|
|
Post by fbczar on Jan 19, 2023 15:14:20 GMT -5
In general I mostly agree with what you said... but with a few limitations... First of all, when it comes to "comparing identical recordings in DSD and PCM"... My question is this: "How were you able to compare 'identical recordings' ? " When you look at "the Red Book CD layer" and "the DSD layer" on SACD discs they are sometimes mastered slightly differently. And also, even if you literally start with one, then convert it to the other yourself, the results will NOT be "identical". First of all there actually is no mathematically exact correspondence between them... And, probably more importantly, ALL converters include a variety of parameters and settings that alter the content further... For examples, I've heard demonstrations where the same DSD original was converted to PCM, using the top two software conversion packages (Weiss Saracon and Korg Audiogate)... And, at the default or similar settings, the results were subtly (but audibly) different. At most I would agree that you will stay closer to the original by doing the fewest conversions. (But, unless you have something that was recorded in DSD, and edited in DSD, or not edited at all, then it's probably been converted a few times anyway.) Second of all is the question of "upsampling in software vs oversampling in the DAC"... I would also suggest that you also cannot make a direct comparison there... While it's a nice idea that "software has more time and processing power to do a better upsampling job" the reality isn't quite so straightforward. For starters... different software programs often produce very different results when upsampling... and some do much better jobs than others... And, beyond that, some programs deliberately make "technically poor choices" because someone feels that the results "sound better". (If you want to see more information than you ever wanted to know about that take a look here: src.infinitewave.ca/ ) In addition to that MOST modern DACs do SOME oversampling most of the time anyway. So, if you upsample in software, they just end up oversampling it again anyway (unless you use an actual "non-oversampling DAC"). (And, either way, the DAC is going to use a gentle filter at a relatively higher frequency - which is why they oversample in the first place.) Third... to be quite honest I would say that the jury is out on whether (and when) downsampling is actually audible. (I would absolutely agree that it's better off to avoid downsampling whenever possible "because it might sometimes be audible".) However many people don't actually agree that downsampling to 44k is audible... and many more consider downsampling to 48k to be "absolutely safe". And, again, a lot is certain to depend on the original source material, and how the resampling is done. HOWEVER, I would also point out that Dirac Live, by its very nature, is doing some pretty heavy processing on your audio signal. So, to put it bluntly, if you're worried about "bit-accuracy" after running Dirac Live... that boat has long since sailed. All that really matters is that the output sounds significantly better than what went in... (And, since the single biggest problems are usually speakers and rooms, and Dirac Live does a good job at improving those, that is usually the case.) And finally - on the subject of various types of noise and interference... There are two distinct parts to that subject. The first is of whether noise affects the DIGITAL signal... And the short answer there is that, in properly designed circuitry, it should not. There have been many products claimed to somehow "improve the digital signal"... including goofy ideas like beveling your CDs and painting the edges green... However, not surprisingly, whenever someone attempts to test them, they generally find that they are unable to correct "bit-errors" because there are no bit errors to correct. My favorite ripping program actually tests each rip against a checksum database... out of around 5000 songs ripped it found THREE with AT LEAST A SINGLE BIT ERROR. (And you just plain can't improve something that's perfect to begin with.) Now it is possible for noise or interference to cause performance problems inside a DAC. But the claims that it actually does so are in most cases exaggerated. The second part of that subject is that noise can get into the ANALOG circuitry in a DAC. This is absolutely a real thing and absolutely can cause real problems. And sometimes filters, installed at various points, absolutely can reduce or eliminate it.
Likewise, a good clean power supply, for the entire DAC, or for certain parts of the DAC circuitry (like the reference voltage input), can absolutely help.
HOWEVER I would also advise some serious caution here when it comes to claims and claimed benefits. For example, if you use a Toslink cable, or a Coax cable with actual galvanic isolation, then it will reduce noise reaching the DAC. (Which may or may not make any difference depending on many other things.) But, while an optical Ethernet cable provides perfect isolation, there is still a digital Ethernet receiver on the other end of it... And, if you use a nice clean power supply for your streamer, the streamer itself is still a digitally noisy little computer... Computers run on high-frequency square waves... so by definition they're incredibly noisy... and the power supply can't fix that. (Luckily, if the other circuitry is well designed, it's not a problem.) I enjoyed your long post and was glad you do not fall in with the bits are bits way of thinking. Most DACs are not impervious to Conducted Interference, Electrostatic Interference and most of all Magnetic Interference from its inputs. A system that does not allow EMI and RFI to get to the DAC is the exception rather than the rule. Ethernet cables often act as antennas relative to RFI and EMI. Most USB cables do not isolate data from power. Power Supplies are subject to Conducted Interference, Electrostatic Interference and Magnetic Interference. Obviously, systems vary as to the EMI and RFI that is present, but to the extent such "noise" can be reduced sound quality, sound staging and imaging can be improved. EMI and RFI interference impact the time domain, as you pointed out. As such, the soundstage and life-like qualities of the sound of a system are diminished when EMI and RFI are present. USB cables that isolate power from data are not terribly expensive. Setting up an Ethernet over Fiber system that galvanically isolates a DAC from ethernet transmitted noise is also simple and inexpensive. Although more expensive, quiet streamers with top notch power supplies can make a considerable difference in a good system vs a computer or even a less quiet streamer. High resolution files or files upsampled before being sent to a DAC allow the DAC to do fewer conversions and sound better. Music