Sampling rates for OS

I wish this would settle the subjective argument! That may be a little ambitious for a crowd of audiophiles :~)

dCS is really doing some hard work on this stuff.

Good reading, and more than enough for me!

Stick this particular audiophile in the 'settled' pile. :)

Wonder if it's as "simple" as going to the quicktime player and under preferences choosing the quicktime settings. Mine is set to 16/44 but it might be a different quicktime engine...dunno.

dCS is really doing some hard work on this stuff.

I bet they do, and when you put everything in a PDF it looks good, right ?

I'm not sure where to begin at pointing out the flaws. There are just too many.
This says February 2009, right ? was the author in lack of Google, or what ...

Maybe others are going to point out what is all wrong in there. I'll stick to this short post.
Call me lazy.

Peter

Peter - Please take this in the spirit in which it's intended. Are you ever satisfied with anything? Maybe instead of just knocking something you could suggest what the correct answer would be so people reading this could learn something. Plus, pointing out some good things in the paper may make your statements easier to swallow.

I've talked to the people at dCS so I stand by my statement that they are doing some hard work on this stuff.

is the computer model/manufacturer & sound card that dcs used for the tests and the instrument they used to record the data. While the differences in sampling rates are very clear on their effects, I suppose it doesn't matter what hardware they used, the effects are the same.
It would be nice to know what they used, purely as a reference and a simple diagram of the test setup. I'm grateful and highly appreciative that dcs have written the test, seeing that many others can just shoot from the hip with findings that exist in their imagination.

Are you ever satisfied with anything?

Of course I am. But if the "errors" are major, nothing can be made of it. So, for example, if the author doesn't recognize the difference between bit perfect because of "bit perfect" (does anyone actually knows what this means -> you could almost grab a dictionary for it) and "not sample rate converted", the whole document is invalidated. The author takes the latter only, and from that point on there's noweher to start.

Can't you agree ?

Hi, my first post...
I could see that the owner of the pdf file is Andrew McHarg and I found an Stereophile's article saying from him: "Andy McHarg joined dCS as a software engineer in 1993. Most of the fabled dCS gear that you read about has his name on the design team. Now Director of Development at dCS, his focus is on signal processing. Andy is the man behind the Scarlatti and Paganini Puccini, and the co-inventor of the Scarlatti Upsampler and Puccini U-Clock. You might say that he knows a few things. " I totally agree with the last sentence...
Best regards from SouthAmerica, Argentina

Hi Maluli - Welcome to Computer Audiophile and thanks for adding that information.

... but a bit flawed. I´m very sure that the author is very capable - at least he comes from a famous company known for engineering perfect consumer and professional stuff. However, this doesn´t hinder him to make conclusions which I find a bit too easy.

First of all, he uses ASIO4ALL. The title of this little addition to the sound interfaces driver suggests that it provides ASIO. This is not true. It simply is a wrapper that enables the audio software to use Kernel Streaming with the interface. As you may know, Kernel Streaming only applies to Windows XP - at least officially. This Kernel Streaming has the unique ability to stream the audio material through the Audio Kernel of WinXP without volume changes or resampling.

The flaw of the article lies within the fact that the autor also used it on Windows Vista which does have a completely different Audio Kernel that most importantly lacks the Kernel mixer of WinXP. However, most interface WDM-drivers still can use it because they are simply ported from the XP driver to the Vista platform. This is very coincidental and in my mind one can´t draw any objective results from it.

Secondly, he claims not to know software that can use WASAPI direct access. I can give at least one example: foobar2000. The author even uses foobar2000 but claims he doesn´t know that there is a WASAPI output plug-in for it. I find this very strange. Can someone not know about this? If the author doesn´t know this, then I´ll say no more. But if he does know it then I´d be saying that he follows some agenda not visible to us when he´s claiming that there is no WASAPI capable software.

Other than that, it is still a very fascinating article :)

Cavaille, you are too late I'm afraid. The article by now is referred to from within various threads here at CA, and it has become the new standard for digital.

Never mind the serious flaws you mentioned, and never mind the other 8 or so at the same level, this is just it.
Now start behaving like it ! :-)

But I tell you the author knows nothing about it all (why suggesting second agendas ?)

Peter

You know what ? maybe I can turn this into a somewhat more constructive essay.

Not that I am going to take this in the spirit it is intended (sorry Chris, it's just all misinformation which is going to lead its own life) ...

So what I will do is mention such a flaw each new day. But this is not the constructive part ...

The constructive part can emerge from others being ahead of me. Not by guessing but by knowing. Looking better at the texts ... knowing the real merits of things. Try to reason it out if you like.
It might prevent stupidly copying others' words a next time.

