JP2001228888A - Speech-encoding device, speech decoding device and code word-arraying method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem where the reproduction quality of input speeches is deteriorated drastically when a speech decoding device makes erroneous recognition mode information by superposition of a transmission error on a speech code, while a distortion minimization section 7 of a speech encoding device selects optimum code information to minimize the power of an auditory sense weighting difference signal. SOLUTION: The storage sequence of the gain code words of gain code books 52 and 66 (or 57 and 71) is permuted, in correspondence with the sequence of the evaluation values related to the gain code words of other gain code books 57 and 71 (or 52 and 66).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio encoding apparatus for compressing the information amount of a digital audio signal, and an audio decoding apparatus for decoding an audio code generated by the audio encoding apparatus and reproducing the digital audio signal. Codeword arrangement method for updating the storage order of codewords in a vector codebook used by a device, its speech encoding device or speech decoding device and improving resistance to bit errors superimposed on speech code It is.

[0002]

2. Description of the Related Art Many conventional speech coding apparatuses adopt a configuration in which an input speech is divided into spectral envelope information and a sound source, and each is encoded in frame units to generate a speech code. On the other hand, the conventional speech decoding device adopts a configuration in which a decoded speech is generated by decoding the speech code and synthesizing the spectrum envelope information and the sound source by a synthesis filter. Also, in order to improve the quality of both the speech signal and the background noise signal having various aspects, a method of preparing a plurality of encoding modes and performing encoding while switching the encoding mode (multi-mode encoding method) is proposed. Some are employed.

FIG. 15 shows, for example, a document “H. Tasak”.
i, "High level description
of Mitsubishi 4-kbit / s s
"peech coder", ITU Telecomm
UNICATION STANDARDIZATION
Sector, Study Group 16, Qu
estion 19-21 / 16 Rapporteu
r Meeting, no. AC-99-016 (1
FIG. 9 is a configuration diagram showing a conventional speech encoding device shown in FIG.

In FIG. 1, reference numeral 1 denotes a noise suppression process for suppressing background noise superimposed on input speech,
The pre-processing unit 2 that executes the low-frequency rejection filter processing for cutting the DC component of the input voice analyzes the input voice that has been pre-processed by the pre-processing unit 1, and analyzes a line spectrum pair (hereinafter, referred to as spectral envelope information of the voice) , LSP), encodes the LSP obtained by the spectrum analysis unit 2, outputs the LSP code to the multiplexing unit 20, quantizes the LSP, and performs quantization. LSP (same as the result of decoding the LSP code) is converted into a filter coefficient (linear prediction coefficient) of the synthesis filter 4,
The filter coefficient is combined with the synthesis filter 4 and the auditory weighting unit 6
Is a spectrum encoding unit that outputs the data to

[0005] Reference numeral 4 denotes a synthesis filter for executing a filtering process on the tentative sound source selected by the changeover switch 19 using the filter coefficient output from the spectrum encoding unit 3 to generate a tentative synthesized sound, and 5 denotes a synthesis filter. A subtracter for outputting a difference signal between the synthesized speech generated by the preprocessor 4 and the input speech after the preprocessing by the preprocessor 1; and 6 calculates an auditory weighting filter coefficient based on the filter coefficient output by the spectrum encoder 3. A hearing weighting unit that performs a hearing weighting filter process on the difference signal output by the subtractor 5 using the hearing weighting filter coefficient, and outputs a hearing weighted difference signal.

Reference numeral 7 denotes an index (gain code, drive excitation code, adaptive excitation code) and an encoding mode for calculating the power of the perceptual weighting difference signal output from the perceptual weighting section 6 and minimizing the power. Distortion minimizing units 8 and 9 for sequentially updating mode information have codebooks for outputting codewords corresponding to indexes updated by distortion minimizing unit 7, and a sound source for generating a temporary sound source from the codewords It is a decoding unit.

