.quasar

I found myself implemented this synthesis technique after reading this paper :

http://recherche.ircam.fr/pub/dafx11/Papers/55_e.pdf

It's a great development of phase distortion synthesis and quite an original way of generating formants. However, as all non linear synthesis technique, it does generate quite a bit of harmonics. By linearly interpolating between 2 adjacent formant integer ratios and oversampling, the aliasing can be reduced quite a lot. But it is quite costly in terms of CPU.

What I found really interesting is that the paper includes a novel single oscillator anti aliasing technique. But I haven't been able to reproduce it yet in gen~. The algorithms doesn't seem too complicated in the paper but every time I tried to adapt it, the result was just the opposite of what it was meant to be . I found a python implementation by one of the author but I can't seem to adapt it either to gen~ codebox.

Anyway I'm including the implementation using the linear formant interpolation technique. I also included the python code in a gen~ codebox elsewhere in the patch in case someone more gifted than me is willing to have a go at it ! ;) I will still keep on trying though.

gen~vps-codebox.maxpat

It's a great development of phase distortion synthesis and quite an original way of generating formants. However, as all non linear synthesis technique, it does generate quite a bit of harmonics. By linearly interpolating between 2 adjacent formant integer ratios and oversampling, the aliasing can be reduced quite a lot. But it is quite costly in terms of CPU. 

What I found really interesting is that the paper includes a novel  single oscillator anti aliasing  technique. But I haven't been able to reproduce it yet in gen~.  The algorithms doesn't seem too complicated in the paper but every time I tried to adapt it,  the result was just the opposite of what it was meant to be . I found a python implementation by one of the author but I can't seem to adapt it either to gen~ codebox.

Anyway I'm including the implementation using the linear formant interpolation technique. I also included the python code in a gen~ codebox elsewhere in the patch in case someone more gifted than me is willing to have a go at it ! ;)   I will still keep on trying though.

I found myself implemented this synthesis technique after reading this paper :
http://recherche.ircam.fr/pub/dafx11/Papers/55_e.pdf
It's a g

«sharing»-vector-phaseshaping-synthesis

Oh! very nice patch! Thanks 
More I play with it more I like it

hey, here's the python code translated to genexpr. should be correct, but doesn't sound so :(

hey,  here's the python code translated to genexpr. should be correct, but doesn't sound so :(
some weird stuff happening...


 That was quick ;) Thanks ! Yeah I see what you mean, it does weird things when the v parameter is being modulated.

At the same time it does seem to work : when the v param is not an integer you do see the anti-aliasing happen if suppress=1. It's just that for some reason, the spectrum changes abruptly when v= 1.5, 2.5, 3.5.

@ VOLKER BÖHM  That was quick ;) Thanks ! Yeah I see what you mean, it does weird things when the v parameter is being modulated. 

At the same time it does seem to work : when the v param is not an integer you do see the anti-aliasing happen if suppress=1.  It's just that for some reason,  the spectrum changes abruptly when v= 1.5, 2.5, 3.5.

[sharing] Vector Phaseshaping Synthesis