Btw, this is *not* against the OP (One and a half) at all. Why shouldn't everybody come up with stuff which looks good, without knowing it all yourself ? What bothers me a bit though, is that this forum by now is full of (ever so nice) people who will be using this kind of information to back up their own misknowledges.

See you tomorrow (if there's anything left of course ! :-))
Peter

PS: If I'm shouting "the author knows nothing about it" I can back that up. Just wait until this thread is finished ...

how come one player could sound differently with the same sound file on a hard disk. (source : STEVE NUGENT)

dCS has clearly fed a 20kHz sine wave into players on different OSs and (emph) measured the result. The whole point of the paper is to achieve 'low' distortion, you need to set the OS output sample rate to match the source file sample rate. By showing us the reference, and then the effects of what not to do, the results are abundantly clear. Using ASIO, or a player that would use WASAPI makes little contribution to fix up the mess caused by setting the incorrect sampling rate in the first place.
Wavelength's recommendations are to set the OS sample rate of the device that you're connecting to with tweaks for say Redbook audio to 88.1, 24bit (for the MAC, or if you use iTunes for Windows). This information totally contradicts the dCS paper's findings. While I have the greatest respect for Gordon Rankin and also the good people at dCS, this latest finding leaves me even more bewildered.

Maybe this information is DAC dependant?

I'm sick of Peter's know it all attitude to put it politely, nor do I care. Does zero for the discussions. People here are finding the best way to play music from their computer. It's new for all of us, with a lot more knowledge to find it's like pioneering to find knowledge at our own pace. Getting to the nitty gritty of computer audio is not easy to understand at first glance. It's a discovery and adventure, so wet blankets, please, be patient, don't spoil the journey.

I'm sick of Peter's know it all attitude to put it politely, nor do I care.

That's cool. But it leaves you with this :

This information totally contradicts the dCS paper's findings. While I have the greatest respect for Gordon Rankin and also the good people at dCS, this latest finding leaves me even more bewildered.

Does zero for the discussions.

I am sorry to hear that. In that case I apologize for not blaming you to post the article in the first place. Apparantly now *you* want to be right on it, well, please do.

Why would I care anyway.

Chris, can you please delete my before post ?
I'll take it in the spirit it's intended now. God knows what that spirit is, but it must be about leaving people in great doubt but with even greater discussions.

Thanks,
Peter

Never mind the serious flaws you mentioned, and never mind the other 8 or so at the same level, this is just it.
Now start behaving like it ! :-)

The author doesn´t know anything about WASAPI? How can he make assumptions from his measured "data" then? More flaws?? Please, point them out. Apparently I overread them. ;-)

I'm sick of Peter's know it all attitude to put it politely, nor do I care. Does zero for the discussions. People here are finding the best way to play music from their computer. It's new for all of us, with a lot more knowledge to find it's like pioneering to find knowledge at our own pace. Getting to the nitty gritty of computer audio is not easy to understand at first glance. It's a discovery and adventure, so wet blankets, please, be patient, don't spoil the journey.

Peter did the right thing. He expressed concerns about this paper. We need "Cassandras" like him. You see, if I claim something like the recent things I´ve posted here (different drives are sounding different etc.) then I expect to be questioned about it. I expect some very healthy doubt about the things I´m claiming.

Because: If people are doubting me that helps myself. Maybe I´ve overseen something important? That´s the beauty of sites like these: through discussion one can reach a better understanding. I won´t write that I support Peter wholeheartedly - he advertises his XXPlayer as the solution for all ill sounds on the computer. Frankly I don´t believe that. On the other hand I admire his consistency towards his project. And I´m sure that he also expects doubts about his findings when he decides to go public with them. Why not?

This is the way things are. And it is good that way. Everything isn´t just black and white, it´s all greyscales. When going through those greyscales one achieves a larger, better informed picture. Therefore I find it very disturbing and disappointing that some engineer who should have much experience and a large scientific background releases such an ill-informed and badly executed paper. And in order to reach a bigger picture I question the motives of this engineer.

The only flaw in the article with respect to WASAPI and ASIO is the omission that FooBar can support WASAPI through a suitable plugin (not "out the box" as I understand it). As for ASIO4ALL - I couldn't see any reference to using this on Vista - though the producer does advertise it as working on Vista with WDM drivers. I don't think it was listed under workarounds.