recorded at higher sampling rates requires less steep filters to deal with very high frequencies than music recorded at a lower sampling rates. We agree that downsampling has the potential to degrade the sound quality of a music file. As such, it is unfortunate that Emotiva could not provide the computing power necessary for Dirac Live to sample at 24/96 in the XMC-2 and RMC-1 rather than downsample to 24/28. Dirac Live is great, but I wish it did not down sample my high resolution files. And finally, in my experience, genuine DSD over USB sounds more like a live performance than PCM when identical recordings are compared. When I bought my XMC-2 the ability to play DSD over USB was a significant reason for my purchase. That Emotiva did not follow through with their promise to activate DSD over USB remains a significant disappointment, but it is rare when an SACD does not sound better than an equivalent PCM recording on the XMC-2 which can play DSD over HDMI. Benchmark has a some articles on system noise and DACs that are interesting: ttps://benchmarkmedia.com/blogs/application_notes/inside-the-dac2-part-2-digital-processing benchmarkmedia.com/blogs/application_notes/149341191-inside-the-dac2-part-1-analog-processingbenchmarkmedia.com/blogs/application_notes/inside-the-dac2-part-3-power-suppliesAnd this Positive Feedback article is a good one: positive-feedback.com/audio-discourse/why-usb-cables-can-make-a-difference/If you send a 24/96 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/96. If you send a 24/192 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/192. A DAC that receives a high resolution file does not downsample that file to 16/44.1 and start over with upsampling. There are several recording studios that record a performance in PCM and DSD at the same time. In those cases it is possible to compare DSD to PCM. In my experience, a true DSD recording has a uniquely silky signature that PCM does not seem to present. The DSD version of the music I own sounds better to me than the PCM version. Maybe all DSD recordings are just better mastered, whatever the reason DSD has a silky, life-like sound. SACD's are particularly interesting. Whether it is an apples to apples comparison or not the SACD's I own sound better than the PCM versions. This is particularly true of the Brothers In Arms album by Dire Straits and Jeff Wayne's Musical Version of The War Of The World. If you get a chance give them a listen. Audible or not, Dirac Live software is fully capable of handling a 24/96 or 24/192 music file without downsampling to 24/48. In the case of Emotiva processors and Dirac Live the limitation lies with the processor. There are other processors that do not have the 24/48 Dirac Live limitation. As to your last point about noise, a computer is about as noisy an environment as is imaginable. The noise level in a purpose built device like an UltraRendu streamer is far less noisy than a Windows or Mac computer. And, while Ethernet over Fiber is not a perfect solution it does reduce noise vs a standard Ethernet cable. The BlueSound node is not expensive and its DAC is a limitation compared to the DAC in an XMC-2. However, if you use the BlueSound Node with the XMC-2's DAC vs a Mac or Windows computer directly into an XMC-2 the BlueSound / XMC-2 combo will sound better. Unfortunately the Bluesound's power supply is less than stellar. Anyway, the bottom line is reducing RFI and EMI in a system allows a system to sound its best, especially relative to soundstage. I think it is great that you are willing to discuss stuff like this. |
|
|
|
Post by leonski on Jan 19, 2023 23:03:22 GMT -5
Concerning RFI / EMI?
Seems like 2 possible entries for such nasties into a system.
Over power lines. Most of thie can be gotten rid of with a good Power Conditioner. Or in some cases an over-capacity
Isolation Transformer......I have both. Isolated circuits may also help. A lot of that 'bad' stuff simply won't get thru
a Toroid....
OR
Thru the air. Transmitted, if you will. good grounded casework may help. Ever lived near a commercial radio transmitter?
Or an airport Radar system? It's like getting a local AM station thru one of your FILLINGS. Tough to figure out the fix.
I'd suggest getting a good AM radio and doing a 'crawl'. If you can hear bad stuff on the radio or it is 'blanked' due to RFI?
Than you have an issue......
|
|
KeithL
Administrator 
Posts: 10,261
|
Post by KeithL on Jan 20, 2023 12:00:54 GMT -5
There's rather more going on inside a modern DAC than "basic DAC theory" would seem to suggest. For example... An R2R DAC really does work pretty much like the diagram you saw in an old textbook... with a ladder of resistors or reference voltages, a row of switches, and some filters. However a Delta-Sigma DAC looks almost nothing like the "one-bit DAC" that most people imagine. Most Delta-Sigma DACs are actually a hybrid and internally use a four or five bit "ladder section" rather than a single bit converter. Technically "oversampling" is just a specific type or category of "upsampling"... However, in practice, they are usually done very differently, and in different places in the process... That difference between "upsampling" and "oversampling" is somewhat arbitrary in theory but rather more distinct in practice and implementation... The term "upsampling" is generally used to refer any time you re-sample a digital audio signal to pretty much any higher sample rate - so, for example, you might upsample a 24/44k signal to 24/96k. But, in practice, the term "oversampling" is generally used to refer specifically to upsampling to an even multiple of 2x the original sample rate - so 2x, or 4x, or 64x. Also, in practice, "oversampling" is almost always done by the digital circuitry inside the DAC chip, or its associated external filter, after the original multi-bit data has been "reduced" to the form in which it will be processed internally. So you might "upsample" a 44k file to 96k using your favorite sample rate conversion program... (and there are a few DAC products that do this using a processor before sending the data to the DAC chip). And a DAC chip might be configured to oversample to fs x64 based on a specific master clock... But, in most cases, the two are really independent, and the DAC chip itself is going to use oversampling internally anyway, after it receives the signal, to raise it to a much higher sample rate for processing. (And, yes, this makes upsampling it before that point somewhat redundant... unless you are using a specific upsampling filter or process which produces a result that you find pleasing.) The argument, which might be true in some specific cases, is that their particular DAC has been "optimized" to work best when fed audio at a specific sample rate. (The original Benchmark DACs - the DAC1 and DAC1/USB - converted all incoming audio data to a