This project implements the basic functionality of TI's Speak & Spell toy introduced in 1978. My S&S only implements the vowel sounds (short and long A, E, I, O, U). I plan to add consonants in the future because those require different signal processing techniques.
The S&S synthesizes speech using filters generated from an autocorrelation of a speech signal. The filters are then convolved with a periodic impulse train to get a synthesized vowel audio signal.
Running genFilters.m will create a filter for every audio signal in the audio directory. Each resulting audio signal will then play.
This project demonstrates my understanding of correlation, filtering, and using efficient methods to solve equations (see Toepsolve.m where I solve Yule-Walker equations using a Toeplitz matrix).