SPEECH ENCODING DEVICE

摘要

Speech quality is improved by suppressing a quantization error. A code candidate storage section (11) obtains a code corresponding to a sample value of a speech signal by storing all the combinations of the code candidates in the vicinity of the sample value. A local decoder (12) decodes the codes stored in the code candidate storage section (11) and generates a reproduction signal. An error evaluation section (13) calculates a sum of squares of differences between input sample values and reproduction signals for each code candidate, detects a code candidate having the minimum sum of squares which minimizes the quantization error, and outputs a code in the code candidate detected.