An adaptive code vector 10 stores a predetermined length of a past excitation and, when receiving an adaptive excitation code from the distortion minimizing unit 7, periodically repeats the past excitation corresponding to the adaptive excitation code. An adaptive excitation codebook 11 that outputs a non-noise driving code vector, which is a non-noise plurality of time-series vectors, receives a driving excitation code from the distortion minimizing unit 7 and outputs a driving code corresponding to the driving excitation code. A driving excitation codebook 12 that outputs a vector stores a code word (a word indicating a gain value) related to the gain, and receives a gain code from the distortion minimizing unit 7 and outputs a gain value corresponding to the gain code. Codebook, 13 is a multiplier for multiplying the adaptive code vector output from adaptive excitation codebook 10 by the gain value output from gain codebook 12, and 14 is output from gain codebook 12. Multiplier gain value excitation code book 11 is multiplied by the driving code vector outputted, 15 adds the multiplication result of the multiplication result and the multiplier 14 of the multiplier 13,
The adder outputs the result of the addition (temporary sound source).

Reference numeral 16 stores a driving code vector, which is a plurality of noise-like time-series vectors, and upon receiving a driving excitation code from the distortion minimizing unit 7, outputs a driving code vector corresponding to the driving excitation code. A code book 17 stores a code word (a word indicating a gain value) relating to gain, and receives a gain code from the distortion minimizing unit 7 and outputs a gain value corresponding to the gain code. The gain value output by the codebook 17 is
Is a multiplier that multiplies the driving code vector output by the controller and outputs the multiplication result (temporary sound source).

When the mode information 19 is received from the distortion minimizing unit 7, the temporary source selected by the source decoding unit 8 or the temporary source output by the source decoding unit 9 is selected according to the mode information. A changeover switch for giving the selected tentative sound source to the synthesis filter 4. A multiplexing unit 20 multiplexes the LSP code coded by the spectrum coding unit 3 with the index and mode information updated by the distortion minimizing unit 7 to generate a voice code. And a multiplexing unit that outputs the speech code.

FIG. 16 is a block diagram showing a conventional speech decoding apparatus disclosed in the above-mentioned literature. In the figure, reference numeral 21 denotes an LSP code multiplexed by the speech encoding apparatus, an index and mode information. The separation units 22, 23 have a codebook that outputs a codeword corresponding to the index separated by the separation unit 21, and are excitation excitation decoding units that generate an excitation from the codeword.

Reference numeral 24 stores a past excitation of a predetermined length, and upon receiving an adaptive excitation code from the separation unit 21, outputs an adaptive code vector which is a time series vector that periodically repeats the past excitation corresponding to the adaptive excitation code. Adaptive excitation codebook, 2
Reference numeral 5 denotes a driving excitation codebook that stores a driving code vector that is a plurality of non-noise time-series vectors and, when receiving a driving excitation code from the separation unit 21, outputs a driving code vector corresponding to the driving excitation code. Is a gain codebook that stores a code word (a word indicating a gain value) related to the gain and receives a gain code from the separation unit 21 and outputs a gain value corresponding to the gain code. A multiplier for multiplying the gain value by the adaptive code vector output from the adaptive excitation codebook 24; a multiplier 28 for multiplying the gain value output by the gain codebook 26 by the driving code vector output by the driving excitation codebook 25; Is an adder that adds the multiplication result of the multiplier 27 and the multiplication result of the multiplier 28 and outputs the addition result (temporary sound source).

Reference numeral 30 denotes a driving excitation codebook which stores a plurality of driving code vectors, which are noise-like time series vectors, and outputs a driving code vector corresponding to the driving excitation code when receiving the driving excitation code from the separating unit 21. , 31 store a code word relating to the gain (word indicating the gain value).
, A gain codebook that outputs a gain value corresponding to the gain code, and 32 multiplies the gain value output by the gain codebook 31 with the driving code vector output by the driving excitation codebook 30, This is a multiplier that outputs a multiplication result (temporary sound source).

