To provide a speech synthesis system capable of improving the sound quality of a synthetic speech and effectively utilizing a recorded original speech.

The speech synthesis system is constituted by being provided with: a database recording device 120 for inputting the original speech and recording waveform data obtained by removing speech components of a glottal sound source from a section corresponding to a voiceless affricate and a section corresponding to the voiceless affricate in speech pieces included in the original speech in a database 110; and a speech synthesizer 130 for extracting the waveform data from the database 110 and synthesizing a voice corresponding to read information.