this may sound like a bad joke but...

[Ronan]

I decided that the thing to do was to virtualize the pots. My preferred implementation is to read the value of the pot with a microcontroller, converting it to a digital number between 0 and 1023. Based on that number, the microcontroller would go slew a couple of control voltages to move filter freqeuncies to the desired vowel. It's a much more usable way to get "position 6 = AHHHHHH" and "position 8 = EEEEEE'" and position 3 = "AWWWWW" as well as moving seamlessly between them, which is what humans do naturally.

It makes the circuits easy - there are a couple of voltage controlled filters, a single pot, and a microcontroller, probably a 14 pin job. And that's all. The complexity is all subsumed into the microcontroller code.

If you are happy to guide the project I would be happy to do a layout and get say 10 PCB's made, and send you a few to use with writing the code, if you would like. I agree with the use of a microcontroller, if the analog vowel pedal is going to be taken to a further degree of development, a micro is a good way to do the control. It might also allow other options like "yoy" or "wah-ee-I", just for example, via a toggle switch. I have programmed Atmel chips in assembly language, and still have a programmer for smaller chips, if that is any help. A bit rusty on it now, but can pick it up all up again.

[R.G.]

I have a persistent personal problem, in that I can think of new things to build much faster than I can actually build them. Then while I'm doing the design, real life (i.e. making money so I can continue eating and living indoors  ) keeps intruding on my design time. I'd love to do that, but at the moment I have three new designs for my day job that need to be finished.

The code is straightforward, but code always needs final tinkering. It goes something like:

Declarations (tell the compiler what variables and features you'll be using)
Initialization (tell all the variables and control words where to start)
Set up time delay to space things out
Main_Loop:
Read the pot
Compute how much it's changed in the last time increment
Look up from which vowel it's coming and to which it's headed
Compute a smooth "slewing rate" between the last value and this one
Change the PWM values on the outputs to head the two voltage outputs in the right direction
Go back and do it again.

Then there's all the details of the tables of pot-to-vowels, and slewing rates to get right.

Not complicated, but it eats up the one thing I'm short on - time.

[Ronan]

OK, I understand that, no problems (re time constraints).

Assign each vowel to a pot position, that's a neat idea. My programming skills don't include calculating a slewing rate, but I could write something basic to at least test the general idea. As for the VCF's, do you think using the bandpass outputs of a pair of LM13700 state-variable filters would be suitable?

[R.G.]

Assign each vowel to a pot position, that's a neat idea. My programming skills don't include calculating a slewing rate, but I could write something basic to at least test the general idea. As for the VCF's, do you think using the bandpass outputs of a pair of LM13700 state-variable filters would be suitable?

I not only think it would be suitable, it's my preferred implementation. The SVF is reliable and repeatable as a VCF. You really want a specific voltage to be a specific frequency with this one. Musical-note/synthesize accuracy is not needed, but it has to be somewhat accurate. I would use identical filters with the high filter having one value of the timing caps, and the low filter having doubled/paralleled ones, because 2:1 cap values are hard to find. I'd use 5% (better, 2%) caps and 1% resistors. For the variable integrators, keep both integrators in a filter inside the same DIP. 

The real problem with using OTA SVFs for the filters is that they have a huge range, maybe 100:1 with no trying. For a vocal formant, you only need about 2:1, so keeping the range down to that small while at the same time getting a usable range on the control voltage, from 1.2V - 5V, needs some thought.  The LM13700 Iabc node can't go below two diode drops without shutting off, so the lower two-diode-drops of your pwm output are useless.  I actually thought about using current-output DACs for this, but PWM is just so much easier in the hardware. All the complexity is pushed into the software, though. You'll probably need to use trimmers for the range resistors to get the right range on the filters.

[digi2t]

Sorry to derail the tread somewhat, but I just came across this;

What struck me was the first squence at the beginning of the video.

You may all now return to normal programming....

[Ronan]

Thanks RG, that is more info than I expected to get me started, much appreciated.

[ayayay!]

N. B. See http://geofex.com/Article_Folders/sing-wah/sing-wah.htm from 2001.

r.g., did you ever get that monstrostity working?

or are ya still waiting for a guinea pig? cuz if ya are, i'm down!!!

but...i dunno about on vero...dino? got your ears on??


and where's ayay been lately?

I feel loved!  I've been MIA because of a new job.  Haven't picked up a soldering iron all year, sadly.  I really, REALLY miss it too.  Good news is as of yesterday, no more traveling!  So I hope to get back to this asap.

[Vince_b]

Ronan, I hope you will keep us updated about your progress. I'm realy interested by this project but I don't have the necessary skills and knowledge to work on it.

[digi2t]

I'll have to echo Jimi, Jono. You have been missed. Nice to see that all is well.

Now get out there, and SOLDER something!!

Cheers,
Dino

[Vince_b]

Back to the original topic...

if you parallel a WAH and a YOY? do you get a WHY?

i know people have tried parallel wahs, but has anyone tried this?

I achieved a "WHY" using a Crybaby and the humanizer of my Boss GT-6. I plugged them in series and I used one after the other, not both pedal at the same time.
http://dl.dropbox.com/u/67108911/WHY.mp3
 

[birt]

that actually sounded quite funny. like a desperate robot in a cartoon