Vector quantization method and speech encoding method and apparatus

A speech encoding method in which an input speech signal is divided on a time axis in terms of pre-set encoding units and encoded in terms of the pre-set encoding units, comprising the steps of:

• finding spectral components of harmonics by sinusoidal analysis of a signal derived from the input speech signal;
• and
• vector quantizing parameters derived from encoding unit-based spectral components of the harmonics as an input vector for encoding, wherein
• said step of vector quantizing includes:
• a pre-selecting step for finding a degree of similarity between the input vector and all code vectors stored in each of a plurality of codebooks and for pre-selecting a plurality of code vectors exhibiting a high degree of similarity from each of the plurality of codebooks in a fixed dimension, wherein the input vector is formed of a parameter on a frequency axis derived from the input speech signal;
• and
• an ultimate selection step of further selecting from the pre-selected plurality of code vectors selected by said pre-selecting step, a code vector that has a minimum error from the input vector in a variable dimension.

A speech encoding apparatus in which an input speech signal is divided on a time axis in terms of pre-set encoding units and encoded in terms of the pre-set encoding units, comprising:

• prediction encoding means for finding short-term prediction residuals of the input speech signal;
• and
• sinusoidal analytic encoding means including:
• vector quantization means form quantizing a parameter derived from spectral components of harmonics obtained sinusoidal analysis as an input vector, wherein said vector quantization means includes:
• pre-selecting means for finding a degree of similarity between the input vector and all code vectors stored in each of a plurality of codebooks and for pre-selecting a plurality of code vectors exhibiting a high degree of similarity from each of the plurality of codebooks in a fixed dimension, and
• an ultimate selection means for further selecting from the pre-selected plurality of code vectors from each of the plurality to codebooks selected by said preselecting means, a code vector that has a minimum error thereof from the input vector in a variable dimension.