Subject: Re: [linux-audio-dev] embedded sound api.
From: andrew_AT_acooke.org
Date: Wed Nov 21 2001 - 18:36:50 EET
On Wed, Nov 21, 2001 at 10:10:51PM +0900, Patrick Shirkey wrote:
[...]
> quality. Also I would want to incorporate a sound conditioner (is that
> the right word?). This would be able to take a sample of the standard
> input and then attempt to equalise or delete it from the track before
> writing to disk.
>
> In my opinion this will be one of the most important features because it
> will allow the user to target sounds they want to capture by getting rid
> of the unwanted peripheral noise. It will also be good for people who
> have cheap mics or happen to be recording in a high solar wind period or
> a lightening storm etc... ;-]
[...]
> What I would like to hear from people who are interested in sharing
> their knowledge is what drawbacks they can see in the software
> design/setup and ideas on how to fully open the internal power with the
[...]
I know nothing about audio, but I have spent a lot of time processing
images (in astronomy), and I suspect some of the lessons carry across.
One of the hard lessons is that you don't get something for nothing -
you can't get a good quality signal out of a bad one. So I'm a bit
worried about the details of your sound conditioner.
More precisely, you usually need to know the details of both noise and
underlying signal very well before you can do anything "clever" with
noise removal ("very well" could mean, for example, that you already
knew the notes you are recording, and when they would occur, or that
you separate record the noise of the lightning away from your main
source *at the same time* - in parallel - as your main recording).
So I suspect the best that you will be able to do in practice is low
pass filtering (you could use the pre-sampled noise to adjust the
filter parameters to some optimal values). I doubt this would sound
much better than a simple electronic filter (something with a couple
of knobs the user can twiddle to adjust performance by ear).
Do any packages exist now that take noisy recordings and make them
not-noisy?
If that (low pass filtering) is what you're aiming at, or there's some
clever processing in audio that I don't know about (perhaps because
you can sample at a much higher frequency than the data your are
interested in?) then fine. This is just a warning in case you were
hoping for spectacular improvements in sound with no real idea how.
Of course, I may simply be wrong - it's happened before.
The rest of it sounds great (although I wonder if people are becoming
so computer-literate that they are prepared to use PCs rather than
specialised devices).
Good luck,
Andrew
-- http://www.acooke.org
This archive was generated by hypermail 2b28 : Wed Nov 21 2001 - 17:16:30 EET