METHOD FOR CREATING INFORMATION FOR VOICE RECOGNITION, METHOD FOR CREATING ACOUSTIC MODEL, VOICE RECOGNITION METHOD, METHOD FOR CREATING INFORMATION FOR VOICE SYNTHESIS, VOICE SYNTHESIS METHOD, APPARATUS THEREFOR, PROGRAM, AND RECORDING MEDIUM WITH PROGRAM RECORDED THEREON

摘要

PPROBLEM TO BE SOLVED: To realize a high quality simple system configuration by introducing a STRAIGHT synthesizing system to a voice synthesizing system using HMM (hidden Markov model), using parameters in the voice recognition identical to the parameters in the voice synthesis from a text. PSOLUTION: The inputted voice is subjected to short-term Fourier transform as parameters for the voice synthesis system using HMM to remove influences of fundamental frequencies to generate a STRAIGHT spectrum. The STRAIGHT spectrum is converted into STRAIGHT mel-cepstrum by logarithmic transformation and frequency extension and contraction discrete cosine transform. The HMM is learned based on the parameters of the STRAIGHT mel-cepstrum and its learning text, and the obtained HMM is used for voice recognition. For voice synthesis, voice is synthesized from the text using the learned HMM. PCOPYRIGHT: (C)2004,JPO