sample rate of 110k, which is a non-standard sample rate, for this reason...) However, suffice it to say that, in practice, most modern DAC chips run at very high clock rates and still apply oversampling to incoming data - even at 96k or 192k. (So, while you might achieve some benefit from using higher-sample-rate audio, you probably will not avoid internal oversampling.) As we seem to agree it really is impossible to perform an "apples-to-apples comparison" between PCM and DSD... For example, even if someone "records the same performance on both DSD and PCM at the same time" they are doing so using different recorders, with different analog-to-digital conversion hardware, and different analog audio paths. And, even if they use two of the same model of recorder, there may be differences between the DSD and PCM signal paths that are not specifically associated with the format itself... And these may include simple differences in the conversion hardware used in each signal path or differences in how the hardware and filters in each are configured. And, even beyond unintentional differences, we may even wonder if the DSD signal path has been "voiced differently to produce a result that is more pleasing to audiophiles". While we know that this often occurs when SACDs and CDs are mastered we cannot rule out that it also happens with the hardware itself. (In simplest terms we can never know "if everyone did their best to produce results that are as accurate as possible" with either format.) Incidentally I absolutely agree that the SACD versions of many albums do sound better than the CD versions. I've heard the Dire Straits album and the SACD version does sound very nice... Likewise for a few of the SACD versions of Dark Side of the Moon... and The Eagles Hotel California... and several of Mike Oldfield's albums. And my favorite version of my all-time favorite album (Renaissance - Scheherazade and Other Stories) is the Audio Fidelity SACD version. HOWEVER, while I cannot say that I've done the comparison with all of them in detail... I have actually taken ISO copies of the DSD content on a few of those... and converted them into 24/96k PCM... And, when I have done so, I have been unable to notice any difference that I would characterize as either "sounding better". (As I've said before there are in fact really tiny differences... which could be attributed to the conversion process or the DACs involved... but I would not say that either version sounded "better".) I would also be remiss if I didn't mention that I'm pretty sure that neither of those albums you mentioned was RECORDED in DSD. Therefore both were in fact converted to DSD from either analog tape of PCM. (I'm too lazy to look but I'm guessing that at least Dire Straits started out on analog tape... which is FAR less accurate than either DSD or PCM.) I'm going to be fair here and start by saying that, although I've read their marketing literature, I have neither listened to, or taken apart and examined, any of the Rendu products. (And, yes, we all know that neither Windows nor Apple computers are designed with much of any consideration towards minimizing noise at all.) HOWEVER, that said, we must agree that any such streaming device is basically a small computer; a digital device which runs on high frequency square waves. Now, on the one hand, I have no doubt that they have put some work into minimizing the "stray and unnecessary noise"... perhaps with significant success. But it's still a computer, taking in Ethernet packets, processing them digitally, and putting out a digital audio signal... (so it's still about as far from analog as you can get). So, considering the fact that a properly designed DAC is designed to be largely immune to this sort of noise (in the context of exposure to it from the audio source)... I'm not honestly convinced about exactly how much optimizing parts of that process matters in actual practice (performance)... For example, if I'm paying extra for better performance, I would like to see actual numbers showing better performance, beyond "it seems like a good idea". (My guess would be that some DACs are far more sensitive to that than others... so your results would also tend to vary rather widely depending on what DAC and other equipment you have.) I'm also sort of forced to draw a humorous parallel to an episode of Get Smart... Where the bad guys invented "a silent explosive" that could "blow things up while making no noise whatsoever"... If you send a 24/96 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/96. If you send a 24/192 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/192. A DAC that receives a high resolution file does not downsample that file to 16/44.1 and start over with upsampling. There are several recording studios that record a performance in PCM and DSD at the same time. In those cases it is possible to compare DSD to PCM. In my experience, a true DSD recording has a uniquely silky signature that PCM does not seem to present. The DSD version of the music I own sounds better to me than the PCM version. Maybe all DSD recordings are just better mastered, whatever the reason DSD has a silky, life-like sound. SACD's are particularly interesting. Whether it is an apples to apples comparison or not the SACD's I own sound better than the PCM versions. This is particularly true of the Brothers In Arms album by Dire Straits and Jeff Wayne's Musical Version of The War Of The World. If you get a chance give them a listen. Audible or not, Dirac Live software is fully capable of handling a 24/96 or 24/192 music file without downsampling to 24/48. In the case of Emotiva processors and Dirac Live the limitation lies with the processor. There are other processors that do not have the 24/48 Dirac Live limitation. As to your last point about noise, a computer is about as noisy an environment as is imaginable. The noise level in a purpose built device like an UltraRendu streamer is far less noisy than a Windows or Mac computer. And, while Ethernet over Fiber is not a perfect solution it does reduce noise vs a standard Ethernet cable. The BlueSound node is not expensive and its DAC is a limitation compared to the DAC in an XMC-2. However, if you use the BlueSound Node with the XMC-2's DAC vs a Mac or Windows computer directly into an XMC-2 the BlueSound / XMC-2 combo will sound better. Unfortunately the Bluesound's power supply is less than stellar. Anyway, the bottom line is reducing RFI and EMI in a system allows a system to sound its best, especially relative to soundstage. I think it is great that you are willing to discuss stuff like this. |
|
KeithL
Administrator
Posts: 10,261
|
Post by KeithL on Jan 20, 2023 12:18:33 GMT -5
