[castleofargh]

 

   
By default, Windows expects you to set a particular sample rate, and plays everything at that sample rate - and setting that sample rate to match the files you're playing would allow you to play them back without conversion. However, there's an easier way. Windows has a special audio mode called WASAPI; and, when you use WASAPI mode, each file will automatically be played back at its native sample rate. The catch is that not all programs support WASAPI mode, including most of the ones that come with Windows, (although all "real audiophile music players" do).
 
If you really want the best sound quality when playing high-res music, the easiest thing is to switch to a program that does support WASAPI mode. (Incidentally, WASAPI mode will take care of matching the sample rate; the bit depth generally takes care of itself in most cases.)
 
If you prefer not to do that, the next easiest thing is to set your default output sample rate to either 24/96 or 24/192 and just leave it there.... Basically, this will play high sample rate files at their proper sample rate, and up-sample those recorded at a lower sample rate (which, while it can't add information, and so can't actually improve them, generally won't do much damage).

Will Kernel Streaming or ASIO also play files at their native sample rate, just like WASAPI, or what?

Why then, when I use WASAPI, does the display on my Fiio X3 (in DAC mode) still always say it's receiving input at the default Windows rate (16/44.1) rather than the rate of the file being played?

they all go with a default value, not necessarily the native value of the file. so you have to look for that depending what you use. through widows mixer, everything turns into what the mixer setting have. for asio, by default it tend to go for the max bitrate the DAC has to offer, as it has no real drawback I find that to be a good choice. but some specific asio drivers may have another default value. wasapi does what you tel it to do. KS to me is a thing of the past, I never noticed any beefit but did have a few bugs.
but again those stuff can make you want to headbutt a wall when recording with all the places where you might have to check for bit depth and sample rate. but for music playback it pretty much always works.
 
 
 
@Music Alchemist  I believe it was a FF animation where I gave a bad score when you loved it. elfen lied has always been up there at the top.

 

[cjl]

  Yikes, there is an awful lot of incorrect science out there......
 
1) While the differences between the various sample rates and bit depths may not be obvious to all listeners, or with all equipment, or will all music, I think it's pretty widely agreed that they are audible a significant percentage of the time. Besides that, whenever you convert one sample rate to another you introduce slight alterations to your content. For both of those reasons, it seems to make obvious sense that you want to play your music bit-perfect. (Assuming that you want it to be as close to the original as possible, then you don't introduce extra changes, right?) I would suggest anyone wondering about this to listen for themselves.

This is false. Sample rates and bit depths above 16/44.1 have never (to my knowledge, though I'd love to be proven wrong) been shown to be audible in a rigorous scientific test. Similarly, competent resampling should not be audible between audibly transparent sample rates.

 

[goodyfresh]

  This is false. Sample rates and bit depths above 16/44.1 have never (to my knowledge, though I'd love to be proven wrong) been shown to be audible in a rigorous scientific test. Similarly, competent resampling should not be audible between audibly transparent sample rates.

Well I mean I looked it up and 16-bit depth has a SNR of 96db or so, and 44.1khz has a maximum reproducibble frequency of 20khz, PRECISELY THE UPPER LIMIT OF HUMAN HEARING.  So yeah, you're almost certainly correct, there's no way anything better than those measurements would actually be audible to even the most skilled of human ears.  Maybe their dogs and cats could hear the difference, though

 

[KeithEmo]

  This is false. Sample rates and bit depths above 16/44.1 have never (to my knowledge, though I'd love to be proven wrong) been shown to be audible in a rigorous scientific test. Similarly, competent resampling should not be audible between audibly transparent sample rates.

 
No - it is, in fact, technically correct. (Well, let's say that the fact that differences exist can be proven easily enough... and there do seem to be quite a few people, including myself, who do claim to hear differences - at least in some specific situations.)
 
Start with a 24/96 file (you pick it); re-sample it to 16/44 using any two audio programs (you pick them); and do a bit comparison between the results. They will NOT be bit-for-bit identical. Even beyond that, most of the higher end programs offer several different choices of which filter slope and type to use, so, even with the same program, you can produce an infinite number of close-but-not-identical results. Likewise, start with any 16/44 file you like, convert it to 24/96 using your favorite program, then convert it back to 16/44 using the same program. What you end up with won't be bit-for-bit the same as the original then either. (This is in direct contrast to converting a FLAC file to a WAV, then back to a FLAC, as many times as you like - and ending up with a file that IS identical to the original.)
 
