XXHighEnd

[BobH]

But how different playback software sound different if delivering "bit perfect" data? The movement of data inside a computer is handled by the OS. The application accesses these facilities through and API to the device drivers. Error checking or lack of cannot be controlled by the application.

 

And that is exactly the point of this discussion. How do you know that bit perfection is being achieved? Why do people recommend certain types of drivers? How did they find out that one driver is 'better' than another and why should that be?

 

The light on the front of a good dac tells you what bit depth and sample rate is being received. It doesn't tell you that the file that left the hard drive is exactly the same, in every way, to the one being received. One way for us to try and find out is to discuss the issues, listen for differences, try stuff out, tell each other about it and see what comes out the other end.

 

 

 

[PeterSt]

No, WASAPI plays what you feed into it. You might mangle the data to your hearts content, implement your own SRC, etc. before you call WASAPI

 

Hi Roseval,

 

True of course. But what is your point here ? I mean, in the context of what I pointed out.

Btw, this can be done always, with or without mixer, with or without other infulences.

 

The point is, when you do not mangle, it is assured the output by the OS is not mangled. Well, that was my point anyway.

 

 

[PeterSt]

How do you know that bit perfection is being achieved?

 

Bob, by looping back the stream from the SPDIF output (USB is more difficult), record it, and compare it with the source.

 

You need to find a common denominating startpoint of the both files though, which needs some experience. This is needed because recording won't start at the exact beginning point of the original file.

If you'd use WaveLab, the end point is not critical because WaveLab will truncate the longest. From there on it checks bit by bit.

 

Note : The good old DTS trick does not apply for Vista, because when an encoded stream is fed to Vista, it will go into Exclusive Mode just because of that (otherwise the encoded data will be destroyed), and it says nothing about your player. Looping back the stream is the only way (for Vista).

 

[PeterSt]

Application A and Application B will use the same device driver to write the file to output buffer

 

and

 

If this is the case and no other intervening activity like SRC is done by A and B, you might indeed wonder if there is any difference in sound quality other than in the ear of the beholder.

 

Not to be nitpicking, but this is just not true down to the merits;

If A is 24/48 and B is 16/44.1 both will play at some common denominator, and at least one of them will be resampled, but probably both. This is KMixer (XP) or Sound Engine (Vista) !

 

This, in the context of my Vista/WASAPI-Exclusive remark.

But as said, nitpicking perhaps. Unless you did not know. :-)

 

 

 

[audiozorro]

There’s a lot in life that is not measurable or provable. There are many things we are only beginning to understand. The more we know, the more there is for us to learn. I am not of the school that believes that if you can’t measure it or explain it that it does not exist. For me many things exist that I can’t measure or prove. While I may understand many things and may be able to measure some things I realize that there I many more things for which I have no answers. The quest for knowledge is a never ending search and the totality of absolute knowledge is beyond human abilities.

 

How does one measure something when one does not understand what they’re measuring? How does one prove something for which they have little understanding? I guess the answer may simply be to believe and say that it does not exist.

 

There will always be naysayers. I for one am very glad that we have individuals that have taken the time and interest to improving the so-called perfect sound forever. My experience, maybe not proof, has been just when some may say "this is the best that is ever possible", someone comes along and makes it better.

 

Now on a personal note, I tend to be more tolerant of some manufacturers’ opinions, that is, if they sell what I consider to be good audiophile products for all of us to benefit. I realize they are in business, this is their livelihood and I would hope that they fully and honestly believe in their products. At times they are fanatical, at times their readings are enjoyable and they are very often characters. I believe Ashley had stated in another post some bad experiences with some other audiophile manufacturers, which may be enough to either taint one’s feelings or place them against the perpetrators for life. His tale of bad experiences gives me some perspective to his opinions.

 

I also realize that we can often get very emotional and opinionated about this hobby we love. Respect and civility has been the norm here and I trust that when apologies are necessary or in good taste, the offender will be a class act and do so.

