thing to go away. but unfortunately most of the time, the bad thing also pass. it means there are also some side effects. BOTH digital and analog filtering have its weakness. what i found does not make sense is applying that filter at high frequencies WOULD NOT affecting lower frequencies in the audible range. why ? because those filters are actually wide bandwith low pass filter. more steep filter will have more side effects. and like you said, even if the dac doesn't have filtering most pre-amps, power amps and speakers has their own filters. even our ears too ! www.metrum-acoustics.com/Design%20Philosophy%20Metrum%20Acoustics.pdfthird, i'm not saying that DSD avoids filtering. like you said before, in a pure DSD process, mixing and equalizing in DSD format is impossible. so in a true DSD playback, the filtering is done in the analog domain (not using digital filter). but unfortunately (just like in that weiss' white paper), some DSD players are not pure DSD like the ones who use Sabre ES9018 chip. they converted DSD to PCM first then applying digital filter which make it not a pure DSD and lossy conversion. |
|
|
|
Post by yves on Jan 24, 2015 4:51:12 GMT -5
I have some very good DSD / SACD examples and some not so good. And I also have some redbook CDs that are as good as anything else I've heard. So how much of the good sound on our SACDs is due the the technology and how much is due to the care, mastering and production? CDs are often compressed and recorded at such high levels causing peaks to clip badly just to to make it louder than other CDs. SACDs are generally not compressed so heavily and are recorded at a lower levels for more headroom for peaks. So is it the vehicle or the driver? and there is the problem CD's are heavily compressed and sa-cd are not . grant it some CD's are quit good but it's getting fare and few wile most sa-cd's are not, but as you say it's in the hands of the mastering that makes the most of what will sound good, great or fantastic . I buy the sa-cd's because they usually are better sounding then the CD. until the loudness war stops sa-sd will find it's market among the audiophiles Yes, this is what I do too. Only problem.. SACDs and DSD digital downloads (where available of course) only sound better than the CD version for *some* music albums, i.e. not really very many at all (except perhaps in some specific music genres like jazz and classical, which not everyone tends to listen to a lot of course), and, in the vast majority of cases where the CD version *does* lose out to them, when people prefer them not just over the CD version, but over *any other* version, on general, this is IMNSHO due mainly to the fact someone forgot (of course) to compare it to the vinyl version.. |
|
)
. we can see now why those oversampling and digital filter applied : to make it less grainy ! but with analog filtering that problem can be solved.
