| 摘要 |
<P>PROBLEM TO BE SOLVED: To correctly prepare a model for judging, in playing a natural musical instrument, whether the instrument is operated as indicated by a playing indication means. <P>SOLUTION: When the model is prepared, played sounds generated by actually playing the natural musical instrument are level-adjusted and retrieved through a microphone 10 and an A/D converter 6. An attack detection section 13 detects an attack by comparing, based on the sound volume of the level-adjusted played sounds and the sound volume of noises retrieved in the level-adjusted state, the retrieved played sounds with a set threshold. A spectrum creating section 12 stores, as a model, the power spectrum of the played sounds when the attack is detected, moves into creating the next model after a state of silence is detected, and removes action sounds, noises, and influences of DC components from the model. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
