| 摘要 |
PROBLEM TO BE SOLVED: To provide a singing scoring device scoring singing, even for a song without melody. SOLUTION: A user's singing data obtained by sampling singing notes sung by a user along with Karaoke accompaniment sound generated by reproduction of accompaniment data, and model singing data which is read from a Karaoke data memory 13 in synchronization with reproduction of the accompaniment data, are divided by frames corresponding to the predetermined number of data. A voice feature amount MFCC of a model singing voice and a voice feature amount MFCC of user singing voice are extracted from model singing data and a user singing data in the divided frames, respectively. Singing is scored by determining a similarity degree of the voice feature amount MFCC of the extracted model singing voice and the voice feature amount MFCC of the user's singing voice. That is, as it is scored as to whether or not the singer correctly sings words of the music, even a song without melody which is called "rap" can be scored. COPYRIGHT: (C)2010,JPO&INPIT |
