| 摘要 |
PURPOSE:To synthesize a voice of high quality by providing a correction filter following a synthesizing filter and obtaining and outputting the coefficient of the correction filter which minimizes the error between the output signal of the correction filter and an original voice signal. CONSTITUTION:The model signal of a residual signal generated from the original voice signal and a synthesizing filter 4 constituted by a forecast parameter extracted from the original voice signal are used to generate a synthesized signal, and this signal is allowed to pass a correction filter 5 to output a synthesized voice signal. The coefficient of the correction filter 5 which minimizes the error between the original voice signal and the synthesized voice signal is obtained. Thus, the quality of the synthesized voice is improved because the modeling distortion of the residual is reduced in the level of the synthesized voice by the correction filter 5. |
