Addition of Prosodic Information to Synthesized Speech Imitated Human Text-Independent

Speech is an important method of human communication. So, research about speech is expected and conducted as man-machine interface. Until now, quality of synthesized speech become higher, but research from the viewpoint of natural speech isn’t conducted. So, this paper proposes the imitation system as an example of man-machine interface system focussing the accent on natural speech. First, this paper describes human speech method and pitch pattern. And intonation structures of sentence is showed by matching of pitch pattern and speech segmentation. This system synthesizes speech on the basis of intonation structures. It uses commercially available LSI synthesis speech, so robots and cars can equip with this system. As a result of experiments, more than 90% speaker-independent speech can be imitated. And the effectiveness of this system have confirmed. And this system shows captured speech in a visual form. So, this system can be applied to the speech interface for hearing-impaired man.