That's pretty much my take on the whole question... I've seen some pretty strange stuff over my career... So I am quite open to the idea that all sorts of strange interference and interactions CAN exist... And that, when they do, they can be really difficult to track down, and even more difficult to fix... But I've also seen a lot of demonstrations of how people can imagine they hear things that really aren't there. And I absolutely admit that this has happened to me occasionally too. NONE of us is immune to it. (I spent a whole day once trying to track down "a system hum" that turned out to be a wire vibrating in the wall.) HOWEVER, that said, I've also seen an awful lot of unnecessary audio fixes and tweaks... - some based on the threat of problems that don't actually exist - some based on the ability to fix real problems that only a very few people actually have - some based on the idea that, even if you don't hear a problem, you'll hear an improvement if you fix the problem you don't hear - and more than a few based on the claim that they can fix a problem for which no fix actually exists My favorite one is the claims about various ways to "reduce bit errors on CDs".... I've probably so far ripped about 5000 songs from CDs... And the program I use actually compares the results to a checksum database (so it will detect and flag A SINGLE INCORRECT BIT)... And, out of those 5000 files, I can recall exactly three songs in total that had one or more bad bits - when read on a crappy computer CD drive. (Two were visibly scratched... and one had an actual error on the master which showed up in the same place on multiple copies of the same disc.) So, that being the case, I can't figure out what the little edge mill, and the green magic markers, and all that other crap that is claimed to "eliminate errors on your CDs", is supposed to actually be "fixing"... (Either some of them actually help with really beat-up CDs... and a lot of people really beat up their CDs... or a lot of people are imagining they hear differences that aren't really there.) Concerning RFI / EMI? Seems like 2 possible entries for such nasties into a system. Over power lines. Most of thie can be gotten rid of with a good Power Conditioner. Or in some cases an over-capacity Isolation Transformer......I have both. Isolated circuits may also help. A lot of that 'bad' stuff simply won't get thru a Toroid.... OR Thru the air. Transmitted, if you will. good grounded casework may help. Ever lived near a commercial radio transmitter? Or an airport Radar system? It's like getting a local AM station thru one of your FILLINGS. Tough to figure out the fix. I'd suggest getting a good AM radio and doing a 'crawl'. If you can hear bad stuff on the radio or it is 'blanked' due to RFI? Than you have an issue...... |
|
KeithL
Administrator
Posts: 10,261
|
Post by KeithL on Jan 20, 2023 13:24:24 GMT -5
I realized that I had written a really long response... but had neglected to touch on Dirac Live.... so here goes... I do agree that it would be nice if we could run Dirac Live at 96k (and that may still happen someday). However, as we've already discussed, at the moment it's a limitation based on the processing power available, and what Dirac Live needs to run... (And, when you want to run Dirac Live on 16 channels, it does need quite a bit of processing power to run.) But, to be honest, either way, I consider that to be something that "would be nice in principle but probably doesn't matter in practice". As I've mentioned before.... Dirac Live does major processing on your audio signal... so it is NOT "a purist solution". If you're going to worry about "purist questions", like whether your audio signal is being resampled or not, then you probably shouldn't be running Dirac Live either. (You should probably be using Reference Stereo mode if that's how you feel.) Dirac Live is going to be altering your signal a lot more than simply resampling it from 96k to 48k..... Once you've decided to process your audio signal via Dirac Live, you've given up on "being a purist", so all you should really care about is how the end result sounds. And, to be quite blunt, I have no reason to even suspect that someone else's processor running Dirac Live at 96k is going to sound better, or even as good as, our processor running Dirac Live at 48k. (I think most of us agree that our processors sound noticeably better than our competitors.) At MOST, if it's actually audible at all, the difference between 48k and 96k is going to be far less than the difference you get with Dirac Live... And if, like most people, you find Dirac Live to deliver an audible improvement, then that should be all that really matters... (Or, to put it another way, if nobody had told you that Dirac Live was running at 48k, would you know... or care?) ................................. If you send a 24/96 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/96. If you send a 24/192 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/192. A DAC that receives a high resolution file does not downsample that file to 16/44.1 and start over with upsampling. There are several recording studios that record a performance in PCM and DSD at the same time. In those cases it is possible to compare DSD to PCM. In my experience, a true DSD recording has a uniquely silky signature that PCM does not seem to present. The DSD version of the music I own sounds better to me than the PCM version. Maybe all DSD recordings are just better mastered, whatever the reason DSD has a silky, life-like sound. SACD's are particularly interesting. Whether it is an apples to apples comparison or not the SACD's I own sound better than the PCM versions. This is particularly true of the Brothers In Arms album by Dire Straits and Jeff Wayne's Musical Version of The War Of The World. If you get a chance give them a listen. Audible or not, Dirac Live software is fully capable of handling a 24/96 or 24/192 music file without downsampling to 24/48. In the case of Emotiva processors and Dirac Live the limitation lies with the processor. There are other processors that do not have the 24/48 Dirac Live limitation. As to your last point about noise, a computer is about as noisy an environment as is imaginable. The noise level in a purpose built device like an UltraRendu streamer is far less noisy than a Windows or Mac computer. And, while Ethernet over Fiber is not a perfect solution it does reduce noise vs a standard Ethernet cable. The BlueSound node is not expensive and its DAC is a limitation compared to the DAC in an XMC-2. However, if you use the BlueSound Node with the XMC-2's DAC vs a Mac or Windows computer directly into an XMC-2 the BlueSound / XMC-2 combo will sound better. Unfortunately the Bluesound's power supply is less than stellar. Anyway, the bottom line is reducing RFI and EMI in a system allows a system to sound its best, especially relative to soundstage. I think it is great that you are willing to discuss stuff like this. |
|
|
|
