| 摘要 |
PROBLEM TO BE SOLVED: To provide a vocalization period specifying device capable of specifying a piece of voice data for each voice period in a large voice data.SOLUTION: In a voice period estimation processing, a piece of syllable data is generated, in which various kinds of information are linked with respective periods (referred to as voice period) estimated that a voice is made corresponding to each of syllables contained in a vocalization of a voice waveform data. In particular, a voice power transition is extracted by disposing the power calculated for each time window along a time axis (S230). Each of the periods specified by a pair of vocalization start time and a vocalization termination time, which continue in order in the vocalization start time and the vocalization termination time specified according to the voice power transition, is specified as a voice period (S260). A piece of syllable data is generated by associating to each of the voice periods a lyric to each of note syllable sets which are extracted by performing a morphological analysis according to the voice period(S270). |
