Vocoder

So, I've been learning the basics of PD. It's a MAX/MSP type program, written originally by Miller Puckette (who also wrote max), and is now open source (ie. free). It has video/3d capabilites as well (also available for free). The documentation is a little arcane, and it's a little geeky, but I'm really intrigued by the possibilites of making procedural sounds and images and thought this would be a cool program to check out -- I've always meant to learn max, but it's so expensive and the free trial only lasts 30 days so....

Sound file: My singing keyboard
The PD Patch (it's a mess, sorry..)

I thought it would be cool to try to make a vocoder as my first experiment. It's pretty glitchy, but I'm happy with it for now. I basically looked into how the real ones worked and figure out how to do it in PD. Figuring out how many eq bands to break up the sounds into is the hardest part (as well as how wide those frequencies should be) -- I tried to look at charts of human voice frequencies and put the most detail in those areas. I mean, I'm sure if I would have used like 256 bands it would sound really detailed, but that's not really the point right?

I was inspired by these two videos, which are both extremely awesome:

Stevie Wonder
Giorgio Moroder