MOTM

I agree with the idea of using a non-harmonically related spacing, 
although unless you individually tuned the filters to be exactly on 
frequency the chances are that the tolerances of the parts would 
take care of it.  I also was thinking about costs.  The key 
component in the modular vocoder idea is the VCA.  These certainly 
wouldn't have to be up to the MOTM standard, but they should still 
be of reasonably good quality.  I think the envelope follwers could 
be done inexpensively since the frequency range they cover would be 
narrow.  
I also like the modular vocoder idea, but I'm trying to put 
additional functions into this module.  I'm really interested in 
trying to break sounds apart and then reconstitute them.  I was 
thinking of looking at the area of vocal formants for the strongest 
concentration of filters, but also looking at the resonances in 
stringed and wind instruments to try to determine the most optimal 
filter frequencies. Making the Q variable somewhere along the line 
for stronger resonances is also a possibility.
Of course the larger the number of filters the greater the 
possibility of noise.  I have some really good Ashley EQs in my 
studio.  Maybe I'll take one of those apart and look at the 
filters.  They are really outstanding. 
What would be really cool is if I could somehow do a Fourier 
transform in analog circuitry, extract the spectrum of a sound in 
real-time, then reconstitute it with the spectrum modified or 
applied to another signal.  Maybe a pitch follower and a series of 
VCBPFs would work.  Hmmm. :/