Upon receiving the mode information from the separating section 21, the 33 selects a temporary excitation output from the excitation decoding section 22 or a temporary excitation output from the excitation decoding section 23 in accordance with the mode information. Temporary sound source is synthesized by filter 35
, And a switch 34 that outputs L
The SP code is decoded, and the decoded result is output to the synthesis filter 35.
, And outputs the filter coefficients to the synthesis filter 35 and the post-processing unit 36. The spectrum decoding unit 35 uses the filter coefficients output from the spectrum decoding unit 34 to perform switching. A synthesis filter that performs a filtering process on the tentative sound source selected by the switch 33 to generate a tentative synthesized sound, and a synthesis filter 36 generated by the synthesis filter 35 based on the filter coefficient output from the spectrum decoding unit 34 and the like. This is a post-processing unit that performs post-processing such as voice emphasis processing on the sound and outputs a reproduction result (output sound) of the input sound.

Next, the operation will be described. The conventional speech encoding device and speech decoding device require about 5 to 50 ms for one.
The processing is executed for each frame as a frame.

First, upon receiving an input speech, the pre-processing unit 1 of the speech encoding device executes a noise suppression process for suppressing background noise superimposed on the input speech, and cuts a DC component of the input speech. To perform low-pass rejection filter processing. When the pre-processing unit 1 performs pre-processing on the input voice, the spectrum analysis unit 2 analyzes the input voice after the pre-processing and obtains LSP which is the spectrum envelope information of the voice.

The spectrum encoding unit 3 encodes the LSP obtained by the spectrum analysis unit 2 and outputs the LSP code to the multiplexing unit 20. Also, its L
The SP is quantized, and the LSP after the quantization is combined with the synthesis filter 4.
, And outputs the filter coefficient to the synthesis filter 4 and the auditory weighting unit 6.

The synthesis filter 4 includes a spectrum encoding unit 3
When the filter coefficient is received from, a filtering process is performed on the temporary sound source selected by the changeover switch 19 using the filter coefficient to generate a temporary synthesized sound. The process of generating a temporary sound source will be described later. When the synthesis filter 4 generates a synthesized sound, the subtracter 5 outputs a difference signal between the synthesized sound and the input voice after the pre-processing by the pre-processing unit 1, and the auditory weighting unit 6 determines that the spectrum encoding unit 3 An auditory weighting filter coefficient is calculated based on the filter coefficient to be output, and the subtractor 5 is used by using the auditory weighting filter coefficient.
Performs an auditory weighting filter process on the difference signal output by the controller and outputs an auditory weighted difference signal.

The distortion minimizing section 7 sequentially updates the index and the encoding mode, thereby obtaining the auditory weighting section 6.
Minimizing the power of the auditory weighting difference signal output by. That is, the index and the mode information are appropriately selected,
Each time the signals are output to the sound source decoding units 8 and 9 and the changeover switch 19, the power of the perceptual weighting difference signal is calculated, and the result of the calculation is searched for the combination of the index and mode information that minimizes the power. When the index and the mode information that minimize the power of the auditory weighting difference signal are determined, the index and the mode information are multiplexed by the multiplexing unit 2.
Output to 0. However, since excitation excitation decoding section 9 does not include an adaptive excitation codebook, it does not output an adaptive excitation code when outputting mode information indicating the second encoding mode.

Upon receiving the index from distortion minimizing section 7, excitation decoding sections 8 and 9 generate a temporary excitation according to the index. More specifically, first, adaptive excitation codebook 10 of excitation decoding section 8 stores a past excitation of a predetermined length, and receives an adaptive excitation code from distortion minimizing section 7, and stores the past excitation corresponding to the adaptive excitation code. Is output as an adaptive code vector. Note that adaptive excitation codebook 10 selects the temporary excitation output from changeover switch 19 in response to the index and mode information after distortion minimizing section 7 selects the index and mode information. The temporary sound source is stored as a final sound source.

