Re: Status report on Gnome Wave Cleaner, and possibility ofplug-ins

[Jeff Welty <welty harbornet com>]

The LADSPA plugin architecture simply won't support these
algorithms.  I have looked at it (with the help of Conrad Parker).
For example, declicking.  Say the window size is 1000 samples.
A click may span samples 995 thru 1005.  The click detector
won't even find that click until it has access to the second
window.  So, it can't output the first window until it has the
second window.  But my understanding is that LADSPA requires
the same number of samples output as were input.

Someone please let me know if I don't understand this correctly!

What the audio cleaning algorithms *can* do, at some point, is
to start outputing samples which have been fully processed.  Which,
is why the need for a new API.  Denoising, in addition needs access
to a completely separate part of the audio input to get a noise sample
which the user has identified, which is sometimes not at the start of
the audio input.

iain wrote:
> 
> On Thu, 2002-09-19 at 05:13, Jeff Welty wrote:
> 
> > Once library functions exist, then I hope that it will be easier to
> > determine requirements for a pluggin API that would support these
> > algorithms.  Probably then also, someone smarter than me could make
> > the libraries into pluggins.  It would also be possible to code
> > stand alone programs that would read from stdin, write to stdout...
> 
> This plugin specification already exists. It's called LADSPA. It would
> probably be best to code your library to a LADSPA plugin, as most/all
> linux audio apps that use plugins really should support LADSPA.
> 
> Details at: http://www.ladspa.org/
> 
> iain
> --
> "Laws are like cobwebs; strong enough to detain only the weak, and too
> weak to hold the strong" - Anacharsis
> 
> _______________________________________________
> gnome-multimedia mailing list
> gnome-multimedia gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-multimedia