Home Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Even if the software were perfect ....

...which as many know is a big assumption to make with software, the resolving power of the method is inherently limited by the resolution of the soundcard used.

To put it as simply as possible, unless the soundcard has enough resolving power and fidelity that you would be able to listen to music through it and use it for all playback chores, without being in the least dissatisfied with the playback quality of the soundcard, then this method would not be able to reliably detect any differences for a system that was at or above (less signal aberrations) the level of performance of the soundcard.

Or put another way, this method is inherently incapable of resolving to or below the limits of the soundcard used.

While this might seem trivial, I have yet to hear a soundcard, even the $500-600 pro grade ones, that are totally transparent and the equal of near-SOTA audio components or systems. Yet the author is claiming that a $200 soundcard is all you need to perform this kind of difference testing.

In order to make this clearer, I will provide an over-the-top-example:
lets say the soundcard used had an ACTUAL linearity such that it literally failed to respond to any data below the 15th bit, or put into a different framework, that pretty much everything below -90 dB full scale was obliterated. Now, given that we really cannot record in the real world all the way up to full scale, we have to leave some headroom, even with familiar signals and events, because the alternative, possible digital recording system clipping, is just not an acceptable alternative, we will probably have less than 14 bits of resolution to record the two events, and to try and create a difference file. If there were ANY difference between the two recorded events that were below approx. -84 to -85 dB of full scale (this presumes allowing approx. 5-6 dB of headroom, which would be pushing the digital recording pretty hot), it literally would not be discernible. If the two recordings were identical above the level of -85 dB, then there would be nothing in the difference file, just residual noise.

The signal below -85 dB could be COMPLETELY DIFFERENT, and we would never even see/hear anything in the difference file, just that touch of residual noise.

I know that some folks are of the opinion, that if all the subtle sonic differences live below -85 dB, then they are automatically inaudible anyway, and thus, my example would be a moot point. But unfortunately for them, this is not the case.

One of my answer's to this is that there several things that occur at levels at or below -90 dB FS in a digital system that are quite certainly audible, and one non-trivial example is digital dither algorithms. As anyone who has actually made recordings for a living can tell you, the choice of the dither algorithm is not at all trivial, and that each one sounds quite different, each one has it's own sonic signature, and each one lives in it's entirety at levels below -90 dB FS. Sony's SBM, Apogee UV-22, POW-r, ExtraBit/MegaBitMax, etc. each one sounds different, each one has it's proponents in various studios, and each one only affects the signal at very low levels, typically below the -90 dB FS point.

If the overall tonality and sound of a recording can be altered by such algorithms that all work at such low levels, then this means that we can not ignore signal aberrations below -90 dB as insignificant, to the contrary, they are capable of having tremendous sonic impact.

Of course, the obvious answer is that you can find a sound card that has 'resolution' below the 15th bit, indeed, according to the specs on a spec sheet, there are cards that are 'linear' down to 17 or 18 bits.
But the bottom line is still: can you listen to the sound cards output and not detect any amusical occurrences, is the soundcard literally completely sonically transparent? If it is not, then it could be masking other components or systems aberrations and differences to whatever extent that the soundcard lacks ultimate sonic perfection.
Note that the listening test only invokes the DAC, the ADC (input) recording portion is totally ignored in this kind of 'test'.

In order to really test the sound card, both the ADC AND the DAC need to be exercised, and for this, what is required, is to record the output of a very high quality 'source', say the output of a reference grade CD or SACD or DVD-A player, and then compare the net result of going in the ADC and out the DAC on the soundcard, and comparing that to the output of the 'original' source player.

I think that by now, you are beginning to get the picture, no sound card out there can pass this test, they all add something that is not 'musical' to the sound, when compared to the original 'source', there are various and sundry issues and problems, and again, the bottom line is that if the soundcard is less than perfect sonically, whatever faults it has, it will basically be incapable of resolving any differences below it's own limit's of true resolution and clarity, and thus, any difference files will be inherently limited to it's sonic accuracy.

One could refuse to acknowledge this, and insist that bits are bits, and that if a soundcard has more than 16 bits of resolution, and the FR is XX dB flat, etc, it is therefore perfect, and can provide all the capability needed in the first place. But now we are back to the same old impasse: if the specs say it is perfect, therefore it must be perfect, and can resolve everything that matters. On the other hand, if it does not record and listen completely transparently, it has some sort of inherent limitations, and thus cannot resolve below it's own limits. Now we are back to where we started, the test is said to be good enough by one group, and it is said to be inadequate by another group.

BTW, all of the above does not even bring into question the HUGE issue of whether or not an arbitrarily timed event can ever be compared to another arbitrarily timed event. Whatever am I talking about? If you record one set of components, and then make a change, and record another set of components, the actual timing of the sampling will not be synchronized.

Say we record a transient event, an infinitely fast and short impulse. In one recording event, it gets recorded exactly when the sampling interval is catching the peak, on another recording event, it captures the peak of the impulse dead in the middle of two of the sampling intervals, that is, you have two samples that would be at (approx.) half height of that single sample.

While this also seems trivial, and that the transient information has been preserved, there is no way to exactly line up the digital samples so that the two recordings could fully lineup if the two recorded events were indeed exactly the same, thus, there would be an irreconcilable difference residue that could not be "adjusted" away.

So even though you could have recorded two exact same events, there would exist a difference that would not be real. OK, in the example I gave, the grand total of the error would be limited to a few samples, and the rest would be random noise floor, etc. But what about a complex piece of music that you were using to make the recordings? ALL the samples would be off "a little bit", and there would be a constant tiny error due to the sampling limitations involved.

But that is not the only problem, soundcards in computers are very much subject to a form of digital recording and playback error known as JITTER. Now, we would have the sample to sample timing error CHANGING, and the jitter differences between the ADC and the DAC would only add to the difference error, not cancel.

Guess what most soundcards do not even bother to spec? ADC or DAC jitter. Who knows what a given soundcards level of jitter actually is?

This kind of error would also tend to blur and limit what the resolving power of the soundcard was and just how effective it could be in creating an accurate difference signal file.

The software could be perfect, and yet the soundcard is in the real world, where nothing is perfect, and thus this type of test IS definitely limited in it's resolving power.

Knowing that there are NO soundcards out there that are sonically perfect, it would be an open question as to just how many were "good enough" to provide meaningful results.

Note that the kinds of faults and problems I am mentioning almost all result in a false null, that is, little or no difference file content.
Of course, far too many objectivists will take such a result as "proof" that there are no sonic differences, but the actual truth would be, we just wouldn't know for sure. In other words, this kind of test actually and literally suffers from the same kind of limitations as all of the DBT type listening tests: failure to achieve a significant result will NOT mean there is no difference, just that we could not detect it with that particular set-up. Period, end of the truely scientific claims that can be made.

However, I have absolutely NO doubt that the vast majority of objectivists will indeed claim that such had occured, and this time they would have the cachet of "scientific proof" that there was no difference, only it really wouldn't be true.........


Jon Risch


This post is made possible by the generous support of people like you and our sponsors:
  Atma-Sphere Music Systems, Inc.  


Follow Ups Full Thread
Follow Ups


You can not post to an archived thread.