摘要 |
PROBLEM TO BE SOLVED: To encode speech to a lower bit rate while maintaining sound quality corresponding to a quantization width of 5 bits, without being influenced by a time variation of statistic nature of speech data. SOLUTION: The speech encoding unit 10 comprises; a primary encoding unit 11 for encoding a speech signal to an adaptive pulse code modulation (ADPCM) code; an interleaver 12 for rearranging a bit string which is encoded to the ADPCM code by the primary encoding unit 11; a comparator switching unit 13 for calculating each statistic of two bit sequences which is input from the primary encoding unit 11 and the interleaver 12, and for selecting and outputting one input sequence based on a comparison result; and a secondary encoding unit 14 for encoding and outputting an output data of the comparator switching unit 13. Entropy or the like is used for the statistic and the sequence with low entropy is selected. As it is selected whether the interleaver 12 is used or not, according to an output of the primary encoding unit 11 which reflects the time variation of the speech data, the encoding is not influenced by the time variation of the statistic nature of the speech data. COPYRIGHT: (C)2007,JPO&INPIT
|