Vocal Filter

[Scott Bernardi]

I've been looking at your Vocal Filter ideas with some interest, because 
I've been kicking around ideas for a filter bank that could be adjusted 
to do formants, and your voltage controlled formant CV generator looks 
like it will fit right in.

I've got a couple of schematics at 
http://members.home.net/sbernardi/elec/og2/filter_bank.gif and 
http://members.home.net/sbernardi/elec/og2/filter_bank_vc.gif.  The 
bandpass filters will be standard voltage controlled state variable 
types, but I've got them configured like parametric equalizers, where 
you can cut or boost the frequencies within the bands.  The amount of 
input signal mixed in with the filtered output can also be adjusted.
Each BP channel would have its own VC input and initial freq and Q pots, 
but then there would also be VC inputs (inverting and non-inverting) and 
initial freq pot that drives all the channels.  Your formant control 
voltages could be just an additional VC input to the independent channels.
Setting the filters at a high-Q cut (i.e., notch) with staggered fc's, 
then sweeping the fc's with an LFO gives a phaser effect.

Some questions about your schematics:
On sheet 4, the bottom opamp - shouldn't that have another input of Vin 
through a 100K?

Some very interesting things to do with vocal type sounds would be the 
compound formants like "a" (eh-ee), "ow" (ah-oo), "oy" (oh-ee), "i" 
(ah-ee).  Can you imagine your synth saying "oy, oy, ow!" ;?)     A 
smooth transition between formants responding to changes in the control 
voltage could be accomplished by placing lag/portamento RC circuits 
between the formant CV's and the bandpass CV inputs.

If I understand the circuits correctly, the output of the formant CV's 
are not going to be smooth transitions but are going to have steps in 
them due to the diodes switching in.  What are they going to look like 
responding to say a 0 to -9v triangle LFO?

'Scuse all the questions, but this circuit has me very interested.

I guess the last question is., will this sound vocal enough without also 
adjusting the Q and gains of the formant peaks, or should I just do a 
vocoder (a major undertaking). I remember building  a variable slope 
filter from Electronotes in my original synth and being very 
disappointed with the results.

patchell wrote:

>     Here are some sketches of what I have done so far.
> 
> http://www.oldcrows.net/~patchell/vocalfilter/vocalfilter.html
> 
>     Looks like that will be it until next week.  Didn't get as much work
> in as I liked in the last 11 days, but mental health is important
> also...
> 
> --
>  -Jim
> ------------------------------------------------
> * Visit:http://www.silcom.com/~patchell/
> *-----------------------------------------------
> *I'm sure glad Merry Christmas comes just once a year
> * -Yogi Yorgensen
> ------------------------------------------------
> 
> 
> 
> 

-- 
Scott Bernardi
sbernardi at home.net