This document discusses the history and methods of speech synthesis. It notes that the first speech synthesizer was created in 1791 and was able to produce 19 consonants and 5 vowels by modulating the vibration of reeds through a leather tube. There are three main methods of speech synthesis: waveform coding, analysis-synthesis, and synthesis by rule. Waveform coding involves storing recorded speech waves and selecting/connecting units to synthesize speech. Analysis-synthesis breaks sounds down into parameters and rules for re-synthesis. Synthesis by rule generates new waves without human recordings. Modern methods like TD-PSOLA and the Harmonic/Noise Model divide the speech spectrum and concatenate frames to alter prosody.