18 Nov
2008
18 Nov
'08
11:46 a.m.
On Mon, Nov 17, 2008 at 08:21:44AM -0500, Fred Gleason wrote:
On Monday 17 November 2008 07:54:18 am Hannu
Savolainen wrote:
One case where you want complete control of
the hardware is any application that generates,
measures, or analyses *calibrated* signals.
Having to set or verify gains and check the
presence of other signals or unwanted system
level processing on a 'control panel' is an
absolute no-go in such an application.
Other fields where you may want some control over
the hardware, and most certainly no hidden layers
that may interfere with what you do are doing (e.g.
some unspecified resampling algo), are:
- professional music recording,
- broadcasting,
- medical applications,
- forensic audio,
- anything requiring official certification and
traceability,
- anything using non-consumer multichannel formats,
e.g. electro-acoustic music,
- psycho-acoustic research,
- acoustic measurements,
- ...
These are just some I've been involved with, there
must be more.
Just one example of how things can go wrong by the
system level trying to be clever: last year a research
assistant here was using an 8-chan sound card to do
some experiments with beamforming. The results were
not what he expected, and we could not explain what
went wrong. In the end we found out that the first
two channels of his card were going through a software
'mixer' (a 'system feature', and impossible to bypass)
and as a result had more delay than the six others.
Ciao,
--
FA
Laboratorio di Acustica ed Elettroacustica
Parma, Italia
Lascia la spina, cogli la rosa.
So what kind of professional do you mean?
For purposes of our discussion, call it a class of application designed for
use in 'unusual' environments. Something other than generating sounds for a
single user on a desktop. Fons' multi-cpu/multi-site application does very
well for an example.