 

Now on to XXHighEnd.

 

 

[BobH]

From your knowledge of how all this stuff works, could 'loop back' recording be done from within the same PC, say using 'Virtual Audio Cable', and still get a reliable result?

 

i.e Media player --> VAC --- > Wavelab

 

 

 

 

[glt]

...I thought that was the assumption in this thread: how can applications claiming to be bit perfect sound different. The underlying assumption is that everything else remains the same: the music file, the motherboard, the sound card, the drivers, the OS, the DAC, the Power Supply, etc.

 

[tfarney]

First of all, zorro --

 

"How does one measure something when one does not understand what they’re measuring? How does one prove something for which they have little understanding?"

 

Well, that's simple enough: Go get people who DO understand. We're not talking about religion, philosophy, or even quantum physics. This is not cosmic, or even theoretical. What we're talking about are systems that are wholly conceived, designed and executed by men (and women). They operate within a set of rules and principles invented by men, and there are plenty of men who understand EXACTLY what they're measuring. There may even be one or two of them here.

 

I'm not among those who understand, but I'm learning...back to it:

 

Dear Packet People -- If I got that packet stuff right, the data is moving around in bunches (packets) from CD to RAM to hard drive to RAM to DAC...and if there are any packets missing, or in the wrong place, the error will be obvious - a drop out or a shut-down, not some analog-like compromise of audio quality...yes?

 

So where, in the continuum, do the packets become a stream? Where is it all de-bunched?

 

I'm listening to John Coltrane right now. I know exactly when he's blowing packets or blowing streams. I'm less sure about my MacIntosh.

 

Tim

 

[The Computer Audiophile]

 

Two comments of the day and both from the same person!

 

 

"Dear Packet People..."

 

and

 

"I'm listening to John Coltrane right now. I know exactly when he's blowing packets or blowing streams. I'm less sure about my MacIntosh."

 

[tfarney]

Thank you...thank you...I'll be here all week. With a few moments off here and there to make a living.

 

Tim

 

 

 

[ldolse]

I work in an industry where I'm dealing with IP packets all the time, so I can take a crack at Tim's question.

 

It's not really about streaming or packets, it's simply about whether error correction is built into the transfer protocol. In some packet based protocols - like TCP/IP - there is a specific packet order, and both sides are aware of how many packets there should be. So they can always be put back in order, and if one is lost or corrupted it can be asked for again.

 

Other forms of data transfers often look much more like streaming than packets, such as file transfers from a hard disk or CDROM. Generally these streams are segmented in some way and there is some form of error correction, often CRC. If some part of the stream is corrupted it is re-requested until the error correction algorithm agrees that it's ok. Note that in both the above examples it's a two way communication, the receiver can make a re-request to the source.

 

When it comes to the digital connection between a digital source and a DAC, there are a number of industry standards, and I'll freely admit at this point that I have limited expertise on how these standards work. I believe most of these are purely focused on sending 'words' (i.e. samples) in real time. There is no ability to ask for a replacement word if one is bad, since these are real time, unidirectional protocols - Source->DAC.

 

Here's what's actually in an SPDIF word, referenced from http://www.epanorama.net/documents/audio/spdif.html.

 

Every sample is transmitted as a 32-bit word (subframe). These bits are used as follows:

bits meaning

----------------------------------------------------------

0-3 Preamble (see reference doc; special structure)

4-7 Auxillary-audio-databits

8-27 Sample (A 24-bit sample can be used (using bits 4-27). A CD-player uses only 16 bits, so only bits 13 (LSB) to 27 (MSB) are used. Bits 4-12 are set to 0).

28 Validity (When this bit is set, the sample should not be used by the receiver. A CD-player uses the 'error-flag' to set this bit).

