.
I fully realize that this particular forum is above the complete nonsense that one finds on so many others on the Net', however, for the sake of science... :-) I set up an audio and visual audition demonstration - attempting to discern an audible or at least visual, difference between two completely and utterly dissimilar power leads.

Have a look (Jan 24th post) http://www.cdnav.com

Andrew D.

.

HERE is the normalized difference of both files.

Comments?

EDIT: gah- turns out thats unnormalized- I'm sure y'all can do it yourselves! Its late and I need sleep. I'll try and post a normalized one in the morning.

That is a gaming/multi-media audio card. It most likely records everything at 48kHz and resamples to whatever you have specified. Such hardware related resampling is not generally of the highest quality, although there is some evidence that the hardware is improving. Most companies do not lie as blatantly as Creative, but unless the manufacturer is very explicit about no resampling, you are safest assuming the worst. The best results with such a card are likely to be from recording at 48kHz, to avoid the resampling. If you need a different end sample rate, do it is software. CoolEdit 2000 does excellent resampling.

Taking care of such details is not going to reveal something that isn’t there, but ignoring it does open any tests to legitimate criticism.

BVM:

Not sure what is meant by 'normalized difference'. Is it one subtracted from the other - kinda’ like this?:

http://theaudiocritic.com/blog/index.php?o...35&blogId=1

AndyH-ha:

Have sent an email to ASUS with your query about resampling, will let you know when they get back to me.
However, I guess it does resample as the unit's spec' sheet shows this:



In fact, here is the whole spec' sheet if it helps clear things up in any way:

http://ca.asus.com/products.aspx?modelmenu...p;l3=0&l4=0

On the other hand, is resampling such a bad thing if done right?

Andrew D.
www.cdnav.com

.

[citation needed]

I’m don’t know what those numbers really mean. The performance of some DSP filters may refer to a particular part and not be very representative of throughput. The dynamic range of 32 bit transforms in CoolEdit is around 245dB. Testing of actual results is the only way to get the full story.

Resampling per se is not necessarily a bad thing in any absolute sense, but that is not the same as saying it is in any way desirable as part of a recording process. I did not say it would necessarily be poor quality with this soundcard, as it is with some cards, only that any aiming for maximum quality has to be through avoiding resampling:

** We know soundcard hardware resampling has generally been fairly poor; even many software programs do not do a very good job of it.

** If this extra step is in the processing, can we really say that the failure to find differences when there are differing analogue treatments might not be an artifact of the resampling? I think not without at least first fully proving the resampling independently of these tests.

Yeh, thats the basis for what I did. You align the two files, put one out of phase and you are left with the difference between the two files. For analysis purposes, I normalized it; a process that scans the file for the highest peak then boosts that peak (and hence the whole track) to (in this case) 0dB.

Normalized differences are invalid unless either you synchronize the tracks using intra-sample delays, or digital's playback clock was synchronized to his recording clock.

The technique of subtracting the before-and-after signals can show convincingly in some cases that there cannot possibly be any audible difference, but the reverse is not true. Even when there is significant difference, that difference still may or may not be audible.

Used to show that line cords do not affect sound reproduction is a valid use of this technique.

Unless he copied both original audio files into a single audio file *before* normalizing the single file...I think.

-brendan