Waveform morphing...

[Johan Persson]

> what do you mean by "morphing"? If you mean simple adding of two waves

I thought of trying simple adding at a start and if the results are good,
there are no need to dig deeper. It has also struck me that it could sound
like a very uninteresting fading effect, but opposed to analog VCO's fading
into each other the two waveforms are exactly synced which probably makes the
fading not as obvious (that is at least what my sometimes illogical brain came
up with...). I have a friend who wrote a piece of software some years ago that
used another technique. From what I recall, it used several points adjustable
both in X and Y, and changed the positions of these points gradually. When I
think about it now, it's basically the same thing as "fade-morphing", but some
of the sounds where quite interesting...


> you can do this with almost any rom wave synth (rompler) that offers 2
> or more patches,layers etc per voice.  Just select different waves for
> the patches of one voice and then program the envelope such that you
> hear first the first patch, then both and finally the second one. Ok,
> this is very simple morphing, but it works as such.  Most experiments
> of this kind that I did where not that much exiting...  But sometimes

I was more thinking of doing the calculations in real-time and not have two
oscillators changing volume. This would just take up one voice! And the
thought was also to be able to change between more than two waveforms in time.




> Or do you mean a method like , well, let's see ah, initial spectrum has
> three formants, and final has only one , but at another location, so
> I morph between the spectra by slowly shifting one formant to the
> final position and fade out the superfluous two.
> Well, that's interesting, but also complicated....

I'm not quite sure what you mean. Anyway, this would probably be more than I
could handle. My DSP skills is not what they should be...



> However,this sounds like a lot of work.  If you do this all digital,
> you'll end up with multirate and interpolation (well, you want to play
> the waves at different pitches) and a hybrid solution requires a quite
> high frequency for variable memory read out that must be exact enough
> to match your scale (hf-vco, or pll vco).

Yep.. It would too much, if it weren't for the fact that I'm going to cheat!!
(hehe...). If I do my testing on a computer, why not have the finished project
running on a computer aswell..? I planned on using my ol' Amiga 500 (it's just
stuffed away in a closet anyway), which has built in four voices audio-chip
and a serial port. Why do a lot of work already done? It's not actually the
sound I'm looking for (there are more interesting ways and machines for
producing sounds), but it's a nice project. If I would be smart, I would give
up these ideas right now and go buy me a microwave (still not an oven).

Johan