摘要 |
PROBLEM TO BE SOLVED: To decrease a computational quantity for post-filter processing. SOLUTION: A device for voice decoding comprises a synthesizing filter 1 which adds a background noise signal to a signal 5 obtained by decoding a received voice signal 3 and performs a synthesizing filter process to output a composite filter processed signal 6, a filter group 8 including as elements a plurality of filters which perform a plurality of filter processes for the synthesizing filter processed signal, and a process stop unit which stops the plurality of filter processes individually stepwise in order on the basis of voiced/voiceless state information corresponding to the voice signal 3 in a process in which the voice signal is shifted to a voiceless state from a voiced state. Since the plurality of filter processes are stopped not at the same time, but individually in steps one after another, abrupt variation of a background voice is relaxed. In this case, an interpolation process is not necessary or toughly performed to eliminate or reduce the computational quantity needed for the interpolation process. COPYRIGHT: (C)2003,JPO
|