| 摘要 |
PURPOSE: To easily decode a voice signal of high quality, relating to an encoding and decoding of the voice signal. CONSTITUTION: Parameter making by a voice generation model is executed, based not only on voice signal to be encoded but also the voice signal to be decoded, that is, synthesized voice signal. The parametric display 207 of synthesized signals is compared with the parametric display 203 of the original voice signal, and an encoding function is controlled by the difference. First, the parameter making 205 by the voice generation model used for encoding is executed, based on the voice signal to be decoded. Then, the parameter value formed from the respective synthesized voice signal is compared with the respective parameter values 203 which is calculated from the voice signals to be encoded. Measurement of well-known distances can be used in the comparing operation. The encoding function is controlled inside a shaping block 202, so as to make the difference indicated by the distance measurement as small as possible. |