Driving excitation codebook 11 of excitation decoding section 8 has:
When a driving code vector, which is a plurality of non-noise time-series vectors, is stored and a driving excitation code is received from the distortion minimizing unit 7, a driving code vector corresponding to the driving excitation code is output. However, driving excitation codebook 11 is provided with an algebraic excitation table in which each time-series vector is represented by a plurality of pulse positions and polarities in advance, so that algebraic excitation based on the driving excitation code output from distortion minimizing section 7 is provided. A sound source may be generated, and the algebraic sound source may be output as a driving code vector.

When the gain codebook 12 outputs a gain value corresponding to the gain code, the adaptive code vector output from the adaptive excitation codebook 10 and the driving code vector output from the driving excitation codebook 11 are multiplied by a multiplier. The gain value is multiplied by 13 and 14, and the multiplier 1 is multiplied by the adder 15.
The multiplication results of 3, 14 are added to each other.

On the other hand, driving excitation codebook 1 of excitation decoding section 9
Numeral 6 stores a driving code vector which is a plurality of noise-like time-series vectors, and upon receiving a driving excitation code from distortion minimizing section 7, outputs a driving code vector corresponding to the driving excitation code. However, the driving excitation codebook 16 is provided with an algebraic excitation table in which each time-series vector is expressed in advance by a plurality of pulse positions and polarities.
May generate an algebraic excitation based on the driving excitation code output by the controller, and output the algebraic excitation as a driving code vector. When the gain codebook 17 outputs a gain value corresponding to the gain code, the driving code vector output from the driving excitation codebook 16 is multiplied by the gain value by the multiplier 18.

In this way, when the provisional excitation is output from adder 15 of excitation decoding section 8 and the temporary excitation is output from multiplier 18 of excitation decoding section 9, changeover switch 19
Is a temporary excitation or excitation decoding section 9 output by the excitation decoding section 8 according to the mode information output by the distortion minimizing section 7.
Selects one of the tentative sound sources output by, and gives the selected tentative sound source to the synthesis filter 4.

The multiplexing unit 20 includes the LSP code coded by the spectrum coding unit 3 and the index and mode information updated by the distortion minimizing unit 7 (the index and mode in which the power of the auditory weighting difference signal is minimized). ) To generate a speech code and output the speech code.

Next, upon inputting the speech code output from the speech encoding device, the separation unit 21 of the speech decoding device separates the LSP code, index, and mode information contained in the speech code.

Upon receiving the index from separation section 21, excitation decoding sections 22 and 23 generate a temporary excitation according to the index. Specifically, first, adaptive excitation codebook 24 of excitation decoding section 22 stores past excitations for a predetermined length, and receives an adaptive excitation code from demultiplexing section 21, and receives past excitations corresponding to the adaptive excitation code. Is output as an adaptive code vector. In addition,
When the changeover switch 33 selects and outputs a temporary excitation, the adaptive excitation codebook 24 stores the temporary excitation as a final excitation.

Driving excitation codebook 25 of excitation decoding section 22
Stores a driving code vector, which is a plurality of non-noise time-series vectors, and outputs a driving code vector corresponding to the driving excitation code when receiving the driving excitation code from the separation unit 21. However, the driving excitation codebook 25 includes an algebraic excitation table in which each time-series vector is represented in advance by a plurality of pulse positions and polarities, so that the algebraic excitation is output based on the driving excitation code output from the separation unit 21. Alternatively, the algebraic excitation may be generated and output as a driving code vector.

When the gain codebook 26 outputs a gain value corresponding to the gain code, the adaptive code vector output from the adaptive excitation codebook 24 and the driving code vector output from the driving excitation codebook 25 are multiplied by a multiplier. The gain value is multiplied by 27 and 28, and the multiplier 2 is
The multiplication results of 7, 28 are added to each other.

On the other hand, driving excitation codebook 30 of excitation decoding section 23 stores a driving code vector which is a plurality of noise-like time series vectors, and receives driving excitation code from separation section 21, and receives the driving excitation code. Is output. However, the driving excitation codebook 30 is provided with an algebraic excitation table in which each time-series vector is expressed by a plurality of pulse positions and polarities in advance, so that the algebraic excitation is output based on the driving excitation code output from the separation unit 21. Alternatively, the algebraic excitation may be generated and output as a driving code vector. When the gain codebook 31 outputs a gain value corresponding to the gain code, the driving code vector output from the driving excitation codebook 30 is multiplied by the gain value by the multiplier 32.

