| 摘要 |
PROBLEM TO BE SOLVED: To provide a satisfactory synthesized sound reduced in a calculation amount and in a required ROM capacity. SOLUTION: In a sound decoder, generated information of a sound vector constituted with a low pulse number is stored into a sound source information storing part 1, any one of sequence of numbers is stored into a diffused vector storing part 2 out of a random sequence of numbers, a sequence of numbers found by learning and a sequence of numbers found by knowlege, and sound source information of the sound source information storing part 1 and a diffused vector of the vector storing part 2 are convoluted by a convolution part 3 to prepare a diffused sound source vector. In a sound encoder, the diffused vector and an impusle response from a digital filter are convoluted by the convolution part 3 to prepare a new impulse response. |
