摘要 |
PURPOSE: A phonetic data compressing method for a voice synthesis system is provided to systematically reduce triphone according to the use frequency of each triphone. CONSTITUTION: A voice composition system generates plural recording files in separate triphone file(1100). The system compresses the individual triphone file in an ADPCM(Adaptive Differential Pulse Coded Modulation) mode(1200). The system deletes a quantization file corresponding to the individual triphone file having the use frequency lower than a set frequency(1400). The system stores the quantization file at a voice DB(1500). |