摘要 |
PROBLEM TO BE SOLVED: To enable fluctuation components to be generated by separating voiced sound waveforms into low frequency components and high frequency components and controlling the components, and also to reduce the storage capacity of a voice synthesizer. SOLUTION: This voice synthesizer is provided with a voice segment database 11 storing voice segments, a low frequency component code book 8 being the vector quantization code book of low frequency component waveforms of sound waveforms, a high frequency component code book 9 being the vector quantization code book of high frequency component waveforms of the sound waveforms, a language processing part 2 obtaining phonemic strings and accent information from an inputted text, a segment read out part 10 reading out voice segments according to the phonemic strings, a code book referring part 12 obtaining synthetic parameters by selecting low frequency components and high frequency components from the low frequency component code book 8 and the high frequency component code book 9, and by adding them, a rythm generating part 5 generating pitch frequencies according to the accent information, and a synthesis part 14 generating a synthetic voice based on the synthetic parameters and the pitch frequencies. |