At 11:47 AM 2008-08-17, Mike Hirst wrote:
>I guess what I am asking here is to do with specific combinations. I
>must admit that this is something I have never questioned. It has
>been my assumption that analogue audio enters one side of the AD
>converter (soundcard/audio interface) and the digital audio produced
>is represented through the software GUI - that the software does not
>change the digital content during capture and more importantly that
>different software does not alter the digital content, it merely
>represents the same content through a different GUI. However it
>seems that what is being said here is that different software will
>not only visually represent the same audio capture differently, but
>will also fundamentally alter or misrepresent the data part of the wav file.
While you've probably thought of this, I think it's good to reiterate
in this thread that there is no "truth" in the digital numbers from
an analog source.
Since the A-to-D clock is not phase-locked to the analog audio
"clock" (since there isn't one), you will most likely get different
numbers (and even different waveforms at the sample level) depending
on the phase of the audio and the A-to-D sample clock.
In other words, a different set of numbers can end up via the
reconstruction filters providing the same original sound--it all
depends on where the signal was sampled.
So, it is impossible to compare different A-D converters from the
same analog source (tape or microphone) by comparing the digital
output. One might come closer if all the converters were locked to
the same reference and the conversions were done simultaneously, but
even then, I suspect that the numbers will differ.
I found this out in real life when I was repairing bad sections of a
tape and I transferred it several times and I was looking to line up
at the sample level and discovered that the samples were different.
The same overall pattern was there, but when zoomed into individual
samples, say around a particular zero-crossing point, the actual
sample values were different (and different enough to see on the GUI,
not just by comparing numbers.
Conceptually think of a sharp transient at 1/2 the sampling
frequency. Say it has a value of 10. If you're lucky, you'll have a
single sample of 10...but you might get samples of 6 and 4 (or do
they need to relate by the square root of two) but you get the idea.
Richard L. Hess email: [log in to unmask]
Aurora, Ontario, Canada (905) 713 6733 1-877-TAPE-FIX
Detailed contact information: http://www.richardhess.com/tape/contact.htm
Quality tape transfers -- even from hard-to-play tapes.