Yes, this is an interesting technique. Thanks.

Maybe it could be done by writing the fsig to a table, and then
implement the spectral flux calculation from there.
Makes me wonder:
The kflag in
kflag pvsftw fsrc, ifna [, ifnf]
Is that used *only* to synchronize the f-rate and the k-rate processing,
or should I expect an additional delay here somewhere ? (meaning, do
the pvsftw add a delay somewhere, or is the data simply written at
f-rate, and can be read at k-rate).

best
Oeyvind


2008/4/22, Jamie Bullock <jamie@postlude.co.uk>:
>
> Hi Oeyvind,
>
> If I understand your question correctly, the spectral flux feature will
> give you the information you need.
>
> http://en.wikipedia.org/wiki/Spectral_flux
>
> I'm not sure what the best way to do this in Csound would be. I'm a bit
> out-of-touch with Csoundy things at the moment. Maybe you could do it
> natively with the pvs opcodes and some maths, or maybe a new opcode
> would be needed. Alternatively, maybe it would be good to write a VAMP
> plugin host for Csound, this would make available a couple of SF
> implementations along with a whole load of other useful features...
>
> Jamie
>
> On Mon, 2008-04-21 at 14:05 +0200, Oeyvind Brandtsegg wrote:
> > Hello
> >
> > I'd like to do transient analysis of an audio signal in Csound,
> > preferably analysis in realtime. I'm unsure about how to go about
> > this. I did a simple test of taking the centroid (pvscent) and looking
> > for steep ascending slopes. This sort of works, but not perfect. Maybe
> > it could be combined with amplitude (ascending) slope detection, but
> > I'm sure there are better ways to do this.
> > It would be useful to have degrees of "transient-ness", and also to
> > have the signal gently slope back to zero as the analyzed audio
> > stabilizes on a "steady state" spectrum.
> > Maybe a new pvs opcode ?
> >
> > Oeyvind
> >
> >
> > Send bugs reports to this list.
> > To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
> --
> www.postlude.co.uk
>
>