| 摘要 |
PURPOSE: To prevent the deterioration in intelligibility of a decoded sound by adding an encoded signal indicating the residual difference signal of a selected sub-band part to a signal to be encoded indicating the time frame part of a sound pattern. CONSTITUTION: The residual quantized error of the sample of each sub-band is formed, and the vector of the residual signal of the sub-band is generated. The estimated value of the (rms) value of the residual error of the sub-band is generated, and the two-sub-bands having the maximum quantized error are selected. The selected sub-band vector is encoded by comparing it with the content of a Gaussian code book, and an index signal for identifying the content of the code book is transmitted as an assembled index signal. Thus, the residual signal of the sub-band having the maximum quantized error is transmitted, and a chirp-like effect can be reduced. Thus, the deterioration in the intelligibility of a decoded sound can be prevented. |
