Do any of you hear any difference between high vs normal resolution material via the DirectStream? I’ve downloaded and listened to samples from Barry Diament’s Sound Keeper Records site at 16/44, 24/96, 24/192 http://www.soundkeeperrecordings.com/format.htm
According to Barry (if I recall correctly), the difference between 24/96 vs. 24/192 is more significant than 16/44 vs. 24/96 as if some kind of threshold is crossed.
For the life of me I don’t hear a difference between the different resolutions. While I can’t rule out my audio system or my aging ears (which I know are not the best anymore), I can distinguish a DSD played from .iso vs. a DSD decoded to .dsf file beforehand.
Of course DirectStream upsamples to 10x DSD which could be why I hear no difference. Barry writes on his site “When comparing the files, be sure your system is not performing any sort of resampling and/or dithering.” Afaik there’s no way to defeat the DirectStream upsampling. So, maybe the upsampling defeats the difference which would be awesome.
Am perfectly okay with not hearing a difference (and avoid buying high res material) but if there is a difference then I’d want to find out if my system is lacking and I got work to do or my hearing is the limiting factor and should just enjoy what’s left.
Any of you hear a difference? What clues should I listen for?
PS. Hope PS Audio, Barry Diament don’t mind me posting a link. Thanks to Barry Diament for making samples available to compare.
I downloaded the Sound Keeper Recordings sample tracks and I am having a hard time discerning any difference between sample rates. Okay, I don’t have the most discerning ears, nor the most resolving system. Not bad, just not great: Foobar2000 running on a general purpose AMD powered Windoze 8.1 computer feeding an LH Labs Pulse X infinity DAC to a PS Audio Trio C-100 driving a pair of Magnepan MMGs (sitting ON the computer desktop, flanking a 24" monitor).
In listening to the sample tracks, to my ear they seem to be very well recorded and mastered. Strangely, it seems that better mastering may have more apparent effect on lower sample rates than it does on higher sample rates. I can tell from the spectrum analyzer that I have running in Foobar that there is more high frequency information in the higher sample rates as you would expect if the analog recording chain is good and not filtered before being fed to the ADC. Don’t count any of that as any profound statement, as I am only an audiophile, not a recording professional or even amateur. Still, I too find it curious that I can hear so little difference (if any) between sample rates.
I might even be tempted to buy one or two of his albums, but if I do I would be tempted to go with the redbook version, or perhaps the 24/96 rate if I were to splurge. That said, I have heard more differences between sample rates on material from other sources and in many cases would be more inclined toward higher than redbook sample rates with those… I can’t explain why I hear so little difference with his material.
LoBro89 said
Any of you hear a difference? What clues should I listen for?
To me Redbook can sound amazing good, but still as you go down in resolution things get a little more ghostly, a little less present, perhaps a little less real. I hear the same soundstage size, same soundstage positioning, same frequency response (e.g. similar lows, similar highs.) Perhaps a little more air on the top with higher res. The differences are more apparent to me after hearing high res for a little while and then going back to a lower resolution.
My observations in this regard are consistent with what Ted has said. I’d only add that I wouldn’t go buying Hi-Res editions of CD that I already own because the differences aren’t worth the price of admission to the higher res format. And also, in my experience with buying from HD Tracks, the CD version can sound better than the higher-res downloaded version. I did complain to HD Tracks about this, that there should be some buyers remorse consideration when their high-res content doesn’t sound as good as the original CD. They basically said “bad luck”. I haven’t bought anything from them since.
After spending a whole day re-wiring yesterday (aka “spring cleaning” as per Paul’s Posts), my PS stack is now entirely powered by PS regenerated power carried on PS power cables. Now that is worth the price of admission.
I do hear differences as I go up the resolution ladder, and I often find that well-done DSD has a bit more analog-like smoothness that I appreciate.
That said, what really matters is not the format or resolution, but the skill of the engineers. Any level of recording can give you good sound that is musically involving, if the engineers (and performers, of course) really know what they are doing. Redbook can sound amazing, especially through a transport like the DMP. I have some 24/96 recordings that you would think are 24/192.
Seek out good performances that are well recorded and well mastered, regardless of the bits involved, and be happy.
I have been reading about, and listening to hi-rez for many years now. I really think that with PCM, 24/96 is sufficient for anything analog, like master tapes. It puts the filters up high enough, and easily covers everything on the tapes.
With DSD, I haven’t heard enough of the different levels, mostly just SACD files which are 64(?) I think. I know they aren’t the double or triple DSD. Right after I got those DSF files, I switched DACs to one that doesn’t do DSD, so now JRiver is outputting them at 24/176, and I think they sound great.
