* Jeanette C. <julien(a)mail.upb.de> [2020-04-22 00:05]:
Apr 21 2020, Peter P. has written:
...
If you don't mind the loss of
realtime-tweaking you could use Fons'
convolver, or if you don't mind non-realtime processing why not have a
look at the latest version of the fastconv binary in the composer's
desktop project (CDP)
...
So, do these convolvers actually support some kind of manipulation of
the loaded IRs, i.e. time stretching/compressing the IR, damping the
response in time, ...?
CDP is a collection of about 150 binaries which each execute
a certain
sound transformation, such as filtering, time stretching, applying
fade-outs etc. If you apply these in advance to the IR before you run
fastconv this would give the manipulations you might be looking for.
The fact that this has to be done in a step prior to convolution does
not matter so much in the end, as the possibilities and parameters are
so manifold that often one's brain seems the limiting factor, not the
lack of realtime-processing. In fact, offline computation can also be
faster than realtime. What you will not get, is a fader which you can
manipulate while listening to the changes in realtime. For that you
would clearly prefer Supercollider or Csound with convolution UGens or
opcodes.
CDP examples:
Convolve two soundfiles, creating outfile.wav
$ fastconv violin.wav ir.wav outfile.wav
Pre-process the impulse response by applying a lowpass filter
$ filter variable 3 ir.wav irLop.wav 0.1 1 500
3 selects lowpass, 0.1 is steepness, 1 is gain and 500Hz cutoff
Then convolve again
$ fastconv violin.wav irLop.wav outfile.wav
Why not convolve a second time?
$ fastconv outfile.wav irLop.wav outoutfile.wav
etc.
cheers, P