Eigenvoice is a technique for generating new voice characteristics with speech database of several speakers. By changing parameters extracted from several speakers in the synthesis stage, We can synthesize speech with various voice characteristics.