To be totally truthful, I've never seen ANY of what I would call "well conducted rigorous testing" on that subject at all. (I've seen plenty of tests that showed that, under certain circumstances, the majority of a certain group of subjects didn't notice a specific difference, and I've seen tests that proved that specific people who believed they were able to hear specific differences under specific conditions were in fact unable to do so, but I've never seen a comprehensive test anywhere near compelling enough to be worthy of being claimed to prove the general case.)
 
The fact that something hasn't been proven to be true in no way proves that it is UNTRUE - that's just flawed logic.

 

[KeithEmo]

 
Will Kernel Streaming or ASIO also play files at their native sample rate, just like WASAPI, or what?

Why then, when I use WASAPI, does the display on my Fiio X3 (in DAC mode) still always say it's receiving input at the default Windows rate (16/44.1) rather than the rate of the file being played?

 
In general, ASIO should play files at their native sample rate, and Kernel Streaming will NOT do so.
(A specific ASIO driver may also include the option to resample "on the way through".)
 
If you're using WASAPI, then WINDOWS should not be resampling the files you're playing. I would also assume that the display on the DAC itself is probably correct. The most likely choice is that your player program is resampling the files. This can happen for several reasons:
 
1) Your player could be set to do so. Many players can be manually configured to resample certain sample rates (this is intended to allow them to play high sample rate files through DACs that don't support them). You should double check the output settings on your player.
 
2) When you connect a USB device to your computer, the two devices exchange information about what sample rates they both support. If this information exchange didn't complete properly, which can happen if you have driver or cable issues, then the computer may "think" that your DAC only supports 44k. If this has happened, then your player program may be automatically resampling what it plays to accommodate what it believes is a limitation of your DAC.
 
(It could be that the player you're using is set to resample - many offer the option. It's also possible that you have a driver issue going on. If the drivers aren't working correctly, or you're using a cable that's too long, it's quite possible that your computer "thinks" your DAC doesn't support the higher sample rates because the USB connection negotiations didn't finish properly. If that happens, then your player will "believe" that the DAC is incapable of supporting the higher sample rates, and may resample them to a lower sample rate before sending them out to enable them to work.)
 
3) Your ASIO driver might have an option to convert the sample rate - so you should also check that.
 
4) I hate to ask, but you ARE playing FLAC, WAV, ALAC, or some other sort of LOSSLESS files, right? (If you play lossy compressed files, like AAC or MP3 files, the numbers you see aren't actually sample rates - they're bit rates. And, when your player program converts them to PCM, they will be converted at the default sample rate set in Windows... at which point they will play through WASAPi at that sample rate. Likewise, if you're playing music through a streaming service, their player program may alter the sample rate for whatever reasons.

 

[cjl]

   
No - it is, in fact, technically correct. (Well, let's say that the fact that differences exist can be proven easily enough... and there do seem to be quite a few people, including myself, who do claim to hear differences - at least in some specific situations.)
 
Start with a 24/96 file (you pick it); re-sample it to 16/44 using any two audio programs (you pick them); and do a bit comparison between the results. They will NOT be bit-for-bit identical. Even beyond that, most of the higher end programs offer several different choices of which filter slope and type to use, so, even with the same program, you can produce an infinite number of close-but-not-identical results. Likewise, start with any 16/44 file you like, convert it to 24/96 using your favorite program, then convert it back to 16/44 using the same program. What you end up with won't be bit-for-bit the same as the original then either. (This is in direct contrast to converting a FLAC file to a WAV, then back to a FLAC, as many times as you like - and ending up with a file that IS identical to the original.)

Of course they won't be bit for bit identical - the file size alone will differ by a factor of 3 or so. Even after converting back, you won't have identical files, since you will have lost some data. Fortunately, I didn't claim they would be identical, just that they would be AUDIBLY identical. I stand by that assertion. Everything lost in the downconversion from 24/96 to 16/44.1 is inaudible.

 

[goodyfresh]

   
In general, ASIO should play files at their native sample rate, and Kernel Streaming will NOT do so.
(A specific ASIO driver may also include the option to resample "on the way through".)
 
If you're using WASAPI, then WINDOWS should not be resampling the files you're playing. I would also assume that the display on the DAC itself is probably correct. The most likely choice is that your player program is resampling the files. This can happen for several reasons:
 
1) Your player could be set to do so. Many players can be manually configured to resample certain sample rates (this is intended to allow them to play high sample rate files through DACs that don't support them). You should double check the output settings on your player.
 
