On Sunday, August 17, 2008 6:14 AM, Mike Hurst wrote:
> However, reading Goran Finnberg's comments re the work dome by George
> Blood, I am surprised to find that different combinations of software
> and hardware can produce different results. Am I to understand that
> Adobe Audition, for example, will capture audio differently when used
> with an M-Audio card, for example, than it would when combined with an
> E-MU or SoundBlaster card? does it then follow that the same card
> combined with different software would produce different results? Is
> this something I should worry about? What is the extent of this
> discrepancy and how best can I avoid it?
- the ADC has an analog section, which can influence the sound
ahead of the converters by introducing analog noise,
distortion, cross-talk, etc.).
- not all converters are equal in terms of their digital precision
- clock synchronization between the ADC and the computer can be
an issue (whether an internal or an external ADC)
So in the first case of Adobe Audition used with a variety of
capture (ADC) cards, there's a good chance that you will indeed
hear subtle (or perhaps not so subtle) differences and get
different results, and the differences will be due to the cards
themselves as well as how they interact with the software (ie.
On the flip side, using the same card with a variety of software
is an interesting question. Each software package should be
receiving the same bitstream, particularly if they are going
through the same hardware interface with the same drivers.
I've not tried this, but doing the test of multiple software
packages with a single ADC would seem to be tricky. To do
this test correctly, you would need all the software packages
to capture the same digital audio stream simultaneously (ie. in
parallel) on the same clock. To do otherwise will introduce
either slight input signal differences or quantization
differences on the part of the ADC - a natural part of
digitization. This test would require either a powerful
workstation so that none of the software packages are starved
for resources (ie. buffers, CPU) as each software is running
at the same time, or you need a well-tested digital distribution
system to multiple computers. Some software packages may take
exclusive control over the audio stream in order to guarantee
sufficient computer resources and to prevent other processes
from introducing audio or noise to the capture stream, making
such a parallel software test impossible on one computer if
this is the case for a particular software package.
I've not seen the George Blood presentation, but would be
interested in hearing about the detailed test and
To your questions: (1) is this something you should worry
about, and (2) how best can you avoid these differences?
Regarding (1): Yes, differences in audio capture are something
you should worry about. "Worry" is perhaps the wrong word
here - "Understand" might be better, so that you can make
decisions on how you can best capture your audio to meet
your needs (access, preservation, budget constraints, etc.).
Regarding (2): I would encourage doing your own tests. Even
if the technical testing and measurements are beyond your skills
or means, your ears (along with others' ears) are still a good
tool (although perhaps constrained by your monitoring equipment
or the acoustics of your environment). Do lots of A-B testing
between the source and a variety of capture combinations. Even
if you can't identify the root cause of the differences, you'll
be able to identify elements in the capture chain which sound
more right or true. And you'll learn a lot and fine tune your
listening at the same time (always good things if you're
involved in audio).
The Audio Archive, Inc.
mailto:[log in to unmask]