[sdiy] Synth UI [was: Modular - sound or song]

Mike profpep at hotmail.com
Thu Apr 9 22:36:37 CEST 2009


I tend to a very split way of thinking about this. Sometimes I go very
'digital purist' and write things in CSound. So far, I've never found
anything I couln't write, but the rub is that you have to know what you want
to do. I would expect that writing for the Chameleon etc is much the same.

I agree with Seb about the UI aspect, there are Digital UI's that would be
impossible in the analogue domain, and pretty much vice versa. The 'Analogue
Controls plus Digital Implentation' is as valid as any other - if makes
sound or controls sound and is easy to manipulate. I prefer the encoder plus
display model if possible, because of the 'instant recall' abilities, a
taste I gained when working with DMX controlled lighting desks, where patch
control is essential. There were some good ideas in that field too, one I
saw was a cylinder half embedded in the panel on which you wrote your cues
and master fader allocations. Rotating it to the a position cue made them
active.
Cheap, flexible and easy. A big LCD would do the same, but at a scary price.

There is one last point in the analogue versus digital debate that no-one
has yet brought up. Especially when working with dynamical systems or
fractal type patches, in the digital domain, there will always remain the
spectre of round off errors. A very long time ago, I got a horrible flaming
from some American academics on a now extinct dynamical systems group, by
postulating that some of their fractals were the product of round off errors
in the programming languages they used. I modelled some of them on an
analogue computer, and never got the chaotic states they were showing. They
were using ranges below 10E-6, and the floating point on their PC's was
throwing up all kinds of artfacts. I was recued by the cavalry from another
American university who did some very long word models on their
supercomputer, and got the same results as me.

On the another track, a simple analogue device can often be way cheaper than
digital for the same job. Back in that era, I built a device that did a
wierd coefficient multiply run, at virtually real time speeds, that was
pulled back into a PC by a National Instruments A->D card. It enabled the
guy doing the research to twiddle the coefficients to find null points in
his model quite simply, interacting his human inutition, with instant
modelling of the results. Nowadays a decent DSP would do the job with ease,
but at the time this was a very economical solution.

I ageree wholeheartely about a synth being a tool, whatever gets the job
done for you is appropriate. The same is true in a the world of engineering
and in craft work. Sometimes a power tool is appropriate, sometimes a hand
tool. A hand plane produces a better finish than a power plane, but is
tedious and difficult when you need to thin down a long edge. A miller is
far too awkward to set up to do a simple clean edge, but I'd be stupid to
try and cut/file a series of long slots if a milling machine was there.
Sometimes you haven't got what you need, and you make what you have fit -
you carefully multitrack a modular, if you want chords,  or stack a load of
effects onto your digital keyboard to get the dirty sound you like.

I love hybrids like Monowave, MIDIbox SID and AVRx too.

A last thought: there are current thories that certain classes of problems
can only be solved by neural networks, essentially analogue devices. Perhaps
the hybrid/mixed world is the real one, and for users it is just a balance
between the two

http://www.eetimes.com/story/OEG19981103S0017


(sorry for rambling, went rather stream-of-conciousness there.

||\/||ike




More information about the Synth-diy mailing list