摘要 |
PURPOSE:To obtain a synthesized voice small in deterioration in quality even when the pitch is varied in voice synthesis based upon a vocal chord sound source wave model. CONSTITUTION:The voice synthesizing device has a vocal chord sound source wave code book 4 wherein vocal chord sound source waves are classified by plural feature categories of time lengths of sound source waves, spectra by frequency bands, etc., and relative distances of respective characteristic values are made to correspond to the large/small relation of codes. A voice analyzing means 1 takes a vocal chord glottis type analysis and outputs the analytic result to a sound source wave time information control means 2. The means 2 varies only a code regarding the time length of a sound source characteristics in the inputted analytic result or both a code regarding a spectrum and a code regarding the time length according to a target pitch and inputs the variation result to a speech synthesizing means 3, which generates a sound source wave actually to synthesize the voice. |