The condradiction between dCS document and Gordon's comments about using 88.2 or 96k out of a Mac might not be so contradictory. You must remember that dCS's own USB interface is part of a high end upsampler which is designed to upsample to DSD format. They state in the document that the MacOS upsampling is much better than under windows however would be not as good as in the Scarletti Upsampler. However Gordon's DACs don't have such an upsampler so maybe they do work better with the MacOS upsampling. Please note this is not anything I know - just a possible explanation to explain the contradiction.

Peter's comments were meaningless. The written equivilent of sticking tounge out because you don't like what someone said. If he has 8 points where he feels the document is misleading or wrong. Let him say them rather than playing a "I'm cleverer than you because I've spotted flaws can you spot them too" game. Yes the document has limits, it's not meant to answer everything. There was an omission about WASAPI and Foobar, but that doesn't make the findings invalid. It still showed (pending Peter's revalations) that you can avoid SRC with Windows Media Player and iTunes. They never claimed it proove all sounds the same or anything more.

Eloise

Peter's comments were meaningless. The written equivilent of sticking tounge out because you don't like what someone said. If he has 8 points where he feels the document is misleading or wrong. Let him say them rather than playing a "I'm cleverer than you because I've spotted flaws can you spot them too" game. Yes the document has limits, it's not meant to answer everything. There was an omission about WASAPI and Foobar, but that doesn't make the findings invalid. It still showed (pending Peter's revalations) that you can avoid SRC with Windows Media Player and iTunes. They never claimed it proove all sounds the same or anything more.

My dear Eloise, I tend to side with Peter. And I would like to explain it a bit. I guess (Peter may clarify that) that he just was tired to read this article. You know for someone like me who deals with audio on the computer since 1995 as a hobby all the things written in the article aren´t exactly new. The audio engine of Windows basically has been the same since Windows 98, with only a few additions. If my memory isn´t faulty, mixing by the Kernel was introduced in 1998 when WDM drivers were introduced.

To be honest I do not know Windows Vista. But I know Windows 7 and I doubt that Microsoft changed that much regarding the audio engine. From my observations (by ear & measurment) Windows 7 does a pretty good job when it comes to audio processing. It´s not very good but decent.

Maybe my assumption is wrong but I think that one has to know about the shortcomings of Windows Audio Engines. I don´t care much for Mac, I´ve never owned one so I can´t say anything about this. But the article is in so far ridiculous as members of Hydrogenaudio.org years ago wrote about the crappy resampler in Windows. I find it very disturbing that so many people don´t know about this. I then have to ask myself if some people here actually are looking elsewhere for information and if the things they are hearing are "real" and not just some distortion found to be pleasing to the subjective ear. Don´t get me wrong, I´m no fan of Hydrogenaudio.org either. Their attitude is sometimes extremely tiring. But at least they do one thing right: they know how to do measurments. And the guy in this paper is far from doing good measurments.

He presents some little and lovely charts - and hooray! (forgive me) the community is in awe. So my conclusion is that people also need to look at other sites for more information, not only here. If some would have been looking over the horizon, the "revelation" of this article could have been prevented. I´m not saying that I´m better than the next man - but for sure I´m better informed.

And only because I like to see over the horizon, I sometimes like to sit between chairs. So in my opinion Peter´s additions weren´t pointless. No, this paper was. It didn´t tell me anything I didn´t know beforehand.

My reading of the document was that it was saying the same as (I read) you saying here: that Windows XP (and Vista) have rubbish SRC which need to be avoided. Though I suspect that you doubt that can be avoided so simply?

As for Peter's comments: my feeling was he (so far) hasn't added anything to the discussion of the document. He said it was wrong but refused to say in what ways is wrong or offer any opposite evidence. I don't read everything published, but so far have seen 2 well presered arguments that iTunes can be bit perfect (this and Kent Poon's) and only Peter's evidence that bit perfect player (his own) output differently.

Bit perfect may not equate to the same sound. But I've only heard objective and anetdotal evidence for this which is counter to my (albeit limited) listening observations. [Edit/addition] That's not to say differences don't exist - just my own experience leads me to doubt it. I'm not following this blindly - but nothing to make me doubt it's validity has been "published" - at least not here.

Eloise

My reading of the document was that it was saying the same as (I read) you saying here: that Windows XP (and Vista) have rubbish SRC which need to be avoided. Though I suspect that you doubt that can be avoided so simply?

My observations are somewhat different: WinXP has a very crappy resampling engine, Vista has a better one. Still, both are far away from such SRC as the one from iZotope. That doesn´t necessarily makes them worse. The fine thing with SRC in those mixers for sure is their convenience. They work without most people noticing them.

For me that is exactly the biggest problem. If one like me wants the perfect sound quality a source can offer (may it be FLAC, WAV or even MP3) they have to be avoided at all costs. Some may observe that with these internal Windows mixers and SRC the sound is "better" because of introduced distortions or rounding errors - but this impression is only a colourization that suits some music.

