SPEECH SYNTHESIZER FOR GENERATING SPEECH SIGNAL WITH DESIRED INTONATION

摘要

PROBLEM TO BE SOLVED: To provide a speech synthesizer capable of generating speech signal having a desired meter.;SOLUTION: A speech synthesizer 40 includes: a text preprocessing unit 32 that preprocesses an input text and outputs a plain text, a sequence of syllables accompanied with time information, and markup tags accompanied with their positions in the sequence of syllables; a meter modelling unit 42 for generating power and phone duration for each of the syllables, and an F0 contour utilizing neutral prosody modelling; a meter reconstruction unit 44 for reconstructing the F0 contour so that a part of the F0 contour corresponding to the marked up syllable is modified according to the markup tags; and a synthesizer for synthesizing a speech signal having powers received from the meter modelling unit 42, and having the F0 contour reconstructed by the meter reconstruction unit 44.;COPYRIGHT: (C)2009,JPO&INPIT