摘要 |
PROBLEM TO BE SOLVED: To conduct an accurate exercise by classifying the portions, in which an introduction an interlude and a model singing exist, even though the phases of the accompaniment of voice signals are not the same and not conducting the comparison discrimination of the voice characteristics which are erroneous by the introduction.interlude. SOLUTION: Respective amplitudes of the signals of L and R channels of reproduced voices are detected by first and second level detection means 50 and 51 and are compared by a comparing means 52. When a model singing is started, only the model singing amplitude differences are detected. Therefore, other than the above case, a control is made so that the voice characteristics are stopped and a comparison is made for the voice characteristics with a high detection precision. |