发明名称 SOUND SOURCE DATA GENERATING METHOD OF VOICE SYNTHESIZER
摘要 PURPOSE:To find the accurate peak position of even a waveform, which has a large frequency of zero crossing and is large in waveform attenuation in pitch, at all times. CONSTITUTION:An original waveform is divided to detect pitch sections. Then the residue of all frames is calculated to extract a residue peak position. The final point of the residue is extracted as the peak position of a section. Further, mutual relations with adjacent sections are calculated from the section in the front-rear direction of a time base. Peak positions are determined one after another by calculation. After the peak positions are determined, adjacent intervals are put in a peak sequence. The peak sequence is averaged to determine a peak and the final peak position is determined. Then the peak position is stored in a file while the phase is matched.
申请公布号 JPH04253100(A) 申请公布日期 1992.09.08
申请号 JP19900408727 申请日期 1990.12.28
申请人 MEIDENSHA CORP 发明人 ISHIDA KIYOSHI;SAWADA YOSHIMASA
分类号 G10L19/08;G10L11/06 主分类号 G10L19/08
代理机构 代理人
主权项
地址
您可能感兴趣的专利