Post by fbczar on Jan 20, 2023 14:15:34 GMT -5
There's rather more going on inside a modern DAC than "basic DAC theory" would seem to suggest. For example... An R2R DAC really does work pretty much like the diagram you saw in an old textbook... with a ladder of resistors or reference voltages, a row of switches, and some filters. However a Delta-Sigma DAC looks almost nothing like the "one-bit DAC" that most people imagine. Most Delta-Sigma DACs are actually a hybrid and internally use a four or five bit "ladder section" rather than a single bit converter. Technically "oversampling" is just a specific type or category of "upsampling"... However, in practice, they are usually done very differently, and in different places in the process... That difference between "upsampling" and "oversampling" is somewhat arbitrary in theory but rather more distinct in practice and implementation... The term "upsampling" is generally used to refer any time you re-sample a digital audio signal to pretty much any higher sample rate - so, for example, you might upsample a 24/44k signal to 24/96k. But, in practice, the term "oversampling" is generally used to refer specifically to upsampling to an even multiple of 2x the original sample rate - so 2x, or 4x, or 64x. Also, in practice, "oversampling" is almost always done by the digital circuitry inside the DAC chip, or its associated external filter, after the original multi-bit data has been "reduced" to the form in which it will be processed internally. So you might "upsample" a 44k file to 96k using your favorite sample rate conversion program... (and there are a few DAC products that do this using a processor before sending the data to the DAC chip). And a DAC chip might be configured to oversample to fs x64 based on a specific master clock... But, in most cases, the two are really independent, and the DAC chip itself is going to use oversampling internally anyway, after it receives the signal, to raise it to a much higher sample rate for processing. (And, yes, this makes upsampling it before that point somewhat redundant... unless you are using a specific upsampling filter or process which produces a result that you find pleasing.) The argument, which might be true in some specific cases, is that their particular DAC has been "optimized" to work best when fed audio at a specific sample rate. (The original Benchmark DACs - the DAC1 and DAC1/USB - converted all incoming audio data to a sample rate of 110k, which is a non-standard sample rate, for this reason...) However, suffice it to say that, in practice, most modern DAC chips run at very high clock rates and still apply oversampling to incoming data - even at 96k or 192k. (So, while you might achieve some benefit from using higher-sample-rate audio, you probably will not avoid internal oversampling.) As we seem to agree it really is impossible to perform an "apples-to-apples comparison" between PCM and DSD... For example, even if someone "records the same performance on both DSD and PCM at the same time" they are doing so using different recorders, with different analog-to-digital conversion hardware, and different analog audio paths. And, even if they use two of the same model of recorder, there may be differences between the DSD and PCM signal paths that are not specifically associated with the format itself... And these may include simple differences in the conversion hardware used in each signal path or differences in how the hardware and filters in each are configured. And, even beyond unintentional differences, we may even wonder if the DSD signal path has been "voiced differently to produce a result that is more pleasing to audiophiles". While we know that this often occurs when SACDs and CDs are mastered we cannot rule out that it also happens with the hardware itself. (In simplest terms we can never know "if everyone did their best to produce results that are as accurate as possible" with either format.) Incidentally I absolutely agree that the SACD versions of many albums do sound better than the CD versions. I've heard the Dire Straits album and the SACD version does sound very nice... Likewise for a few of the SACD versions of Dark Side of the Moon... and The Eagles Hotel California... and several of Mike Oldfield's albums. And my favorite version of my all-time favorite album (Renaissance - Scheherazade and Other Stories) is the Audio Fidelity SACD version. HOWEVER, while I cannot say that I've done the comparison with all of them in detail... I have actually taken ISO copies of the DSD content on a few of those... and converted them into 24/96k PCM... And, when I have done so, I have been unable to notice any difference that I would characterize as either "sounding better". (As I've said before there are in fact really tiny differences... which could be attributed to the conversion process or the DACs involved... but I would not say that either version sounded "better".) I would also be remiss if I didn't mention that I'm pretty sure that neither of those albums you mentioned was RECORDED in DSD. Therefore both were in fact converted to DSD from either analog tape of PCM. (I'm too lazy to look but I'm guessing that at least Dire Straits started out on analog tape... which is FAR less accurate than either DSD or PCM.) I'm going to be fair here and start by saying that, although I've read their marketing literature, I have neither listened to, or taken apart and examined, any of the Rendu products. (And, yes, we all know that neither Windows nor Apple computers are designed with much of any consideration towards minimizing noise at all.) HOWEVER, that said, we must agree that any such streaming device is basically a small computer; a digital device which runs on high frequency square waves. Now, on the one hand, I have no doubt that they have put some work into minimizing the "stray and unnecessary noise"... perhaps with significant success. But it's still a computer, taking in Ethernet packets, processing them digitally, and putting out a digital audio signal... (so it's still about as far from analog as you can get). So, considering the fact that a properly designed DAC is designed to be largely immune to this sort of noise (in the context of exposure to it from the audio source)... I'm not honestly convinced about exactly how much optimizing parts of that process matters in actual practice (performance)... For example, if I'm paying extra for better performance, I would like to see actual numbers showing better performance, beyond "it seems like a good idea". (My guess would be that some DACs are far more sensitive to that than others... so your results would also tend to vary rather widely depending on what DAC and other equipment you have.) I'm also sort of forced to draw a humorous parallel to an episode of Get Smart... Where the bad guys invented "a silent explosive" that could "blow