29 Subcode-data

30 Channel-status-information

31 Parity (bit 0-3 are not included)

----------------------------------------------------------

 

It's interesting that there is a parity bit, this provides some level of error detection, but in the case of single bit parity, two wrongs would often make it right again, so I wouldn't call that great detection. Even if the parity bit is often effective, this only detecting the error, not correcting it.

 

Anyway, for 44Khz stereo audio that means two 32bit words need to be transmitted 44,000 times per second. These bits are being spooled out onto the wire one by one as an analog signal. That's 2,816,000 bits per second. The DAC then needs to convert this signal from an analog representation of 1's and 0's to a digital representation in it's buffer. According to the guys making fancy digital cables you'll find the dac is dealing with all sorts of things like reflections, noise, etc, while it's attempting to figure out what's really a 0 and what's really a 1. Then there are the discussions about how difficult it is to represent 0 and 1 in an analog signal in the first place.

 

I'll have to admit I'm more in the subjective camp, but I do try to use as much objectivity as I can apply. Based on the above, I think that the only thing that can really apply to changing 'bit perfect' sound is a) Jitter, and b) errors. I actually don't believe we're dealing with errors all that often (that's a gut 'subjective' guess). I have a hunch that anyone who has there rig properly set up would be able follow Peter's technique and find the source and output match.

 

That leaves only jitter. I for one am a believer that humans can probably hear much lower levels of jitter than what the industry accepts. When you think about what's really happening on that wire it's not that difficult to accept that there are a lot of things contributing to jitter. The problem with ABX'ing jitter is you would need some incredibly accurate equipment that could actually introduce varying amounts of jitter and leave everything else the same. It sounds like What PeterSt is trying to do with XXHighend might actually make that possible, but you'd still need the ridiculously expensive equipment he mentions to monitor said jitter.

 

[The Computer Audiophile]

A software salesman told me once, "Where there's mystery there's margin."

 

[audiozorro]

Maybe it because I lived through the total harmonic distortion specification wars of the 1960s and 1970s. Perhaps it’s because I remember many of the claims from the measurement experts in audio when they said all solid state amps sound the same, all cables sound the same, and power cords could not possibly make an audio difference.

 

Perhaps I have grown tired of all the claims that 16/44.1 digital is good enough and exceeds anything human hearing can discern. Maybe I have never accepted that one can determine the musical fidelity of audio equipment solely on the basis of measurements and not by listening.

 

It may have been SteveN who gave a pretty good description of the complexities and many forms of jitter in a previous thread. I don’t think he even discussed the interactions and spectrum of the distortion harmonics of complex jitter. Maybe it’s because it is usually in the manufacturer’s best interest to get everyone to agree to a problem that can be measured with scientific instruments so that the simple solution is just to produce a product that has superior specifications.

 

Maybe it’s because I believe in bit perfect data but I not sure I agree that we understand everything we need to know about bit perfect music. I understand analog waveforms and digital waveforms, the differences between 24/192 and DSD. I also confess that I may often strive for high levels of musical satisfaction and excitement, even if its due to colorations or distortions that make the recorded playback to be more like or even better than the real performance.

 

Yes, I often rely on my God given ears. I wish I had the golden ears of several musicians that I know or the superior hearing abilities of the many blind people I have read about. I know I don’t like my music to sound harsh or be tiring, even if one wishes to claim this is a more accurate representation of the recorded music.

 

I know I like soundstage, depth, tonal balance, harmonics, dynamics, midrange purity, air and separation, naturalness, rhythm and pace, and imaging, even though I know some of these cannot be measured by any scientific instruments to date. I am also fully aware that some of these attributes can be measured by scientific instruments but I most certainly know that all can be heard with human ears.

 

I think it is great to have the likes of technical experts, like Keith Johnson, Kent Poon, SteveN or PeterSt, to find explanations and where possible, scientific measurements, for the things many of us hear. All of these individuals have dedicated consider effort in bring us the best digital music possible. What I find tiring are those individuals that would prefer to spend their efforts to convince people that they couldn’t possibly hear what they hear and if it can’t be measured with scientific instruments it doesn’t exist.