So one has to use players which can avoid those mixers. You can do it with Kernel Streaming, WASAPI and ASIO while the latter one is the only one who "talks" directly to the hardware. Software being able to use at least one of those three is foobar2000, Winamp and MediaMonkey.

And let me tell you, I truly believe that there are differences between bit-perfect. The reasons for this may be many. According to the opinion of science, technicians and measurments there shouldn´t be any at all. I wonder how Peters software changes the sound while being bit-perfect. He won´t say it of course, that would take away a marketable advantage.

I had written ... "My reading of the document was that it was saying the same as (I read) you saying here: that Windows XP (and Vista) have rubbish SRC which need to be avoided." And you wrote in reply ...
"My observations are somewhat different: WinXP has a very crappy resampling engine, Vista has a better one. Still, both are far away from such SRC as the one from iZotope. That doesn´t necessarily makes them worse. The fine thing with SRC in those mixers for sure is their convenience. They work without most people noticing them.

For me that is exactly the biggest problem. If one like me wants the perfect sound quality a source can offer (may it be FLAC, WAV or even MP3) they have to be avoided at all costs. Some may observe that with these internal Windows mixers and SRC the sound is "better" because of introduced distortions or rounding errors - but this impression is only a colourization that suits some music."
Surely we are saying the same thing here ... or have I missed something about the WinXP and Vista resampling engines?

You then went on to say... "So one has to use players which can avoid those mixers. You can do it with Kernel Streaming, WASAPI and ASIO while the latter one is the only one who "talks" directly to the hardware. Software being able to use at least one of those three is foobar2000, Winamp and MediaMonkey."
I think this is where your opinion differs from the dCS document. dCS assert that if the sample rate that is incoming to the KMixer (or Vista's mixer) is the same as the outgoing signal, then KMixer works in a bit-perfect way. dCS are not the only people to hold with this view and express it - from memory I've read the same from Benchmark and also read somewhere it being stated that Microsoft say this is the case.

Finally you say... "And let me tell you, I truly believe that there are differences between bit-perfect. The reasons for this may be many. According to the opinion of science, technicians and measurments there shouldn´t be any at all. I wonder how Peters software changes the sound while being bit-perfect. He won´t say it of course, that would take away a marketable advantage."
My personal opinion / experience is that I've not found a difference between players when bit-perfect conditions are met. However I acknowledge and respect your counter opinion. Maybe my system is either able to read the bits in a way to counter the effects of software, maybe it's not good enough to notice them. Maybe it's just me. Anyway I'm still open to such differences being there in the future. Having said all that ... the dCS document doesn't claim all "bit-perfect" (whatever you interpret their use of that term to mean) sounds the same - all they are saying is that using Windows and Mac, standard software (i.e. iTunes and Media Player) can output a signal in a way that avoids the operating system's flawed SRC.

As I say ... if you disagree with the findings, then you need to present a real argument ... not just a "it's wrong" statement.
Eloise

I´m sorry. I did not want to use the usual clashing of arguments.

I indeed try to avoid Kernel Mixing or else like the devil tries to avoid Holy Water. I think I´m hearing a difference, in my opinion ASIO always sounds best. I can´t prove it of course and I can´t provide measurments. I´d love to but my means are insufficient for that.

And I didn´t say that the dCS document is just "wrong" - it was just done in a way where mistakes could sneek in. Furthermore, if I consider the authors background then I conclude that I would have done it more painstakingly if I had the means like him, especially if I´d be going public with it. There are some loopholes I talked about - and only these loopholes are the reason for my critique.

My dear Eloise, I deeply regret that I wrote some of the sentences I wrote - because they just were not true. I wrote for example that this article wasn´t needed at all because I thought it to be common knowledge. Obviously I was very mistaken there - because I just stumbled upon an article on hydrogenaudio.org which discusses this very dCS paper. Some of them were quite surprised of the findings and some of them didn´t even know it. I apologize for my misinformation and my belief.

The people over at hydrogenaudio.org were extremely disturbed by the test results for Windows Vista. They ran a few tests themselves and couldn´t confirm the results mentionend in the dCS paper for Vista. In their tests, Vista performed much better at resampling. They found another flaw which I didn´t mention (I did not see it): the article doesn´t describe the hardware that was used.

The Vista testresult could very well caused by a hardware resampler on the sound interface - and not by Vista.

Read it all here: http://www.hydrogenaudio.org/forums/index.php?s=2a0ed13ef665428678a47711...