Is there a way to convert human voice sample to formant voice? May be by extracting peculiarities from waveform? Is there a tool or a program library you can suggest?
You can try Praat (http://www.fon.hum.uva.nl/praat/). You will get f1, f2... frequencies from voice samples.