MOTM

I'ts a bit too much to cut and paste a bunch of responses, so I'll just
point some things out.

I of course have to use a bunch of filtering in a lot of stuff I do.  But
the ironic problem you get into is the more powerful, flexible, and more
bands a filter has, the harder it is to acheive what you want.  It's like
additive synthesis.  If you have 128 harmonics to play with, then changing
the level of one of them may barely even be audible.  Then you have to
adjust more, but that takes more time.  And if you are trying to acheive a
certain imitative sound, say a violin, there is no way you can just listen
to a violin with your ears and then know you need to set band #38 to -21dB
for example.  In my case with my latest Nord violin sounds, I needed to make
adjustments to my 42 band filter bank, then digitize some notes from the
Nord, take a fourier transform to visually see the levels of the harmonics,
compare them to the levels of a real violin, see where the differences are,
then go back make more adjustments and so on back and forth.  It's like
tedious manual resynthesis.  And then you want voltage controled freq, level
and bandwidth ontop of that?  You'll need to spend days or weeks on one
sound.  And if you're NOT trying to imitate some complex acoustic
instrument, then trying to shape your non-imitative sound with such detail
doesn't make sense because there isn't really any detailed specific target
in your mind in the first place.

As a result, in a lot of my sounds I've tried to come up with clever tricks,
which is probably why even on something like a $600 Alesis Ion some people
were suspicious I might be using samples somehow in my demos.  You start to
learn how to use phase shifters and flangers for complex filtering, and what
kinds of regular LP/BP/HP configurations to use in addition to EQ ontop of
that stuff.

Somebody mentioned Teisco having a filter bank.  So did the Polymoog,
Spectralis, and a number of modular synths had EQ and formant filter
modules.  But if you listen to the history of synth recordings, the synth
has been so underutilized except by two or three people, that it doesn't
seem like people want to spend the time and effort in learning and doing
anything groundbreaking or difficult.  And as pointed out, damn sampling
killed off a lot of people's motivation to do their own sounds.  It's easier
for them to work with a known entity.

But as a synthesist, you basically need to rely on outboard gear anyway, and
if that's the case, then there currently isn't anything that can't be done.
In fact SOS magazine just ran an article describing  what's available in
Spectral Multi-band plug-ins.  I have Spectral Delay myself.  If you want a
1000 band filter bank where each band is capable of 0dB to full level to
give you virtually any shape you want, then that's available.  Or even
vitural analog synths like the Nord or Reaktor will allow you to build up
huge filter banks for processing your modular.  Or just use multiple pieces
of gear and effects to fake the sound you want the best you can.

Note that I just noticed that Samson has reduced the price of their D-3500
to a great price of $200.  It's a dual 15 band digital parametric EQ with
patch storage, and capable of +16 to -48 dB of boost/cut (plus you can
overlap bands for more cut/boost).  Just another example of a potentially
sophisticated tone shaping piece of gear.

As for the attitude that if you want a clarinet sound just hire a clarinet
player, besides samples, this is another reason synthesis is dead.  The
synthesizer was supposed to be a device to create any sound you want, but
nobody uses it that way.

And about allowing the freqs of a filterbank to track the keyboard.  When it
comes to doing human vocal sounds that usually works best.  Anytime I've
tried to keep filters fixed and create separate patches for male and females
voices, the fact that all the different note pitches have the same filtering
leads to an annoying strong boxy or vocoder type sound.  I seem to have to
go with a single patch that morphs from male to female as I play up the
keyboard to get a more natural sound.  In real life a choir of singles have
different resonances which combine and smooth each other out anyway.

-Elhardt