As described above, when the provisional excitation is output from adder 29 of excitation decoding section 22 and the temporary excitation is output from multiplier 32 of excitation decoding section 23, changeover switch 3
Reference numeral 3 denotes a temporary excitation or excitation decoding section 2 output by the excitation decoding section 22 according to the mode information output by the separation section 21.
3 selects one of the tentative sound sources output, and gives the selected tentative sound source to the synthesis filter 35.

When the separating section 21 outputs the LSP code, the spectrum decoding section 34 decodes the LSP code,
The decoding result is converted into a filter coefficient of the synthesis filter 35, and the filter coefficient is output to the synthesis filter 35 and the post-processing unit 36. Upon receiving the filter coefficient from the spectrum decoding unit 34, the synthesis filter 35 performs a filtering process on the temporary sound source selected by the changeover switch 33 using the filter coefficient, and generates a temporary synthesized sound. The post-processing unit 36 performs post-processing such as voice enhancement processing on the synthesized sound generated by the synthesis filter 35 based on the filter coefficients and the like output from the spectrum decoding unit 34, and reproduces the input sound (output sound). Is output.

FIG. 17 is an explanatory diagram showing an example of a gain codebook used by a conventional speech encoding apparatus and speech decoding apparatus. In particular, FIG. 17A shows an example of the gain codebooks 12 and 26, and FIG. 17B shows an example of the gain codebooks 17 and 31.

In this example, each gain codebook stores 128 gain codewords. However, the gain codewords stored in the gain codebooks 12 and 26 are composed of codewords indicating a set of two gain values that are multiplied by the adaptive code vector and the drive code vector.
Is composed of a codeword indicating one gain value by which the driving code vector is multiplied. Although the index and the evaluation value rank are not actually stored in each gain codebook, they are described for convenience of explanation. The index has a value of 0 to 127 in order from the upper code word. The evaluation value is the value of the power (sum of squares) of the gain codeword. For example, the power ranking of a codeword having an index of “1” is “102”.

As an operation of each gain codebook, when a certain gain code is input, a gain codeword stored at an index position corresponding to the gain code is output. The gain codeword stored in each gain codebook is created by learning so that distortion between the learning speech and the encoded speech is reduced. Then, in order to minimize the deterioration of the output voice due to a code error when transmitting the voice code, the gain codeword is appropriately rearranged.

For example, an expected value of the magnitude of the degradation that occurs when a one-bit error is actually given to the gain code is calculated,
Furthermore, an expected value of the magnitude of the degradation that occurs when two randomly selected gain codewords are exchanged is calculated, and when the expected value of the latter is reduced as compared with the expected value of the former, the gain codeword of the gain is actually calculated. Swap the storage order. This operation is repeated until the decrease in the expected value becomes very small. Conventional speech encoding devices and speech decoding devices use a gain codebook in which such gain codewords are rearranged.

[0036]

Since the conventional speech coding apparatus and speech decoding apparatus are constructed as described above, the distortion minimizing section 7 of the speech coding apparatus uses the power of the auditory weighting difference signal. Although the optimal mode information is selected so as to minimize it, if the transmission path error is superimposed on the audio code and the audio decoding device mistakenly recognizes the mode information, there is a problem that the reproduction quality of the input audio is significantly deteriorated. In addition, in order to minimize deterioration due to code errors, codewords are rearranged for each codebook.However, since rearrangement is not performed in consideration that mode information may be erroneously recognized,
There has been a problem that it is not possible to increase the resistance to mode information errors.