2) When you connect a USB device to your computer, the two devices exchange information about what sample rates they both support. If this information exchange didn't complete properly, which can happen if you have driver or cable issues, then the computer may "think" that your DAC only supports 44k. If this has happened, then your player program may be automatically resampling what it plays to accommodate what it believes is a limitation of your DAC.
 
(It could be that the player you're using is set to resample - many offer the option. It's also possible that you have a driver issue going on. If the drivers aren't working correctly, or you're using a cable that's too long, it's quite possible that your computer "thinks" your DAC doesn't support the higher sample rates because the USB connection negotiations didn't finish properly. If that happens, then your player will "believe" that the DAC is incapable of supporting the higher sample rates, and may resample them to a lower sample rate before sending them out to enable them to work.)
 
3) Your ASIO driver might have an option to convert the sample rate - so you should also check that.
 
4) I hate to ask, but you ARE playing FLAC, WAV, ALAC, or some other sort of LOSSLESS files, right? (If you play lossy compressed files, like AAC or MP3 files, the numbers you see aren't actually sample rates - they're bit rates. And, when your player program converts them to PCM, they will be converted at the default sample rate set in Windows... at which point they will play through WASAPi at that sample rate. Likewise, if you're playing music through a streaming service, their player program may alter the sample rate for whatever reasons.

Thanks for the further info!  And to #4, come on man, I'm not some chump or anything, OF COURSE I only use FLAC or WAV files unless I for some reason simply CANNOT obtain lossless files for a particular album.  All but a very tiny fraction of my music library is, bare-minimum, at unmodified lossless CD-Quality.

Oh and good news.  I've now re-configured the implementation of WASAPI in PotPlayer and now the display of my DAC properly displays the sample-rate for the file format of whatever I am playing

  Of course they won't be bit for bit identical - the file size alone will differ by a factor of 3 or so. Even after converting back, you won't have identical files, since you will have lost some data. Fortunately, I didn't claim they would be identical, just that they would be AUDIBLY identical. I stand by that assertion. Everything lost in the downconversion from 24/96 to 16/44.1 is inaudible.

Oh man, I never meant to start an argument, but this is interesting   Haha

 

[nick_charles]

   
To be totally truthful, I've never seen ANY of what I would call "well conducted rigorous testing" on that subject at all. (I've seen plenty of tests that showed that, under certain circumstances, the majority of a certain group of subjects didn't notice a specific difference, and I've seen tests that proved that specific people who believed they were able to hear specific differences under specific conditions were in fact unable to do so, but I've never seen a comprehensive test anywhere near compelling enough to be worthy of being claimed to prove the general case.)
 
The fact that something hasn't been proven to be true in no way proves that it is UNTRUE - that's just flawed logic.

 
By and large the experimental approach starts out with what is called the null hypothesis, which basically posits that there is no difference between two conditions such as a control condition and an experimental condition or say redbook and upsampled (or whatever) . Then we run tests and analyze the data and we can either conclude that we find a statistically significant difference and "reject the null hypothesis" or that we "fail to reject the null hypothesis", of course this is all done with a specific set of conditions. 
 
Of course one study is not enough but if we run sufficient variations of the tests with big enough samples (bigger is better) and they all end up with us failing to reject the null hypothesis while we never say the case is proven we pragmatically eventually conclude that the preponderance of evidence supports the null hypothesis (until something contradicts it) at this point we take it as "generally accepted" and study something more interesting unless we receive contradictory evidence. we no longer need to drop apples from trees 
 
While we can argue the toss about studies like Meyer and Moran the more interesting question is given the resources that folks like Sony (SACD) , Meridian, Neil Young and so on have where are their controlled tests that conclusively support the audible benefits of XXXXXX vs YYYYYYY ?? The best you get is the same old dog and pony shows.

 

[goodyfresh]

   
By and large the experimental approach starts out with what is called the null hypothesis, which basically posits that there is no difference between two conditions such as a control condition and an experimental condition or say redbook and upsampled (or whatever) . Then we run tests and analyze the data and we can either conclude that we find a statistically significant difference and "reject the null hypothesis" or that we "fail to reject the null hypothesis", of course this is all done with a specific set of conditions. 
 
