摘要 |
PROBLEM TO BE SOLVED: To code voice data efficiently with a small calculation amount and to avoid the effect of data missing. SOLUTION: A frame data clock of a prescribed length is segmented from received voice data and a spectrum envelope parameter is extracted (steps s1-s3), a position and a strength of a main pulse are estimated from the spectrum envelope parameter (step s4), the position and the strength of the main pulse are estimated from a prescribed timewise preceding block and the position and the strength of the main pulse are estimated from a prescribed timewise succeeding block with respect to the main pulse (steps s5-s6). Then the main pulse, the preceding pulse and the succeeding pulse are used as one set, and the pulse group of one set is used to generate a synthesis pulse train in one frame and the synthesis pulse train and the spectrum envelope parameter are respectively coded (steps s7-s14). |