Individual formant shifting

[barrabasito]

I have tried several vocal synths, but none of them reaches a real voice sound quality, so i decided to sample my own voice and alter it in real time. I sampled an open vowel (a spanish 'a') in order to start with a broad spectrum. I found out that i can get decent spanish 'e' and 'i' varying the Q factor of a notch filter at a specific frequency.
But i was no lucky when trying to get 'o' and 'u'. It seems the relative position of F2 respect F1 is an important feature that distinguish those vowels, but the formant shifters i know about move all formants as a whole, not independently. So I wonder if individual formant shifting has been tackled for real time audio processing.