| 摘要 |
PROBLEM TO BE SOLVED: To discriminate voiceless sound with high accuracy. SOLUTION: A modulation spectrum specifying section 34 specifies a modulation spectrum MS for each unit section TU of input sound VIN. An index calculation section 42 calculates an index value D1 according to intensity of a component whose modulation frequency is within a predetermined range in the modulation spectrum MS. An index calculation section 44 calculates an index value D2 according to the number of zero crossing of the input sound VIN for each unit section TU. An index calculation section 46 calculates an index value D3 according to flatness of a frequency spectrum S0 of the input sound VIN for each unit section TU. A voiceless sound determination section 54 determines whether or not the input sound VIN of each unit section TU is voiceless sound, based on the index value D1, the index value D2 and the index value D3. COPYRIGHT: (C)2009,JPO&INPIT
|