More specifically, since the rearrangement of the codewords in the gain codebooks 12 and 26 and the rearrangement of the codewords in the gain codebooks 17 and 31 are performed independently, the power (evaluation value) of the gain codeword is obtained. Paying attention to the rank of, as shown in FIG. 17, there is no correlation between the gain codebooks 12 and 26 and the gain codebooks 17 and 31 at all. For this reason, for example, when decoding a gain code having an index of “0”, if the mode information is erroneously recognized and the second coding mode is selected instead of the originally selected first coding mode, The gain value of “121” is selected instead of the gain value of the evaluation value rank of “41”. This allows
The amplitude of the output voice changes greatly, causing large local degradation.

The present invention has been made in order to solve the above-described problems, and a speech encoding apparatus and a speech decoding apparatus capable of suppressing deterioration of speech reproduction quality even if mode information is erroneously recognized. And a code word arrangement method.

[0039]

In the speech coding apparatus according to the present invention, the storage order of codewords of a plurality of codebooks is rearranged in accordance with the order of evaluation values for codewords of another codebook. It is as if it were.

The speech coding apparatus according to the present invention uses the power or the average amplitude of the code word as the evaluation value for the code word.

[0041] In the speech coding apparatus according to the present invention, the plurality of codebooks are codebooks for outputting excitation gains.

According to the speech encoding apparatus of the present invention, the storage order of the codewords of the plurality of codebooks is adjusted so that the total value of the deviations of the evaluation values of the corresponding codewords among the plurality of codebooks is minimized. It is made to be rearranged.

In the speech coding apparatus according to the present invention, when a sound source is generated from a codeword and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value.

The speech coding apparatus according to the present invention has mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the index after the mapping, thereby forming a plurality of codebooks. A state equivalent to the updated storage order is constructed without updating the storage order of the codewords in advance based on the rank of the evaluation value.

The speech coding apparatus according to the present invention has a mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the index after the mapping, so that a plurality of codebooks are output. A state equivalent to the storage order after the update is constructed without updating the storage order of the codewords in advance so that the total value of the deviations of the evaluation values is minimized.

A speech decoding apparatus according to the present invention is arranged such that a plurality of codebooks are rearranged in a storage order of codewords corresponding to the order of evaluation values of codewords of another codebook. It is.

The speech decoding apparatus according to the present invention uses the power or average amplitude of a code word as an evaluation value for the code word.

[0048] In the speech decoding apparatus according to the present invention, the plurality of codebooks are codebooks for outputting excitation gains.

According to the speech decoding apparatus of the present invention, the storage order of the codewords of the plurality of codebooks is set so that the total value of the deviations of the evaluation values of the corresponding codewords among the plurality of codebooks is minimized. It is made to be rearranged.

In the speech decoding apparatus according to the present invention, when a sound source is generated from a codeword and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value.

The speech decoding apparatus according to the present invention has mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, so that a plurality of codebooks are output. A state equivalent to the updated storage order is constructed without updating the storage order of the codewords in advance based on the rank of the evaluation value.

The speech decoding apparatus according to the present invention has mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, thereby obtaining a plurality of codebooks. A state equivalent to the storage order after the update is constructed without updating the storage order of the codewords in advance so that the total value of the deviations of the evaluation values is minimized.

The codeword arranging method according to the present invention examines the evaluation values of the codewords of each codebook, and, according to the order of the evaluation values of the codewords of the other codebooks, at least one or more codebooks. The storage order of the codewords is rearranged.

In the code word arranging method according to the present invention, the power or the average amplitude of the code word is used as the evaluation value for the code word.

In the codeword arrangement method according to the present invention, the plurality of codebooks are codebooks for outputting excitation gains.

The code word arranging method according to the present invention calculates the sum of the deviations of the evaluation values of the corresponding code words among a plurality of codebooks, and calculates the difference until the sum is reduced and minimized.
The storage order of at least one or more codebooks is updated.

In the code word arranging method according to the present invention, when a sound source is generated from a code word and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value.