Of course one study is not enough but if we run sufficient variations of the tests with big enough samples (bigger is better) and they all end up with us failing to reject the null hypothesis while we never say the case is proven we pragmatically eventually conclude that the preponderance of evidence supports the null hypothesis (until something contradicts it) at this point we take it as "generally accepted" and study something more interesting unless we receive contradictory evidence. we no longer need to drop apples from trees 
 
While we can argue the toss about studies like Meyer and Moran the more interesting question is given the resources that folks like Sony (SACD) , Meridian, Neil Young and so on have where are their controlled tests that conclusively support the audible benefits of XXXXXX vs YYYYYYY ?? The best you get is the same old dog and pony shows.

That is one of the best concise explanations of the use of stastics in the scientific-method that I have ever seen

 

[arnyk]

 
No - it is, in fact, technically correct. (Well, let's say that the fact that differences exist can be proven easily enough... and there do seem to be quite a few people, including myself, who do claim to hear differences - at least in some specific situations.)
 
Start with a 24/96 file (you pick it); re-sample it to 16/44 using any two audio programs (you pick them); and do a bit comparison between the results. They will NOT be bit-for-bit identical. Even beyond that, most of the higher end programs offer several different choices of which filter slope and type to use, so, even with the same program, you can produce an infinite number of close-but-not-identical results. Likewise, start with any 16/44 file you like, convert it to 24/96 using your favorite program, then convert it back to 16/44 using the same program. What you end up with won't be bit-for-bit the same as the original then either. (This is in direct contrast to converting a FLAC file to a WAV, then back to a FLAC, as many times as you like - and ending up with a file that IS identical to the original.)
 

 
That's all true. However just because two audio files are different is well known to not be a sure indication that they sound different.
 
However in order to find this out, you usually have to do a rational listening tests, which most audiophile casual evaluations are not. I've listed the reasons why here dozens of times, even several times today, and nobody seems to be able to be rebut them with reliable evidence and rational logic. They don't even seem to try.  Be my guest!
 
 

Originally Posted by KeithEmo
 
To be totally truthful, I've never seen ANY of what I would call "well conducted rigorous testing" on that subject at all. (I've seen plenty of tests that showed that, under certain circumstances, the majority of a certain group of subjects didn't notice a specific difference, and I've seen tests that proved that specific people who believed they were able to hear specific differences under specific conditions were in fact unable to do so, but I've never seen a comprehensive test anywhere near compelling enough to be worthy of being claimed to prove the general case.)
 
The fact that something hasn't been proven to be true in no way proves that it is UNTRUE - that's just flawed logic.

 

You can easily do the test for yourself, and that should be highly convincing.
 
The steps are simple, easy to execute and have been done many times that the post above seems to have  dismissed because you were not personally involved:
 
(1) Take a high resolution music wave file of your choice.
 
(2) Down sample the high resolution files to 44/16 using one of the many tools that are designed for the purpose such as the highly regarded freeware software named Sox, in accordance with the best parameters that are recommended for the purpose.  This will damage the bit perfection of your original source file fatally in ways that can never be undone.
 
(3) Again using Sox, upsample the 44/16 file to the original high sample rate.  You now have a highly damaged low resolution file in a high resolution container.
 
(4) Ensure that your processing has not caused level changes in the 20-20 KHz audio band, or timing errors or other easily detectable errors that can be easily avoided.
 
(5) Compare the two files using one of the highly regarded freeware tools that are available for the purpose such as Foobar2000 with the ABX plug in.  This will be a DBT that is free of the usual errors that audiophile casual evaluations usually have.
 
(6) Report your results. Post your ABX logs.
 
If you have any questions about this process you can ask them and get them answered by qualified individuals on the Hydrogen Audio forum, or here.

 

[KeithEmo]

  Of course they won't be bit for bit identical - the file size alone will differ by a factor of 3 or so. Even after converting back, you won't have identical files, since you will have lost some data. Fortunately, I didn't claim they would be identical, just that they would be AUDIBLY identical. I stand by that assertion. Everything lost in the downconversion from 24/96 to 16/44.1 is inaudible.

 
Honestly, I would be willing to believe that the results of a truly "optimum" down-conversion process might be inaudible (either to everyone, or just to me); hwo

  You can easily do the test for yourself, and that should be highly convincing.
 
The steps are simple, easy to execute and have been done many times that the post above seems to have  dismissed because you were not personally involved:
 
(1) Take a high resolution music wave file of your choice.
 
(2) Down sample the high resolution files to 44/16 using one of the many tools that are designed for the purpose such as the highly regarded freeware software named Sox, in accordance with the best parameters that are recommended for the purpose.  This will damage the bit perfection of your original source file fatally in ways that can never be undone.
 
