< Back to IRCAM Forum

SuperVP - generalized cross synthesis

Hi All

Could somebody may explain the order of the param file (amplitude scaling/frequency scaling … of sound 1 and 2)(s.screenshot I mean third slot)

the included file has this matrix
0 1 0 1 0 0
0.5 1 0 1 0 0
2 0 1 0 1 0
4 0 1 0 1 0

but unfortunately the example in the documentation is not working… :smiley:

and how to change between different cross modes? t= constant cross? and the rest? (s screenshot: I mean 9. slot)

would be great

Thanks :slight_smile:

best L



Well the parameters for the Generalized Cross Synthesis are as follow, according to my notes and this AudiosSculpt reference
http://support.ircam.fr/docs/AudioSculpt/2.8.3/co/cross-synthesis.html. :

If you consider S1 is the fft spectrum of sound1, S2… of sound2 and Sout the result then,

in mode <add> (just click on the mode slot, the 9th, and a menu will appear) :

Sout = XS1 + xS2
Where X is the amp coefficient “Amplitude Scaling” of sound 1
and x is the amp coefficient “Amplitude Scaling” of sound 2
so that’s an amp mix

whereas in mode :

Amp(Sout) = Xamp(S1) + xamp(S2) + qamp(S1)amp(S2)
Freq(Sout) = Y
freq(S1) + y

X and x as above
Y is the phase coefficient ("Phase Scaling”) of sound 1,
y is the phase coefficient (“Phase Scaling”) of sound 2,
q is the multiplication amplitudes coefficient “Q Factor”.

and at last the order of parameters for the OM SuperVp object is :

(time X x Y y q).

I Hope I’ve made no mistake typing all these params !



Dear Nadir

Thanks a lot for this very nice and detailled description!!
Now it is working :slight_smile:

Is it also possible making a stretch factor or matching the second sound to the first if the files do not have the same length?

All the best



You’re welcome.

You may have noticed the ‘match’ and ‘stretch factor’ params in the Audiosculpt General Cross Synthesis Interface. You may also apply a time compression/dilation factor with the -D flag in supervp command line ((a code which is (very!) less explicit than the interface but there’s a full documentation : supervp -h in terminal, you need to install the command-line tools in the kernel folder within AS or shipped separately in the forum’s apps)).

I can’t find a way in OM supervp-cross-synthesis object (there’s no optional hidden slot for the compr/dilat parameter, in the code neither: select object and type “e”… ). So, you need to make that operation beforehand with supervp-timestretch to match the length of 1st sound (be aware that the duration of the sound include the silence within.)


Dear Nadir

Yes… I did :wink:

the first soltion sounds complicated…

I think I will do it by hand before with timestretch…

Thank you!

All the best

You may use the OM sound-dur-ms object to get the duration of your sounds then some math to calculate your factor i.e.

S2_Duration * F = S1_Duration => F = S1_Durtation / S2_Duration

am I right ? Hold on…


Cool idea! That works nice. :slight_smile:

Thank you, Nadir!

All the best