[Nika]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by Noiz2:


The filtering that happens at the output stage of the D/A averages between the steps created by your samplerate. A higher samplerate has smaller steps more information and therefore less averaging going on. Any "real" information that falls between the samplerate "steps" is lost so by it's nature a higher sample rate will give a more acurate picture of your original source. <HR></BLOCKQUOTE>

There can't be any "real" information between the steps since it was bandwidth limited on the way in and throughout the system. Recognizing this, there is only one possible way for the intermediary points to be constucted with an FIR - the most accurate way.

If I tell you that a number sequence is going to not contain any interval larger than 1, and I give you 1, x, 3, x, 5, x, 7, x, 9, x, 11. Then there is only one possible solution for each of the intermediate numbers. Sure, you could say that providing the following solution will provide for less "guesswork", but is it really "guesswork", or "averaging" at all?

1, 1.5, x, 2.5, 3, 3.5, x, 4.5 etc. etc.

There is no benefit to providing the D/A with more data than is necessary for it to reconstruct a waveform properly. And if the human ear AND the A/D are both band limited then there is no benefit to providing additional out of band material for the D/A. It will not "give a more acurate picture of your original source" after the A/D's filter. And I would argue that the work of the A/D's filter is capable of being audibly transparent.

Nika.