I was reading part of an email from Cookie Marinco about the amount of storage used to record in DSD. I realize storage is not really an issue for a studio, but for home use there are limits for some of us. I’m on a tight budget, so I only have one external drive for backup. I would really like to have at least two more 2 Tb drives, so I could conveniently store one off the premises, and alternate them. Of course if I was rich, I would buy some dedicated players.
One more thing I would like is a external disc drive that would not run off USB. I have seen a few nice ones, but all over $150. I think Melco recommends a Buffalo model.
The improvements occasioned by higher resolution of a good recording are fairly subtle. They are rendered even more subtle when played on a superb DAC such as the DirectStream which makes 44/16 sound excellent.
As already mentioned, the quality of the recording matters much, much more.
wingsounds13 said
....driving a pair of Magnepan MMGs (sitting ON the computer desktop, flanking a 24" monitor).
I am trying to imagine how you fit a pair of MMG's on your desk
They do kind of dominate the desktop, however, my computer desk is a solid core door sitting on a cheap pair of 2 drawer filing cabinets so I have a bit more desktop real estate than most desks. This is not an ideal setup either, as they are only about 16~18 inches away from the wall. Still… MMGs playing near field do sound pretty good. I am thinking about building some dowel array diffusers to put behind them to see what that does for this crazy setup. Yes, I admit that MMGs on the desktop is a crazy setup, but they only cost about a third of what the mini-maggies would have.
Of course DirectStream upsamples to 10x DSD which could be why I hear no difference. Barry writes on his site “When comparing the files, be sure your system is not performing any sort of resampling and/or dithering.” Afaik there’s no way to defeat the DirectStream upsampling. So, maybe the upsampling defeats the difference which would be awesome.
I think this is an important thing to note. In my own experience with a few different DACs, the biggest difference I hear between 16/44.1 and higher resolution files is a certain kind of “smeary” sound that I’ve come to believe isn’t lack of resolution in the source material, but rather an indicator of mediocre digital filtering as part of the PCM reconstruction process that messes with timing and phase of the resulting analog signal. Upsampling in software like Roon (if the DAC is capable of higher resolution input) seems to clear up the smearing noticeably, as it makes the filtering done by the DAC less audible, because it can be more gentle when it isn’t operating so close to the audible frequencies. In the case of the DirectStream, the upsampling algorithm seems to be quite good, as it’s the only DAC I own where upsampling 16/44.1 material before it gets to the DAC doesn’t result in noticeably clearer or “better” sounding music. I do notice subtle differences in the sound in the case of the DirectStream, but I couldn’t really call either one better than the other, so I just let the DS do the work.
Yes. The DS does wonders w 16/44. I have some albums on cd from the mid to late 70s that I stopped listening to because of the flat presentation. Not no mo. These things sound great. I kid you not. Saying that - the best recordings I have are double DSD. As others have said - there are slight differences - but in some cases the higher res recordings are just more immersive in my opinion.
Great responses! Ted’s response is great to know what to listen for. Difference is mostly in the higher frequencies (duh!) explains for some (probably for me) why we’re not hearing much of a difference, if at all. Though Ted’s description of the difference I would expect to be able to detect - will give it another try or two or three.
For the record for playback I’m using i7 Windows, 16GB Ram, .WAV files on external SSD, Roon, Bridge II with CAT 7 crossover LAN cable between computer and bridge).
Double blind tests don’t always give reliable results:
You need to train yourself for what to expect with each version. Make sure you listen over and over to each until you are very comfortable that you know what’s coming next. Then try to A/B them.
But longer term listening is more reliable for many of us: Listen for a while (perhaps hours or days) to, say DSD and then a while to something else. Then DSD again, then the something else. You need to get used to the sound of one before and then get used to the other. Then when you go back and forth you may more clearly hear the differences.
Also pay attention to whether your feet start tapping on one and not the other. (Or as my wife says, whether she want’s to dance to one or the other.)
Still, at times, I know I’m not hearing a difference when other people say they are. But I don’t worry about it we all listen for different things and we all have different thresholds for how much of a difference matters.
I thought I could tell the differences until i did a double blind test with a friend… we played the following versions of the same song on my DSD Snr over the Bridge II via Roon:
Mastered for iTunes purchase
CD rip in FLAC
HDTracks purchase
DSD purchase on Acoustic Sounds
I got them all wrong. I was most embarrassed at picking the DSD purchase for the Mastered for iTunes copy played and picking the iTunes copy for the HDtracks played.
An absolute disaster of a day, and this is very tough to admit.
It was an album I know really well, Thriller. I think it was the 1st track from memory.
We all think he can hear the differences but you may be surprised in a double blind test. I don’t think I ever want to do one again lol. Expectation bias is a real thing.
