| 摘要 |
<P>PROBLEM TO BE SOLVED: To compress waveform data by using a scale factor giving a quantization precision. <P>SOLUTION: A linear prediction sample S<SB>n-1</SB>is subtracted from an original waveform sample S<SB>n</SB>to obtain a residual sample dn. In a quantization part 62, the residual sample dn is normalized and quantized on the basis of a scale factor SF to obtain a sample L<SB>n</SB>of a k-bit residual code. The sample L<SB>n</SB>is decoded to calculate an error to the original waveform sample S<SB>n</SB>, and the scale factor SF is successively changed to perform retrieval so that this error is within an allowable quantization error. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
