摘要 |
PURPOSE:To allow natural reading aloud by providing a function to select the timbre suitable for an input sentence in accordance with the linguistic characteristics of the sentence. CONSTITUTION:The inputted sentence is divided to words in a language analysis section 1 while a word dictionary 2 registered with the information on reading of words, grammar, etc., is retrieved. The sentence is further provided with reading, accents, intonations, etc., and is converted to a rhythm symbol array embedded th the information to assign various attributes relating to synthesized voices, such as speed of speaking and timbres. A voice synthesizer 3 has 3 kinds of voice element pieces for man's voices, woman's voices and child's voices, interprets the rhythm symbol array, selects the voice element pieces corresponding to the assigned voices and outputs the synthesized voices in the assigned timbres. The natural reading aloud is executed in such a manner. |