摘要 |
In an encoder and a decoder which use a code excitation linear prediction coding (CELP) method, there is provided a means which adaptively transforms on the basis of voice analysis information (LPC) code vectors outputted from the code book in which the stored codes are fixed, such as a statistic excitation code book among the code books prepared as excitation signals, thus generating high-quality reproduced voice at a low encoding speed. Further, in order to obtain similar effects, a pulse excitation code book consisting of isolated impulses is provided in addition to the adaptive excitation code book and statistic excitation code book. Hence, the statistic excitation code book and pulse excitation code book can be selectively used, and the vocal tract parameters are line spectrum pair parameters. <IMAGE> |