 

I freely acknowledge the large number of audiophiles that prefer the sound of analog tape and vinyl. I also accept those audiophiles that feel that SACDs and high rez PCM are sonically super to Redbook CDs or lossy digital. Why many of these folks who profess a desire for high fidelity prefer the sound of these formats can be discussed endlessly, supplanted with all kinds of measurements and theories. But that many of these folks who profess a desire for high fidelity prefer the sound of these formats is a fact.

 

 

[glt]

re: Anyway, for 44Khz stereo audio that means two 32bit words need to be transmitted 44,000 times per second. These bits are being spooled out onto the wire one by one as an analog signal. That's 2,816,000 bits per second. The DAC then needs to convert this signal from an analog representation of 1's and 0's to a digital representation in it's buffer.

 

You mean "... these bits are being spooled out onto the wire one by one as a DIGITAL signal. ..." "...The DAC then needs to convert this signal from a DIGITAL representation of 1's and 0's to an ANALOG representation in it's buffer..." (which means the DAC needs to convert the digital representation of the analog signal to the actual analog signal)

 

[ldolse]

In the end it's just a signal, and a signal is a waveform - in the Computer's memory and the DAC's buffer those are ones and zeros - on the various cables they flow through they are an analog(i.e. waveform) representations of 1's and 0's.

 

[ldolse]

Actually that website I referenced before has a good ascii representation of what the the analog representation of a digital signal looks like. I can't paste it in here as the markup required to display it isn't allowed:

 

http://www.epanorama.net/documents/audio/spdif.html

 

It's about a quarter way down. As you can see, it's just a square wave, and that section describes how ones and zeros are represented by the modulations of that wave.

 

 

 

 

[glt]

It is a waveform alright and the wave form can only have high, low and transition values.

 

[PeterSt]

After a good night of sleep I look back here, and find a bunch of new posts which are all of them so amazingly good ...

 

How does one measure something when one does not understand what they’re measuring? How does one prove something for which they have little understanding?

 

Without degrading each and every other sentence written, this would be my favorite. And in between the lines - sometimes more explicitly - this is exactly what is in my book;

 

I am ignorant enough to think to know (hehe) that by know we base ourselves on "techniques", science and perceived knowledge, which is just a tad too old for modern days. The measuring standard for DACs of which I think is wrong is a clear example of it, as long as I am correct on that one myself of course.

 

Indeed the very next thing is about understanding what we're actually dealing with. And indeed again, how to measure something if we don't know what the "something" is.

 

I think it was Tim who said go to those who know but it is not as simple as that. The "those who know" could be those who perceive the actual matter, and are not the engineers of a DAC. All is bits and pieces, and nobody can know everything. The all result however, is what we listen to, so we "know" most as long as we are open to it. The problem of course is, that "we" usually are not those engineers and it is easy to see that if the one asks a relevant (smart) question, the other doesn't have the real answer, but it encourages for better research.

 

The vagueness of the above will not be vague anymore when I say I'm referring to myself and not having the proper answers to the jitter question/suggestion from Roseval. With both feet back on the ground I realized that my own thinking on this matter has been too simple perhaps, but having to write about it sets my mind.

 

Perhaps skip below alinea, because it may be too difficult for me to put in understandable english.

 

Personally I think it is great from Idolse to spend so much time on presenting an outlay on packet transfer, by accident growing into an SPDIF outlay which for sure helps at understanding, might it be about error correction or jitter. It is these posts which may encourage for "aren't you wasting your time here ?" followed by a remark "better go to those who know in the first place" (but note Tim's suggestion was earlier in the thread, and I'm not talking in that context now). No, I think this is just very good and besides it gives insight for those who want to know, it may stimulate the brain of others to the sense of "hey, wait a minute, that's in the equation as well !".

