| 摘要 |
PROBLEM TO BE SOLVED: To reduce deterioration in decoded voice quality caused by discontinuity of a filter coefficient by decoding a characteristic parameter showing a spectrum envelope characteristic of a composite signal with a value smoothened in the time direction. SOLUTION: A parameter decode circuit 54 outputs the filter coefficient and an RMS obtained from a signal code line inputted from an input terminal 52 to smoothing circuits 64, 66. The smoothing circuit 64 smoothens the filter coefficient to output it to a synthetic circuit 68. However, when the matter that the signal code line isn't transmitted with a DTX decision code inputted from the input terminal 50 is shown, the filter coefficient of the former frame is used. The synthetic circuit 68 inputs an exciting signal from a mixer circuit 61 to the filter constituted of the filter coefficient from the smoothing circuit 64, and decodes the signal to output it from an output terminal 70. When a silence voice section is decoded, after the intermittently transmitted filter coefficient is smoothing processed like the RMS, a synthetic filter is used, and the discontinuous change of the filter coefficient is prevented, and the decoded voice quality is improved. |
