A SPEECH FORMANT SYNTHESIZER BASED ON HARMONIC + RANDOM FORMANT-WAVEFORMS REPRESENTATIONS

This paper describes a new type of speech synthesizer: a parametric-concatenation (PACO) speech synthesizer, which is suitable both for formant synthesis and concatenation synthesis. This synthesizer is based on a hybrid quasi-harmonic and random formant-waveforms model of the speech signal. The synthesizer can be controlled by acoustic parameters: formant parameters and voice source parameters expressed in frequency domain. These acoustic parameters are converted into sinusoidal and formant waveforms parameters. The keypoint of this method is that spectral amplitudes are set according to a parallel formant model (both on sinusoidal waveforms and random formant-waveforms), whereas the spectral phases are set according to a serial formant model for sinusoidal waveforms (and are randomly distributed for formant waveforms). This approach avoids the phase interference problems inherent to parallel synthesis, while keeping the advantage of formant amplitudes control. An automatic analysis-synthesis system is also proposed for segments coding. Our model has been successfully implemented both as a formant synthesis system and in a concatenation synthesis Text-To-Speech system.