But hey, it must be the DSD Snr working it’s magic because the most important thing was they ALL sounded really great to me. It was really difficult to pick which was which in the test - pretty sure they were all guesses, that’s how difficult it was.
Blind tests are valid and useful but, as Ted notes, you need to know what you are listening for. Otherwise the blind test results will be random. The persons tested must be able to perceive the differences for the results to be valid.
The average person will not hear a difference between a good MP3 and a CD on a typical system. But many of us can because we have experience and know what to listen for. A blind test of the first group may provide evidence there is no difference to be perceived by randomly chosen people. Perhaps helpful, but it does not test whether there is a difference between Redbook and MP3.
My experience is like Ted’s. Unless the difference is profound, I need to listen for a long time to learn everything I can about the differences. Even if the difference is profound, it takes time for me to appreciate the various differences and nuances.
After switching back and forth like this, a blind test can be very helpful to test whether what you hear is real or your own perception bias.
I continue to find interesting that the differences I hear between Redbook and high resolution become smaller the better the playback equipment. My guess is many experience this with the DirectStream which produces spectacular results with Redbook.
Yawn. The required demonizing of DBTs has been accomplished, again. They have been faulted for alleged problems that actually affect any kind of listening test performed by any means. The difference is that DBTs are far more transparent in the sense that if the listener fails to hear a difference, that fact is clearly and unambiguously reported.
This is hardly the only case in life where objectivity and reliability make many people uncomfortable.
One of the effects of the false information that people are led to believe by inherently flawed listening tests with woefully inadequate bias control is shown in a previous post. There we can find an omnibus of false claims, such as the idea that listening to recorded music and in particular perceptually coded music results in brain damage. This latter claim is encapsulated in pseudio-scientific language that proves that the person who has manufactured the self-serving fantasy does not understand masking.
You see masking takes place in the ear, primarily in the Basilar Membrane. The operational strategy of the human senses is that the external sense organs are designed to perform aggressive data reduction, so that the stimulus that is passed to the brain is that which is most beneficial for that basic need: Survival. Therefore sonic stimuli have their masked components removed before transmission to the brain. The brain receives the masked results whether or not the source has had its masked components removed by the means of recording process.
Some of the most important characteristics of a listening test is that the program material being used be familiar to the listeners, and that it also have content that is actually changed by whatever audible failing the equipment under test actually has. Technical tests can be used effectively to select recordings that are more likely to produce positive results.
Therefore, if a listener is most familiar with classical music then he will probably get the most positive results possible with classical music, and if he is most familiar with rock and roll, then rock and roll music will probably allow him to do his best. In either case, the requirement that the music be actually significantly be adversely affected by the performance characterisic being studied is extremely important. Equipment often performs well, “except for…” so selection of the recordings of music and other sounds used in listening tests is one of the most important things.
As far as phase perfection goes, it is first and foremost practically impossible in any real world audio system. But not to worry, the human ear lacks the ability to discern it in terms of missing hardware in the ear, above about 1 KHz. At lower frequencies, the brain can be very effective at “listening through” local reverberation and produce sensitive detection. Both of these facilities have served us well for 100,000s of years which is why humans are the apex predators of the Earth.
One of the effects of the false information that people are led to believe by inherently flawed listening tests with woefully inadequate bias control is shown in a previous post. There we can find an omnibus of false claims, such as the idea that listening to recorded music and in particular perceptually coded music results in brain damage.
Which post?
As far as phase perfection goes, it is first and foremost practically impossible in any real world audio system. But not to worry, the human ear lacks the ability to discern it in terms of missing hardware in the ear, above about 1 KHz. At lower frequencies, the brain can be very effective at "listening through" local reverberation and produce sensitive detection. Both of these facilities have served us well for 100,000s of years which is why humans are the apex predators of the Earth.
Tho the ear/brain is insensitive to steady state phase, it is sensitive to the shape of transients which require the correct relative phase of it's components. We can easily hear when a radio station uses an all pass phase dispersion filter (with a flat gain response.) For me the intelligibility of speech (e.g. announcers) goes way down in a noisy environment like a car with such a radio station.
Yawn. The required demonizing of DBTs has been accomplished, again.
Not in the least. Blind, double-blind, and triple-blind tests are all exceedingly valuable, and can be in audio as well.
Some of the most important characteristics of a listening test is that the program material being used be familiar to the listeners, and that it also have content that is actually changed by whatever audible failing the equipment under test actually has.
Yep.
Tougher to control for is the abilities of the participants. If someone cannot distinguish two sources we do not know if this means only this person (or group) cannot, or if there truly is no difference. It always remains difficult to prove a negative.
