发明名称 |
PIECE CONNECTING TYPE VOICE SYNTHESIZER AND ITS METHOD |
摘要 |
<P>PROBLEM TO BE SOLVED: To conduct high speed voice synthesis even though the number of voice pieces of selection objects is large. <P>SOLUTION: A voice synthesis system 20 includes a voice piece DB34 and a memory 48 which stores acoustic featured value data of the voice piece candidates stored in the voice piece DB34 and waveform data of the voice piece candidates selected by a prescribed reference. A sub-cost includes an access speed cost related to the access speed to a storage device where the waveform data are stored. Each acoustic featured value data include a flag which indicates whether the waveform data are stored in the frequency information incorporated voice piece DB34 or the memory 48. The system 20 also includes a voice synthesizer 38 which selects and synthesizes a voice piece candidate that has a minimum cost computed from a plurality of sub-costs based on the acoustic featured values in the memory 48. The waveform data stored in the memory 48 are selected by a load device 46 based on appearance frequency information 44 during the testing of each piece of the voice piece DB34. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
申请公布号 |
JP2005266010(A) |
申请公布日期 |
2005.09.29 |
申请号 |
JP20040075185 |
申请日期 |
2004.03.16 |
申请人 |
ADVANCED TELECOMMUNICATION RESEARCH INSTITUTE INTERNATIONAL |
发明人 |
NISHIZAWA NOBUYUKI;KAWAI HISASHI |
分类号 |
G10L13/06 |
主分类号 |
G10L13/06 |
代理机构 |
|
代理人 |
|
主权项 |
|
地址 |
|