things up while making no noise whatsoever"... If you send a 24/96 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/96. If you send a 24/192 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/192. A DAC that receives a high resolution file does not downsample that file to 16/44.1 and start over with upsampling. There are several recording studios that record a performance in PCM and DSD at the same time. In those cases it is possible to compare DSD to PCM. In my experience, a true DSD recording has a uniquely silky signature that PCM does not seem to present. The DSD version of the music I own sounds better to me than the PCM version. Maybe all DSD recordings are just better mastered, whatever the reason DSD has a silky, life-like sound. SACD's are particularly interesting. Whether it is an apples to apples comparison or not the SACD's I own sound better than the PCM versions. This is particularly true of the Brothers In Arms album by Dire Straits and Jeff Wayne's Musical Version of The War Of The World. If you get a chance give them a listen. Audible or not, Dirac Live software is fully capable of handling a 24/96 or 24/192 music file without downsampling to 24/48. In the case of Emotiva processors and Dirac Live the limitation lies with the processor. There are other processors that do not have the 24/48 Dirac Live limitation. As to your last point about noise, a computer is about as noisy an environment as is imaginable. The noise level in a purpose built device like an UltraRendu streamer is far less noisy than a Windows or Mac computer. And, while Ethernet over Fiber is not a perfect solution it does reduce noise vs a standard Ethernet cable. The BlueSound node is not expensive and its DAC is a limitation compared to the DAC in an XMC-2. However, if you use the BlueSound Node with the XMC-2's DAC vs a Mac or Windows computer directly into an XMC-2 the BlueSound / XMC-2 combo will sound better. Unfortunately the Bluesound's power supply is less than stellar. Anyway, the bottom line is reducing RFI and EMI in a system allows a system to sound its best, especially relative to soundstage. I think it is great that you are willing to discuss stuff like this. Whether you upsample to 24/192, double the sampling rate of the file, or upsample to the maximum sample rate your DAC can handle ( all of which can be done in Audirvana or HQPlayer), when that upsampled file is submitted to the DAC the DAC will not perform upsampling at a rate lower than the file submitted. The concept being that the DAC has less work to do and will perform better as a result. So at least in some sense I think we agree. While there may be no "perfect" way to compare DSD to PCM in my experience the sound differences are always the same. So all else being equal I prefer a DSD file. No doubt the resolving power of the system the file is played on makes a difference. I was super excited by the XMC-2 because it was supposed to be able to play DSD over USB. It is a shame it never will. Emotiva must have thought DSD had value at one time. By the way, the Dire Straits, Brother's In Arms album was the first 100% digital album and as you know the SACD is vastly superior to any CD, even the Japanese remasters. Was it originally DSD? I do not know. I appreciate the fact that the XMC-2 plays SACD so well. The Rendu's may well be "small computers", but the difference in the internal noise generated by their computing processes is as the east is from the west relative to a standard Mac or Windows machine. I guess you could say a Ferrari is just another car when comparing one to a Yugo. No doubt, most purpose built streamers are quieter than standard computers, but some are better than others. In my experience an UltraRendu is much better than a BlueSound Node but both are better than my Mac or my Windows machines. Streamers with linear power supplies do have an advantage. A comparison would be easy to do. Just connect a streamer to you computer via USB and then connect the streamer to an Emotiva processor via USB. Surely someone at Emotiva has a dedicated streamer. That might not give you the data you would like to read, but if you trust your ears I am sure a listening comparison would point to the legitimacy of my argument. Of course there are much better ways to setup a dedicated streamer than a direct connection to a computer, but even that type of setup will sound much improved. All you have to do is listen. |
|
|
|
Post by fbczar on Jan 20, 2023 14:28:45 GMT -5
I realized that I had written a really long response... but had neglected to touch on Dirac Live.... so here goes... I do agree that it would be nice if we could run Dirac Live at 96k (and that may still happen someday). However, as we've already discussed, at the moment it's a limitation based on the processing power available, and what Dirac Live needs to run... (And, when you want to run Dirac Live on 16 channels, it does need quite a bit of processing power to run.) But, to be honest, either way, I consider that to be something that "would be nice in principle but probably doesn't matter in practice". As I've mentioned before.... Dirac Live does major processing on your audio signal... so it is NOT "a purist solution". If you're going to worry about "purist questions", like whether your audio signal is being resampled or not, then you probably shouldn't be running Dirac Live either. (You should probably be using Reference Stereo mode if that's how you feel.) Dirac Live is going to be altering your signal a lot more than simply resampling it from 96k to 48k..... Once you've decided to process your audio signal via Dirac Live, you've given up on "being a purist", so all you should really care about is how the end result sounds. And, to be quite blunt, I have no reason to even suspect that someone else's processor running Dirac Live at 96k is going to sound better, or even as good as, our processor running Dirac Live at 48k. (I think most of us agree that our processors sound noticeably better than our competitors.) At MOST, if it's actually audible at all, the difference between 48k and 96k is going to be far less than the difference you get with Dirac Live... And if, like most people, you find Dirac Live to deliver an audible improvement, then that should be all that really matters... (Or, to put it another way, if nobody had told you that Dirac Live was running at 48k, would you know... or care?) If you send a 24/96 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/96. If you send a 24/192 file to an upsampling DAC the DAC will not perform any upsampling at or below 24/192. A DAC that receives a high resolution file does not downsample that file to 16/44.1 and start over with upsampling. There are several recording studios that record a performance in PCM and DSD at the same time. In those cases it is possible to compare DSD to PCM. In my