 

More to the matters :

 

The problem with ABX'ing jitter is you would need some incredibly accurate equipment that could actually introduce varying amounts of jitter and leave everything else the same. It sounds like What PeterSt is trying to do with XXHighend might actually make that possible

 

And this is so true. In between the lines I referred to this already I think : we may have the unique situation that the one DAC is not static for jitter (specs ?!?!) and the amount can be controlled.

I say "may have", only because it is not 100% clear to me whether it is jitter only that is under influence. I have a fair idea about other subjects, but they too are disproved by my own DAC so far.

 

The more we know, the more there is for us to learn. I am not of the school that believes that if you can’t measure it or explain it that it does not exist. For me many things exist that I can’t measure or prove. While I may understand many things and may be able to measure some things I realize that there I many more things for which I have no answers. The quest for knowledge is a never ending search and the totality of absolute knowledge is beyond human abilities.

 

Sorry for nagging, but I think this is another of such great sayings I discovered this morning.

 

The point (or my point anyway) here is that the perception from an audio chain as a whole indeed can never be measured, but, the elements contributing to it can. That is, I am fairly sure of that. The real matter is : recognizing the elements.

It is exactly this what I am stribing for, and once some major elements can be recognized and next explicitly measured, it comes down to the all so important quest of interpreting.

 

One example which I mentioned somewhere before :

 

Looking at harmonic distortions of an NOS DAC (especially the filterless) will make every signal processing engineer say "no good !". I too would say that, seeing what such a measurement shows. Now :

While all starts with the super complex reasoning about transients (squarish) which by itself is the best, that just creating false harmonics on the Nyquist mirror, it just *is* so that it sounds the best. And I mean, not subjective to likings, but just to the merits of natural sounding instruments and all. So what is going on ?

 

What *will* be going on, is that the super-bad figures must be masked by something we don't know, and that whatever it is, is far more important to natural sound. Part of it, if not all, lies in just those transients, and the thing to do is proving by measurement that the bad figures (the mirrored false harmonics) cannot be heard because they, for example, are 20 dB down.

That any self respected engineer will say "yeah, but they are 50dB *above* the noise floor" (which just *is* how engineers present their figures) may be ever so wrong because the amplitude of this falseness rides along with the good stuff which is always 20dB higher. Of course it will be better if this distortion is not there at all, but if it can't be avoided with the benefit of keeping the dynamics, it is just the better solution and proven by measurement.

Please keep in mind, this is just an example of how interpretation could go, and the 20dB figure was just made up by me.

 

An example in a complete different direction is about measuring itself, and the impossibility to measure "music" :

When a device unter test undergoes a sweep (which is a relatively slow feeding it with e.g. 20Hz to 20KHz the frequency getting higher and higher) this only measures the individual frequencies, and not the kazillion interactions that might take place with real music. The point here is, these interactions are IMO the most important because they are about resonances. And, this is of importance for any device, not only loudspeakers.

While such a measurement theoretically can be setup to a certain extend by means of software that uses a number of sweeps, each one a little faster sweeping than the other, going back and forth until all the combinations have been done (may take a day), in practice it might be undoable because there's also the dimension of waveform regarding "sine" vs "square" and anything in between.

But I think everybody can feel that with such a test we would be more mimicing music, and a simplest form of it could unveil things never seen so far with the single sweep.

 

We'd better work on a time machine first.

Peter

 

 

 

[BobH]

I hope you don't mind me nagging, but I thought I'd ask this again, just in case you missed it in the previous pages - it ended up quite a long way back.

 

I was interested in your advice, about using loop-back recording to check for a bit perfect stream, and was wondering whether this could be done from within the same PC, say using 'Virtual Audio Cable', and still get a reliable result?

 

i.e Media player --> VAC --- > Wavelab

 

Hope you don't mind me asking again.

 

[PeterSt]

No Bob, I sure read it, but forgot to answer upon it. Good that you reminded me.

