[sdiy] Timbral musings

[Czech Martin]

Ian, just
some ideas, not at all thought to the end:

-Wave shapers work in time domain, do not necessarily
have strong and interesting effect in the frequency domain.
What looks interesting on 'skope is perhaps dull in
the speaker. This can be frustrating.
-at some setting there may be partial cancellation effects.
If this is true, it is clear that the parameters are critical.
-after all, a wave shaper does convert a static signal
into another static signal, some harmonics will change,
but one should not expect wonders.
-I think that the ear/brain looks for shifting patterns.
The quality of vocoder speech is poor (from a signal
analysis point of view). But since the characteristic
movement of patterns is preserved, it is still
intelligible.
-interesting sounds have interesting movement of patterns.
Wave shapers will not introduce non-harmonic partials
(single input, try multiple input). A clipper circuit
that will clip a saw wave will only shift the harmonics
content to some degree, they are already there.
So perhaps using a sine as input is more interesting,
we know this from fixed carrier "FM".
-wave shapers that introduce new partials that weren't
there before are perhaps more interesting,
this will certainly happen if a sine wave is used.
-the wave shaper should have a parameter setting
that allows unaltered signal fed through.
-masking will make little changes in partial amplitude
inaudible, another hint not to start with too rich
input


Just a few ideas from my many hours listening experiments
I did with DIY wave tables, hand-drawn waves and
distortion functions etc.
It's like digging in a heap of crap, but sometimes
you find a little piece of gold in it.

m.c.


-----Original Message-----
From: Ian Fritz [mailto:ijfritz at earthlink.net]
Sent: Samstag, 15. Februar 2003 19:13
To: synth-diy at dropmix.xs4all.nl
Subject: [sdiy] Timbral musings


Hi folks --

I've been thinking some about a dilemma I've noticed with regard to 
waveshapers. Let's try out some ideas here.

The dilemma is this:  On the one hand, if you listen to a waveshaper as it 
is tuned (or modulated) through a range of timbres, the tuning often seems 
to have a "weak" effect. Similarily if you take a multistep waveshaper and 
shove the sliders around you get the impression that all the sounds are 
more or less similar.

On the other hand, if you put the waveshaper in a patch and try to get a 
certain sound you may have in mind, you notice that small differences in 
waveshape make a significant change in the sound. In fact it can take very 
careful adjustment to get the sound you are after.

So how do we reconcile these effects?  I think the answer has to do with 
the physics and psychoacoustic of how we hear and analyze sound.

When you consider listening to a sound in a room, you might naively expect 
to get very confused because of all the reflections, phase cancellations, 
etc.  The reason you don't, as I have read, is that the ear/brain actually 
averages spectra over different positions of your head as you move around 
slightly.

Thus it ends up taking quite a bit of time to analyze a sound you are 
hearing. I think I notice this effect when my alarm radio comes on in the 
morning. Even if I am awake, I cannot immediately tell what instruments are 
playing -- it takes a couple of seconds to lock onto the sound.

 From these ideas, I think that when you sweep a waveshaper your 
head/ear/brain simply can't work fast enough to keep up with the 
changes.  There are big differences in sound, but they just don't register. 
That is why my demo clips for the "Wavolver" waveshaper emphasize 
steady-state tones (with quick sweeps in between).  I was just listening to 
those again, and it really does seem like it takes several seconds to 
adjust to the changes.

The counter-argument here is that filter sweeps give a "strong" 
effect.  Perhaps this is because the filter is so drastic in subtracting 
high frequencies, whereas a raw shaper waveform always has lots of harmonics.

I would really be interested in hearing what you good people think about 
these ideas.

Regards,

   Ian

   (Wavolver clips are here: http://home.earthlink.net/~ijfritz/sy_cir6.htm  )