experience, a true DSD recording has a uniquely silky signature that PCM does not seem to present. The DSD version of the music I own sounds better to me than the PCM version. Maybe all DSD recordings are just better mastered, whatever the reason DSD has a silky, life-like sound. SACD's are particularly interesting. Whether it is an apples to apples comparison or not the SACD's I own sound better than the PCM versions. This is particularly true of the Brothers In Arms album by Dire Straits and Jeff Wayne's Musical Version of The War Of The World. If you get a chance give them a listen. Audible or not, Dirac Live software is fully capable of handling a 24/96 or 24/192 music file without downsampling to 24/48. In the case of Emotiva processors and Dirac Live the limitation lies with the processor. There are other processors that do not have the 24/48 Dirac Live limitation. As to your last point about noise, a computer is about as noisy an environment as is imaginable. The noise level in a purpose built device like an UltraRendu streamer is far less noisy than a Windows or Mac computer. And, while Ethernet over Fiber is not a perfect solution it does reduce noise vs a standard Ethernet cable. The BlueSound node is not expensive and its DAC is a limitation compared to the DAC in an XMC-2. However, if you use the BlueSound Node with the XMC-2's DAC vs a Mac or Windows computer directly into an XMC-2 the BlueSound / XMC-2 combo will sound better. Unfortunately the Bluesound's power supply is less than stellar. Anyway, the bottom line is reducing RFI and EMI in a system allows a system to sound its best, especially relative to soundstage. I think it is great that you are willing to discuss stuff like this. The question relative to a higher sampling rate for Dirac Live has nothing to do with a "purist"viewpoint. It has to do with whether or not a 48K sampling rate is better than a 96K rate relative to sound quality and whether or not downsampling is bad. As is, we will never know. I think Dirac is great, but I also am disappointed that my XMC-2 cannot provide a 24/96 sampling rate for Dirac. When you have lots of high resolution files and you think you are getting a processor that can play them at a 24/96 rate with Dirac it is a downer. Thanks again for taking the time to discuss all this. |
|
KeithL
Administrator
Posts: 10,261
|
Post by KeithL on Jan 20, 2023 18:21:04 GMT -5
I think we may have to agree to disagree there... at least to a point. When I purchase or create a 96k file it is with the idea of preserving more of the "original information". When I run Dirac I assume I am "trading the original information for something new that I hope will sound nicer". (After all, what Dirac Live is actually doing is "correcting for deficiencies in my room and speakers - at the cost of extra digital processing".) To me that's something more of an "either/or" choice... Much like whether I play something in Reference Stereo mode or using Dirac Live... If I want an accurate rendition of the original them I'm NOT going to allow Dirac to modify it... In fact, I'm probably going to use Reference Stereo, so I can be sure that nothing alters it. And, if I DO allow Dirac Live to modify the signal, it's because I'm convinced that those changes will be an improvement... But, in return for that audible improvement, I have already given up any notion of "an accurate rendition of the original". So all I'm interested in is how well the "new modified version" sounds. As I said, I too might prefer if we could run Dirac Live at 96k, without sacrificing anything else... But, at the moment, that is not an option... I realized that I had written a really long response... but had neglected to touch on Dirac Live.... so here goes... I do agree that it would be nice if we could run Dirac Live at 96k (and that may still happen someday). However, as we've already discussed, at the moment it's a limitation based on the processing power available, and what Dirac Live needs to run... (And, when you want to run Dirac Live on 16 channels, it does need quite a bit of processing power to run.) But, to be honest, either way, I consider that to be something that "would be nice in principle but probably doesn't matter in practice". As I've mentioned before.... Dirac Live does major processing on your audio signal... so it is NOT "a purist solution". If you're going to worry about "purist questions", like whether your audio signal is being resampled or not, then you probably shouldn't be running Dirac Live either. (You should probably be using Reference Stereo mode if that's how you feel.) Dirac Live is going to be altering your signal a lot more than simply resampling it from 96k to 48k..... Once you've decided to process your audio signal via Dirac Live, you've given up on "being a purist", so all you should really care about is how the end result sounds. And, to be quite blunt, I have no reason to even suspect that someone else's processor running Dirac Live at 96k is going to sound better, or even as good as, our processor running Dirac Live at 48k. (I think most of us agree that our processors sound noticeably better than our competitors.) At MOST, if it's actually audible at all, the difference between 48k and 96k is going to be far less than the difference you get with Dirac Live... And if, like most people, you find Dirac Live to deliver an audible improvement, then that should be all that really matters... (Or, to put it another way, if nobody had told you that Dirac Live was running at 48k, would you know... or care?) The question relative to a higher sampling rate for Dirac Live has nothing to do with a "purist"viewpoint. It has to do with whether or not a 48K sampling rate is better than a 96K rate relative to sound quality and whether or not downsampling is bad. As is, we will never know. I think Dirac is great, but I also am disappointed that my XMC-2 cannot provide a 24/96 sampling rate for Dirac. When you have lots of high resolution files and you think you are getting a processor that can play them at a 24/96 rate with Dirac it is a downer. Thanks again for taking the time to discuss all this. |
|
|
|
Post by leonski on Jan 20, 2023 18:55:05 GMT -5
ONE solution to Airborne or RFI? Have a large isolation transformer or more....feed all the circuits in the listening room. THAN? Have the room constructed as a Faraday Cage...... Now all you have to worry about is a piece of gear inside the room emitting radiation which interferes with another piece...... www.livescience.com/what-is-a-faraday-cageBased on certain readings, this is how Secure Rooms at embassys are organized and with walls-within-walls and other security techniques. |
|
|
|
Post by drumace on Jan 27, 2023 20:31:59 GMT -5
|
|
|
|
Post by leonski on Jan 28, 2023 21:38:41 GMT -5
I still don't understand how 48k is anything But a 'technical' limit.
