make synthesis and sample basis speech, by looking at spectral qualities of https://en.wikipedia.org/wiki/Help:IPA

[ttm]

input -> process -> speech

a sample based method (for music) to obtain the process is to record the vogals and each consonant with 'a'. E.g.: aeiou bacadafaga.....

another method to obtain the process is to hold the a spectrum shape for each vogal and noise profiles for each consonant.

the input should include some sound to use as the voice profile and the text to be spoken.