Speech coder

Claim 2

A speech coder, comprising: a spectral parameter calculator obtaining a spectral parameter from an input speech signal and quantizing the spectral parameter; a divider diving M non-zero amplitude pulses of an excitation signal of the speech signal into groups, each of said groups of pulses having a number of pulses fewer than M; and an excitation quantizer calculating the positions of the pulses in each of said groups and simultaneously quantizing the amplitudes of the pulses using the spectral parameter, selecting and outputting at least one quantization candidate by evaluating distortion through addition of; (1) the evaluation value based on an adjacent group quantization candidate output value, and (2) the evaluation value based on the pertinent group quantization value, and encoding the speech signal using the selected quantization candidate.

Claim 3

The speech coder as set forth in claim 1, wherein the pulse amplitude quantizing is executed by using a plurality of codevectors which are preliminarily selected from the amplitude codebook for each group.

Claim 4

A speech coder, comprising; a spectral parameter calculator obtaining a spectral parameter from an input speech signal and quantizing the spectral parameter; a divider dividing M non-zero amplitude pulses of an excitation signal into groups of pulses, each of said groups of pulses having fewer than M pulses; and an excitation quantizer: calculating a plurality of sets of positions of the pulses in each group and, simultaneously, quantizing the amplitudes of the pulses in each group using a codebook and the spectral parameter, and selecting at least one quantization candidate by evaluating distortion through addition of the evaluation value based on an adjacent group quantization candidate output value and the evaluation value based on the pertinent group quantization value, thereby selecting a combination of a position set and a codevector for quantizing the speech signal.

Claim 5

A speech coder, comprising; a spectral parameter calculator obtaining a spectral parameter from an input speech signal and quantizing the spectral parameter; a divider dividing M non-zero amplitude pulses of an excitation signal into groups of pulses, each of said groups of pulses having fewer than M pulses; and an excitation quantizer: calculating a plurality of sets of positions of the pulses in each group and, simultaneously, quantizing the amplitudes of the pulses in each group using a codebook and the spectral parameter, and selecting at least one quantization candidate by evaluating distortion through addition of the evaluation value based on an adjacent group quantization candidate output value and the evaluation value based on the pertinent group quantization value, thereby selecting a combination of a position set and a codevector for quantizing the speech signal.

Claim 6

The speech coder as set forth in claim 3, wherein the pulse amplitude quantizing is executed by using a plurality of codevectors which are preliminarily selected from the amplitude codebook for each group.

Claim 7

A speech coder, comprising: a spectral parameter calculator obtaining a spectral parameter from an input speech signal for every determined period of time and quantizing the spectral parameter; a mode judging unit judging a mode by extracting a feature quantity from the speech signal; a divider dividing M non-zero amplitude pulses of an excitation signal into groups of fewer than M pulses; and an excitation quantizer calculating a plurality of sets of positions of the pulses in each group and, simultaneously, quantizing the amplitudes of the pulses in each group using a codebook and the spectral parameter, and selecting at least one quantization candidate by evaluating the distortion through addition of the evaluation value based on an adjacent group quantization candidate output value and the evaluation value based on the pertinent group quantization value, thereby selecting a combination of position set and a codevector for quantizing the speech signals.

Claim 8

A speech coder, comprising: a spectral parameter calculator obtaining a spectral parameter from an input speech signal for every determined period of time and quantizing the spectral parameter; a mode judging unit judging a mode by extracting a feature quantity from the speech signal; a divider dividing M non-zero amplitude pulses of an excitation signal into groups of fewer than M pulses; and an excitation quantizer calculating a plurality of sets of positions of the pulses in each group and, simultaneously, quantizing the amplitudes of the pulses in each group using a codebook and the spectral parameter, and selecting at least one quantization candidate by evaluating the distortion through addition of the evaluation value based on an adjacent group quantization candidate output value and the evaluation value based on the pertinent group quantization value, thereby selecting a combination of position set and a codevector for quantizing the speech signals.

Claim 9

The speech coder as set forth in claim 5, wherein the pulse amplitude quantizing is executed by using a plurality of codevectors which are preliminarily selected from the amplitude codebook for each group.

Claim 10

A speech coding method, comprising: dividing M non-zero amplitude pulses of an excitation signal of an input speech signal into pulse groups L pulses each, L being less than M, collectively quantizing the amplitudes of said L pulses, selecting at least one quantization candidate by evaluating distortion through addition of an evaluation value based on an adjacent group quantization candidate output value, and an evaluation value based on the pertinent group quantization value, and outputting the quantization candidate.

Claim 11

A speech coding method, comprising: dividing M non-zero amplitude pulses of an excitation signal of an input speech signal into pulse groups L pulses each, L being less than M, collectively quantizing the amplitudes of said L pulses, selecting at least one quantization candidate by evaluating distortion through addition of an evaluation value based on an adjacent group quantization candidate output value, and an evaluation value based on the pertinent group quantization value, and outputting the quantization candidate.