It is well over double the highest frequency you could reasonably be interested in......
That is? If 48khz is the sample rate limit or if a frequency limit, than the sample rate would be about double THAT.....or 96khz....
|
|
KeithL
Administrator
Posts: 10,261
|
Post by KeithL on Jan 30, 2023 13:03:41 GMT -5
It's a nice thought... but you're oversimplifying the actual practical details. For one thing there are very few things these days that don't generate some form of high frequency "interference". To start with ANY DIGITAL DEVICE essentially works by virtue of high frequency square waves... Therefore they are all very effective sources of noise... In our example that would mean leaving the streaming device, which is itself a source of noise, outside the room... And somehow figuring out how to get the music into the room but not any noise... (Remember that, for example, that rules out optical cables... because, even though the cable is perfectly isolated, its receiver is itself a digital noise source.) (And, obviously, your cell phone, remote control, and TV, all being great sources of interference, must all stay outside the room.) Also, in all fairness, building a "totally shielded room" is virtually impossible in practice... Besides which the REAL goal is to make sure that "no INTELLIGIBLE signal escapes the room". So the "practical compromise" is to shield the room as well as you can... Then deliberately add a source of random noise such that you effectively "mask any remaining signal with the noise"... (A Faraday Cage is only 100% effective if you literally never allow anyone, or any data, to leave the room, and never open the door.) HOWEVER, this is all a bit of a rabbit hole, because in practice we both do our best to limit noise emissions, and to limit the sensitivity of our gear TO those noise emissions. Therefore the only simple question is of whether a given piece of gear offers a real benefit in this regard... or whether "it just sounds better on the marketing brochure". And part of the problem there is that we really don't have sufficient data to make an informed decision. Neither of us actually knows the noise emission spectra of a variety of different streamers... nor how sensitive a given piece of gear is to what part of the noise spectrum. I'm reminded of a few totally idiotic "solutions" that add shielding against microwave interference to audio gear... They're only idiotic because they are attempting to shield against noise that simply doesn't matter - because audio circuitry is rarely, if ever, sensitive to it. (Which makes the question of whether that noise is even there in significant quantities to begin with doubly moot.) I should also note that, while isolation transformers OFTEN reduce the levels of RF interference (because they have lots of inductance). Not all isolation transformers are equally effective at doing so. You have to think about capacitive leakage between the windings... and about "incidental leakage" around the entire transformer... for example in the feed wiring. And also the fact that the isolation transformer is eliminating the original ground path... And either replacing it with a new and perhaps more interesting ground path... Or failing to do so (in which you may have issues due to "not having a real ground".) Isolation transformers are indeed often good solutions to certain problems... but they are far from "a universal panacea". ONE solution to Airborne or RFI? Have a large isolation transformer or more....feed all the circuits in the listening room. THAN? Have the room constructed as a Faraday Cage...... Now all you have to worry about is a piece of gear inside the room emitting radiation which interferes with another piece...... www.livescience.com/what-is-a-faraday-cageBased on certain readings, this is how Secure Rooms at embassys are organized and with walls-within-walls and other security techniques. |
|
KeithL
Administrator
Posts: 10,261
|
Post by KeithL on Jan 30, 2023 13:13:05 GMT -5
The "frequency limit" is "the Nyquist frequency" - which is half of the sample rate. So, at least theoretically, a 48k sample rate will allow you to accurately reproduce up to 24 kHz. (Which "provides a bit of extra clearance beyond the 22 kHz limit of a 44k sample rate".) In practice the biggest reason 48k is "a magic number" is simply that it has been a standard in the VIDEO PRODUCTION industry for a long time. (So most audio FOR VIDEO has traditionally been edited and distributed at 48k.) The 44.1k used for CDs was chosen as being "as close as possible" to what was needed to "pass 20 Hz to 20 kHz". This choice was made back in the day when there was serious concern about exactly how many minutes of music would fit on a CD disc. All disagreement beyond that is based on: - the idea that we really can hear well above 20 kHz - the idea that, even though we only hear up to 20 kHz, the "extra frequency headroom" avoids certain problems that "affect audible performance below 20 kHz" (this is the very real logic behind oversampling - which, by raising the sample rate much higher, enables us to design simpler and cheaper filters that work better below 20 kHz) In fact "it's ALL just 'technical limits' "... Although that could mean "what the hardware can do" or "what the software you have that goes with that hardware supports"... (it's not usually anywhere near as simple as "just get a more powerful computer and run it faster".) I still don't understand how 48k is anything But a 'technical' limit. It is well over double the highest frequency you could reasonably be interested in...... That is? If 48khz is the sample rate limit or if a frequency limit, than the sample rate would be about double THAT.....or 96khz.... |
|
|
|
Post by leonski on Jan 31, 2023 18:47:01 GMT -5
Just got back from a run to get some motor oil and filters. changes will be needed before summer hits.
I was at the local WalMart which has thousands of square feet of solar. Driving under is nice in the heat of summer.
At that point I was trying to listen to some AM radio. News? Weather? Sports? The usual stuff.
But the RFI was crazy. And centered under the panel where you have Dozens of 'distributed' inverters.
These things generate a real RFI mess.
I'd urge anyone with home solar to do an RFI crawl with a Good AM Radio. Is it possible that a Solar PV installation
is bad for your stereo?
|
|