(3) Again using Sox, upsample the 44/16 file to the original high sample rate.  You now have a highly damaged low resolution file in a high resolution container.
 
(4) Ensure that your processing has not caused level changes in the 20-20 KHz audio band, or timing errors or other easily detectable errors that can be easily avoided.
 
(5) Compare the two files using one of the highly regarded freeware tools that are available for the purpose such as Foobar2000 with the ABX plug in.  This will be a DBT that is free of the usual errors that audiophile casual evaluations usually have.
 
(6) Report your results. Post your ABX logs.
 
If you have any questions about this process you can ask them and get them answered by qualified individuals on the Hydrogen Audio forum, or here.

 
That might actually be true - if we can find a true optimum conversion. However, conversions are never "optimum", if we can ever even figure out precisely which settings would be optimum. I once had the chance to do just such a comparison. It involved the (whole other) question about DSD. A DSD source file - claimed to have been recorded and mastered in DSD - was converted to 24/96 using both of the two "top" commercial conversion programs (Weiss Saracon and Korg Audiogate). Amazingly, the two resulting 24/96 PCM files sound slightly different. I wouldn't claim that I could identify one or the other in a standard ABX test, or even that one or the other is "better", but simply that they are not identical.... and that's a conversion from the same source to the same sample rate - the only difference being the algorithms used in the two programs (the "most neutral" option was claimed to have been chosen for each, at least Saracon itself offers multiple filter choices).
 
(Note that this is different than an ABX test. We're talking about switching directly back and forth, with perfectly matched levels, and with no wait, and hearing a slight difference. An ABX test expects the listener to be able to identify one source or the other; which is actually often more difficult than simply comparing them side by side and noting a slight difference. (If I were to hold up two very similarly colored tiles, one after the other, and ask you to identify which was which, in a standard ABX test protocol, you would be able to identify which was which with a certain minimum amount of variation between them. However, if I were to hold up the two tiles next to each other, or even overlapping, and ask you if they were "identical or different" - most people would be able to identify a much smaller difference.)
 
Now, we can go off into an endless discussion about whether the correct or optimum filter options were chosen, or even if one or the other program might introduce more or less coloration than the other, but the fact remains that the output from different programs does sound slightly different. Therefore, or especially, since we don't actually know which program, or which settings, a given company uses to do their conversions, we can't really know how similar the conversions produced by that company sound. So, if you really want to do an ABX test to see if the differences are audible, and make a definitive generalization that they are not, then WHICH programs shall we test, and WHICH filter settings shall we choose, and WHICH headphones and headphone amp, or WHICH speakers and speaker amp shall we use to conduct the test, and WHAT source material. I suggest that a "fair and complete sample" might include the top five conversion programs, the filter setting claimed by each to be "neutral" and two others, at least three each of planar, dynamic, and electrostatic headphones, and at least five different combinations of speakers and speaker amps. If none of a hundred test subjects can reliably identify a difference with ANY of those combinations, each using their own "favorite" test music, then I would be willing to accept a generalization that "the vast majority of people can't hear any difference". Other than that, even if I personally were unable to hear a difference, it wouldn't in any way prove that other people cannot.
 
Even further than that, even if it turns out that some "perfect optimum conversion" is totally inaudible, that still isn't enough, because it could turn out that a particular company deliberately chooses conversion settings that are NOT optimum. We could be sinister and suggest that some "evil company" might actually deliberately degrade the sound quality of their lower-resolution downloads so that their more expensive high-res ones sound better. Or we might wonder if they simply deliberately adjust each separately to suit the desires of their target market (perhaps they boost the treble a bit on their high-res downloads so they sound "clearer" than the 16/44 ones, and reduce the treble a tiny bit on the 16/44 ones "to suit customers who don't like that hi-res sound"; much as many early CDs were remastered with a slight treble boost to deliberately make them sound "better than records"). Since, being end users, we mostly only have access to the version we purchase, we need the answer to this question as well (so, if we want to know if particular versions are "audibly identical", we need to know what program, and what settings, each of our music sources uses).
 
When two of anything measure identical by every practical measurement, but certain people still claim to hear (or see) a difference, then it's fair to suggest that, if they prove unable to do so in an ABX test, then they're probably imagining it. However, when there is a measurable difference, and some people claim to be able to hear or see it, you have to exhaust a lot of different possible test scenarios before you can reasonably claim that "the difference cannot be detected - at all - ever". Feel free to claim that it's minor, or insignificant to you, or statistically insignificant, but none of those is the same as "not there".

