PURPOSE: To offer the voice rule synthesis device which generates the power of a phoneme closer to a natural voice that a person speaks.
CONSTITUTION: When information on a phoneme, a rhythm, a language, etc., is inputted from an input part to a rhythm parameter generation part 2, the phoneme continuation length setting part 21 of a rhythm parameter generation part 2 sets the continuation length of the phoneme by referring to a phoneme continuation length setting rule dictionary 31, a fundamental frequency setting part 22 sets a fundamental frequency by referring to a fundamental frequency setting rule dictionary 32, and a power setting part 23 sets the power by referring to the power setting rule dictionary 33. Respective parameters generated by the rhythm parameter generation part 2 are supplied to a voice parameter generation part and processing for connection, compression, expansion, etc., is carried out according to the voice parameter connection rule in a voice parameter connection rule dictionary to generate voice parameters. The generated voice parameters are supplied to a voice synthesis part to synthesize the voice, which is outputted by an output part.
JPH0934492 | PITCH PATTERN CONTROL METHOD |
JPH1078795 | SPEECH SYNTHESIZING DEVICE |
JPS6448099 | VOICE RECOGNIZING/ANSWERING DEVICE |
UMIKI NOBUYOSHI
KOSAKA YOSHINORI
JPH03293398A | 1991-12-25 |