On Fri, 8 Jul 2011, Alexandre Torres Porres wrote:
It seems now it's a matter of aligning the phases, but not sure how, good night
Phase alignment has to be done by a slight temporary pitch shift. You can shift up or shift down, but in any case, you need to shift.
If you do a plain linear cross-fade of signals (or a plain linear cross-fade of spectra, which is the same thing), it will automatically do this temporary pitch-shift, but will also do a temporary amplitude-drop at the same time.
This is because if you have two dots on a circle representing two phase alignments at the same amplitude, and you go from one to the other, you will go inside of the circle, where the amplitude is lower. The centre of the circle represents silence. In the extreme case of complete reversal of phase, the crossfade will quickly go through the silence point. But each frequency will have a different crossfade, so, overall, the amplitude drop will rarely be great, except if you tend to have few partials.
I'm thinking about phasor diagrams like http://en.wikipedia.org/wiki/Phasor , which may be considered as a use of Argand diagrams (complex numbers plane).
To keep constant amplitude, if you want to, is like rotating dots so that they keep the same amplitude (distance from centre). This is usually not easy to do, but you already have a components representation using sigmund~, so you have the opportunity to fudge the freqs a bit...
| Mathieu Bouchard ---- tél: +1.514.383.3801 ---- Villeray, Montréal, QC