 

[KeithEmo]

  Thanks for the further info!  And to #4, come on man, I'm not some chump or anything, OF COURSE I only use FLAC or WAV files unless I for some reason simply CANNOT obtain lossless files for a particular album.  All but a very tiny fraction of my music library is, bare-minimum, at unmodified lossless CD-Quality.

Oh and good news.  I've now re-configured the implementation of WASAPI in PotPlayer and now the display of my DAC properly displays the sample-rate for the file format of whatever I am playing
Oh man, I never meant to start an argument, but this is interesting   Haha

 
Excellent.
 
I'm not sure which is sillier, the fact that Windows STILL doesn't include built-in support for USB Audio Class 2 (which is why we still need to use separate drivers to get 24/192), or that, even though WASAPI mode has been around since Windows 7, the internal Windows player apps STILL don't use it by default. We would have so many less problems if good audio support was built into Windows itself.
 
ASIO is always a hot topic - especially with "musician types" - because it has historically been more popular, more available, and quite possibly more reliably with "pro" equipment. Most pro equipment offers ASIO, and a lot of it offers nothing else. Therefore, "pro types" are used to thinking of ASIO as "their go-to choice" as a sound driver. ASIO also often allows some pretty sophisticated configuration options - sometimes including the ability to swap or mix channels. However, WASAPI is simpler, and easier to configure in most cases. And, in "consumer equipment", like audiophile DACs, the ASIO support is less consistent - and less certain to exist at all. Of course, the arguments about whether it sounds better occur simply because it's another choice to compare

 

[arnyk]

 
Honestly, I would be willing to believe that the results of a truly "optimum" down-conversion process might be inaudible (either to everyone, or just to me); hwo
 
That might actually be true - if we can find a true optimum conversion.

 
The above appears to be a common form of a deflection of the reasonable procedure that was suggested. The word optimum has been abused to create an impossible task.
 
While the word "optimum" sounds professional and scientific, if one really understands the meaning of the word optimum, nothing in the real world is truly optimum, especially once you go outside of the abstract world of math.
 
So, by demanding an  "...truly optimum down-conversion process", what might be a simple chore has been cleverly and surreptitiously transformed into Mission Impossible.
 
All we really need is a conversion process that is itself sonically transparent. The procedure I described above can be used with slight changes to test that, as well.
 
So, it is incumbent on the deflector to make the procedure appear to be either completely invalid or impossibly difficult.

 

[arnyk]

   
Excellent.
 
I'm not sure which is sillier, the fact that Windows STILL doesn't include built-in support for USB Audio Class 2 (which is why we still need to use separate drivers to get 24/192), or that, even though WASAPI mode has been around since Windows 7, the internal Windows player apps STILL don't use it by default. We would have so many less problems if good audio support was built into Windows itself.
 
ASIO is always a hot topic - especially with "musician types" - because it has historically been more popular, more available, and quite possibly more reliably with "pro" equipment. Most pro equipment offers ASIO, and a lot of it offers nothing else. Therefore, "pro types" are used to thinking of ASIO as "their go-to choice" as a sound driver. ASIO also often allows some pretty sophisticated configuration options - sometimes including the ability to swap or mix channels. However, WASAPI is simpler, and easier to configure in most cases. And, in "consumer equipment", like audiophile DACs, the ASIO support is less consistent - and less certain to exist at all. Of course, the arguments about whether it sounds better occur simply because it's another choice to compare

 
Ignores the fact that the primary reason for the development of ASIO was to reduce latency in live recording/real time monitoring applications.
 
https://en.wikipedia.org/wiki/Audio_Stream_Input/Output
 
"Audio Stream Input/Output (ASIO) is a computer sound card driver protocol for digital audio specified by Steinberg, providing a low-latency and high fidelity interface between a software application and a computer's sound card. "
 
Since the latency (minimum time between recording and playback) during audiophile playback for listening enjoyment is already measured in days or years, sound card driver latency isn't much of a real world issue outside of its intended use which is live recording with real time monitoring of the recording.
 
In many cases the ASIO and  other driver scheme (example: WSASPI, MME, DirectSound, etc.) performance are indistinguishable for every other parameter than latency.

 

[castleofargh]

yeah when you don't have some lingering effects turned ON from your soundcard, my experience with windowzzz mixer isn't bad. but still it really is ludicrous how windows deals with usb sound.