Abstract: A phonetically-driven speech synthesizer and method substantially entirely embodied in a programmed microprocessor. Information indicative of control parameters for each of a plurality of phonemes is stored in a phoneme parameter matrix and selectable by phoneme code. Time-invariant programming controls operation during successive operating cycles of equal time duration to obtain parameter information in a predetermined sequence for each selected phoneme, update control signals as a function of such parameter information, and operate a lattice filter vocal tract as a function of updated control signals to generate phonetic sounds. Separate sources of vocal and fricative sounds comprise corresponding look-up tables which are accessed are required during each operating cycle.