A speech synthesis apparatus, which can embed unchangeable additional information into synthesized speech without causing a deterioration of speech quality and restriction by bands, includes a language processing unit which generates synthesized speech generation information necessary for generating synthesized speech in accordance with a language string, a prosody generating unit which generates prosody information of speech based on the synthesized speech generation information, and a waveform generating unit which synthesizes speech based on the prosody information, in which the prosody generating unit embed code information as watermark information in the prosody information of a segment having a predetermined time duration within a phoneme length including a phoneme boundary.