| Speech signal processing has been studied for decades and the speech synthesis technology has been greatly improved both on the accuracy and naturalness. However, the intelligent of synthesized speech is required since the development of technology. If speech signal processing is combining with emotional compute and make it possible for the computer to express emotional speech, the quality of speech synthesis will improved greatly.In order to synthesis emotional speech, the analysis approach is as follows: extracting the emotional characteristics from the expressive speech signal→mapping characteristic parameters to the different classes of emotion→synthesis the emotional speech with the relevant characteristic parameters.In the beginning of this paper, the research background and history of emotional speech processing are reviewed. After that, some methods of classifying emotion spaces are introduced and the emotional classification used in this paper is presented based on which we have built an emotional speech corpus for the experiment. Through analysis the emotional speech in our corpus, we extracted the prosody feature and voice quality parameters and compared the affective effect of those two kinds of parameters. Then we used an improved Fujisaki model to describe the fundamental frequency and designed a rule for the parameter's setting. In the progress of speech synthesis, a pitch synchronous method of the initial portion and final portion is applied. Finally, modify the affective features using PSOLA to synthesis the emotional speech.The main innovative points of this paper are as follows:1) Modified the tone command of Fujisaki model to make it suitable for Mandarin and combined the model with emotional feature parameters in order to enhance the model's affective expressive ability;2) Proposed a normalized tone model to represent the basic type of tones and tone sandhi and proved its validity by simulation experiment;3) Proposed a pitch synchronous pattern of finall portion based concatenative method to synthesis Mandarin... |