SYSTEM AND METHOD FOR AUTOMATIC TIME ALIGNMENT OF MUSIC AUDIO SIGNAL AND SONG LYRICS

摘要

PPROBLEM TO BE SOLVED: To provide a system for automatic time alignment of a music audio signal and song lyrics that can increase alignment precision over that of the past. PSOLUTION: A fricative-free segment extraction unit 4 extracts segments in which there are no fricatives from the music audio signal. An alignment unit 17 is equipped with a vocal audio model 15 that estimates phonemes corresponding to characteristic quantities for time alignment. The alignment unit 17 takes the characteristic quantities for time alignment obtained from a time-alignment characteristic quantity extraction unit 11, information related to vocal segments and non-vocal segments obtained from a vocal segment estimator 9, and a phoneme network SN as inputs and executes an alignment operation under at least the condition that no phonemes are present in non-vocal segments and the condition that no fricative phonemes are present in segments in which fricatives do not exist. PCOPYRIGHT: (C)2009,JPO&INPIT