 

Without trying it myself, I am not sure. The VAC website states a.o. :

 

- All transfers are made digitally, providing NO sound quality loss (a bitperfect streaming)

- Windows 2000/XP/2003/Vista platforms (32-bit and 64-bit)

- Since VAC 4 is a WDM driver ...

 

 

... I can not imagine this allows bit perfect transfer in Vista, since WDM inherently is not.

The first line of the quote is a tricky one, because it very well may indicate "... under the conditions of bit perfect in the first place", meaning that one has to start off with ASIO to do it. But still ...

 

It may be faster to try than to reason out (if I had been in front of my Vista machine), but I also expect problems with Exclusive Mode. I mean, Exslusive Mode = Exclusive Mode and that just doesn't allow anything else in the "mix" which also counts for recording, although the latter depends ...

 

This is a matter of trying I guess, which I will do somewhere the upcoming weekend.

I will try my best in judging the reliability of things, which may take some experience only those programming within the Exclusive Mode stuff can judge decently. To give an example :

 

It can very well be that VAC catches the audio stream even before it goes into a possible OS mixer, which would be the first justificaton of VAC being bit perfect to begin with (this is about the output, and not about the input which *also* needs to be bit perfect, otherwise it's all useless (for testing)). Thus, when indeed it grabs the stream in front of the mixer, it is useless because indeed that part would be bit perfect, but does not simulate our reality at audio playback.

 

I will let know the results when I have them !

Peter

 

 

[ldolse]

The problem with that approach is you're really just testing the software - you're eliminating the connection to the DAC and the DAC's interpretation of that signal. It would be difficult to do without at least having a virtual soundcard to enable for WASAPI.

 

You could something like NetJack - http://jackaudio.org/, or PulseAudio - http://www.pulseaudio.org/. Both of these could provide a virtual soundcard that XXHighend can attach to. However NetJack isn't ready for XP much less Vista, and PulseAudio doesn't explicitly list Vista support, so not sure if it's an option.

 

 

[ldolse]

Our posts crossed - VAC does look pretty promising, at least they list Vista support.

 

[PeterSt]

Indeed it is so that when such software needs to be in between the player and the driver (WASAPI) this is not going to work. But I thought it merely would be a branche.

 

I don't think it is necessary to let the DAC interpret the signal in order to judge for bit perfectness, but, when you'd be able to physically loopback (cable) from the soundcard (this is sufficient) this would avoid the thinking "but maybe it loops back too early" -> in front of the mixer).

 

But this is why it needs some experience, like with the DTS test I mentioned earlier ... this test just does not proof anything for WASAPI, because it will switch into exclusive mode just because of that and things will be bit perfect. This really does not happen when using MM etc. for normal audio playback. Only with DTS ...

 

But we'll see. And thanks for the "backup" links !

 

[PeterSt]

...I thought that was the assumption in this thread: how can applications claiming to be bit perfect sound different. The underlying assumption is that everything else remains the same: the music file, the motherboard, the sound card, the drivers, the OS, the DAC, the Power Supply, etc.

 

True of course. But this doesn't take away the kind of necessity to check that. So it's a kind of other way around : if you drag some sliders in XXHighEnd and the sound changes, it would be quite convenient to check whether all "bit perfect lights" stay on, right ?

 

If I took your remark wrongly, skip my response !

 

 

[BobH]

Peter, Idolse - thanks for the comments. As I don't yet have Vista, I shall try this with XP and a selection of players and settings and see what transpires.

 

It, (loop back recording), seems like one way in which we can test for changes, in a way that is reasonably accessible for all. I don't have the necessary hardware inputs and outputs to do it 'properly' but if VAC gets the stream at the right time then the only thing not being checked, if the thinking is good, would be the final output driver stuff, be it usb or spdif or whatever.

 

But I'm hoping it will still provide a way of testing up to the media player output and providing a way of checking the impact of various settings.

 

I shall have a play!