| 摘要 |
PURPOSE: To enable voice synthesization much closer to a natural voice by containing a sampling cycle in compressed data and synthesizing voices in the cycle equal to that sampling cycle while referring to the sampling cycle at the time of voice synthesization. CONSTITUTION: An A/D converter 1 converts an input analog natural voice waveform to a digital waveform and resamples it so that amplitude values at the starting end and terminal of a basic frequency can be prescribed values. Besides, a frequency at the time of resampling is detected by a frequency detector 2, the cycle converted from that frequency is contained in data to be code- compressed by an encoder 3, and waveform information is outputted. At the time of synthesization, sampling cycle information at the head of basic cycle waveform information, for example, is transmitted to a clock generator 6 by a decoder 5. Further, the waveform information is transmitted to a D/A converter 7. Corresponding to that sampling cycle information, the device 6 generates a clock. At the device 7, the waveform is extended and converted to the analog voice waveform by the clock in the same cycle as the resampling cycle. Thus, the voices can be synthesized in the same cycle at the time of encoding. |
