Mike Hirst writes:
>To be blunt, 
>a good soundcard can be let down by poor software.

Bingo!  Even very expensive and "industry standard" software may not be
handling all the bits properly depending on what process within the
software is being used.  One doesn't know who did all the coding, how
careful he/she was handling the data (double-precision?  dither rounding? 
truncation?) or which subroutines are so badly written that they undo all
the good work and careful data handling that some other fine programmer
took great care to maintain.

It's especially dangerous to capture through the OS's "mixing console." 
Even if you have all other sources muted, there will likely be noise,
dithering, level changes or outright truncation occurring with your input.

As an example, a client once sent me the "digitized" file of a precious
analogue source which he would not let out of his hands for me or anyone
else to transfer for him.  He just wanted me to clean up the resulting
transfer for archive and potential release.  I offered advice on good,
clean cards, A-D converters, etc. and my advice was used to put together
what should have been a nice transfer.  Unfortunately, the capture was
done using the "mixing console" of the OS and the first file I was sent
had lots of noise, digital hash, peak distortion and occasional "system
sounds" mixed in with it.  I tried to talk the client through the process
of eliminating the trouble and we made some progress, but the final file
still had lots of problems that I know from experience were not caused by
the hardware but were added by the software used to capture the file.

Definitely do NOT take anything for granted in this realm.  Run your own
listening tests.  Compare the analogue source with your digital capture. 
If you hear a significant (noticeable?) difference, try another input
signal path to see if it makes a an improvement.  Try other software!  You
may be very surprised by what you find.  Just be sure to keep your
monitoring setup constant so that you don't introduce additional variables
into your informal testing.

Good luck and tell us what you discover.

Charles Lawson <[log in to unmask]>
Professional Audio for CD, DVD, Broadcast & Internet