Doepfer

Hi Bakis, list,

Here are a few more suggestions for new modules. First of all I have 
to say that the ideas are not at all 'mine', but are all inspired by the 
insights of William Sethares. If there was something like a Nobel 
prize for music, Mr. Sethares would surely get it for his deep 
insights in the relation between timbre (spectrum) and scale. For 
more info check his website:

http://eceserv0.ece.wisc.edu/~sethares/ttss.html

or his book 'Tuning Timbre Spectrum Scale' 

or this java applet that shows how to derive a scale from an fm-
sound:

http://eceserv0.ece.wisc.edu/~sethares/forrestjava/html/TuningAndTi
mbre.html


Now: suggestions for modules:


1. Equal Temperament Quantizer

Think of a single A-156 without the three switches, but with one extra 
added knob (or perhaps two knobs/jack input for cv control). 

The A-156 is based on 12 tone equal temperament (12-tet) which 
means that it devides an octave in 12 equal steps of 100 cents. This 
is very good for natural sounds, which is why the 12-tet is so very 
very popular. But there are more possible equal temperaments. And 
allthough it is allready possible to create other et's than the 12-tet by 
attenuating the signal from the A-156, this will only give you et's with 
steps that are *smaller* than 100 cents. If you want bigger equal 
steps this is not doable with the A-156. 

So an equal temperament quantizer would have an added knob to 
decide in how many steps an octave would be devided.

n=1: only octaves (1200 cents)
n=2: octave and tritone (600 cents)
n=3: 3 steps of 400 cents 

etc. probably up to 12, or maybe up to 24 or 25

Note for Dieter: maybe this is allready possible with the upcoming 
TKB-controller? And if so, does it work just like a quantizer, or is 
there a difference? And I guess it would be possible to derive note-
sequences from any cv-source, not just from the TKB?   


2. Timbral Quantizer 

If you check out the relation between the harmonic spectrum of a 
certain sound (frequency and level of partials) and the intervals or 
chords made with that sound which are most pleasing or consonant 
sounding, you find that for most 'natural' sounds (like an acoustic 
guitar) the 12 tone temperament is great, because in 12-tet the 
partials don't tend to clash, so to speak. (Generally speaking of 
course, because you *can* come up with awful chords if you really 
want to)

But for e.g. metallic sounds, or indeed for lots of *electronic* 
sounds, this is not the case. This is why it is especially interesting to 
experiment with different scales and tunings if you are into metallic 
sounds (gamelan, metallic xylophones etc.) or if you're into 
electronic music!! 

The beauty is that there is now mathematics available (thank you Mr. 
Sethares) to derive a fitting customized scale, that is optimized for 
consonance, from almost any type sound/timbre. This is usually 
done using Fourier analysis. But I don't want to propose a Fourier 
analysis module, because it is possible to derive the necessary 
information in another way, if you know the 'recipe' or 'settings' of a 
sound/timbre. This method could definitely work for additive 
synthesis and fm (maybe for analog fm).

So I would propose a dedicated fm-operator or sine bank or ring 
modulator that can be 'read' (preferably real time) by a timbral 
quantizer so the quantizer can calculate a scale that 'fits' the current 
sound/timbre best. The number of steps in the scale would be 
optional, and probably voltage controllable.


3. Expansion module for all quantizers (hopefully including A-156) 
for creating scales and arpeggio's

A module that would probably contain 24 or 32 jack outputs (+ led's) 
that output only a certain step/value/note or whatever you want to 
call it in the current tuning, through all octaves. By selecting certain 
outputs you select steps/values in a scale. E.g. in 12-tet output 1 
only outputs c, output 2 only outputs c#, output 3 only outputs d-
values etc. Each individual step/value can now be further processed, 
or certain steps can be grouped using multiples etc. 

For simple tunings (say 5 steps in an octave) it should be possible to 
'fill up' all the availabe outputs (all 24 or so) so you can even make 
different scales/pick out different combinations of notes in different 
octaves, which means you could do al kinds of arpeggio's, spanning 
perhaps 4-6 octaves.

OK, that's enough ;-)

By the way, if the list likes the idea, I bet Mr. Sethares would be 
willing to participate in the design of the timbral quantizer. He can be 
found at the alternate tunings yahoo group

Greetings,

Anton