摘要 |
PURPOSE:To reproduce a smooth and natural voice employing a small amount of sound recording by recording voice by voice units which are divided with easily connectible voice and connecting voice waveforms at these divided sections when reproducing them and to reduce the connecting distortion by beforehand storing the connection positions where the connecting distortion is made small every combination of voice units. CONSTITUTION:The information, which represents are contents of voice to be reproduced, is inputted to an output voice information input section 1. Out of voice units beforehand assigned and stored in a voice signal storage memory 4, the reading order of the addresses, to which voice units that constitute a desired sentence voice when connected are assigned, are decided based on the inputted information on the content of the reproduced voice. The addresses are read in the order above by a voice signal reading section 2. These voice units are connected together by a voice connecting section 5 and the connected voice signals are converted into analog signals by a DA converting section 3 and voice is reproduced. |