[0058]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. FIG. 1 is a block diagram showing a speech encoding apparatus according to Embodiment 1 of the present invention. In the figure, reference numeral 41 denotes a noise suppression process for suppressing background noise superimposed on input speech, A pre-processing unit that executes a low-frequency rejection filter process for cutting a DC component analyzes an input voice that has been pre-processed by the pre-processing unit 41, and obtains a line spectrum pair (hereinafter, L) that is spectral envelope information of the voice.
SP), and 43 encodes the LSP determined by the spectrum analyzer 42,
The LSP code is output to the multiplexing unit 60, the LSP is quantized, and the quantized LSP (same as the result of decoding the LSP code) is converted into a filter coefficient (linear prediction coefficient) of the synthesis filter 44. And a spectrum encoding unit that outputs the filter coefficients to the synthesis filter 44 and the auditory weighting unit 46.

Reference numeral 44 denotes a synthesis filter that performs a filtering process on the temporary sound source selected by the changeover switch 59 using the filter coefficients output from the spectrum encoding unit 43 to generate a temporary synthesized sound, and 45 denotes a synthesis filter. A subtractor that outputs a difference signal between the synthesized sound generated by the pre-processing unit 41 and the input voice that has been pre-processed by the pre-processing unit 41; A hearing weighting unit that performs a hearing weighting filter process on the difference signal output by the subtracter 45 using the hearing weighting filter coefficient, and outputs a hearing weighted difference signal.

Reference numeral 47 denotes an index (gain code, driving excitation code, adaptive excitation code) and an encoding mode for calculating the power of the perceptual weighting difference signal output from the perceptual weighting unit 46 and minimizing the power. Distortion minimizing units 48 and 49 for sequentially updating mode information have a codebook for outputting a codeword corresponding to the index updated by the distortion minimizing unit 47, and a sound source for generating a temporary sound source from the codeword. It is a decoding unit.

Reference numeral 50 denotes an adaptive code vector which is a time series vector which stores a past excitation of a predetermined length and receives the adaptive excitation code from the distortion minimizing section 47, and periodically repeats the past excitation corresponding to the adaptive excitation code. An adaptive excitation codebook 51 that outputs a driving code vector that is a plurality of non-noise time-series vectors, receives a driving excitation code from the distortion minimizing unit 47, and outputs a driving code corresponding to the driving excitation code. A driving excitation codebook 52 for outputting a vector stores a codeword relating to the gain (a word indicating a gain value), and receives a gain code from the distortion minimizing section 47 to output a gain value corresponding to the gain code. Codebook 53 is a multiplier for multiplying the gain value output from gain codebook 52 by the adaptive code vector output from adaptive excitation codebook 50, and 54 is Driving the gain value to force excitation codebook 5
A multiplier for multiplying the driving code vector output by 1; 55
Is an adder that adds the multiplication result of the multiplier 53 and the multiplication result of the multiplier 54 and outputs the addition result (temporary sound source).

Reference numeral 56 stores a driving code vector which is a plurality of noise-like time-series vectors, and upon receiving a driving excitation code from the distortion minimizing section 47, outputs a driving code vector corresponding to the driving excitation code. A codebook 57 stores a codeword (a word indicating a gain value) related to the gain, and receives a gain code from the distortion minimizing unit 47 and outputs a gain value corresponding to the gain code. A multiplier that multiplies the gain value output from the codebook 57 by the driving code vector output from the driving excitation codebook 56 and outputs the multiplication result (temporary excitation).

When the mode information 59 is received from the distortion minimizing section 47, the excitation decoding section 4 receives the mode information in accordance with the mode information.
A switch for selecting a temporary sound source output by 8 or a temporary sound source output by the sound source decoding unit 49 and providing the selected temporary sound source to the synthesis filter 44. The pre-processing unit 4
1, spectrum analysis unit 42, spectrum encoding unit 43,
Synthesis filter 44, subtractor 45, auditory weighting unit 46,
Encoding means is composed of the distortion minimizing section 47, the excitation decoding sections 48 and 49, and the changeover switch 59. A multiplexing unit 60 generates a speech code by multiplexing the LSP code encoded by the spectrum encoding unit 43 and the index and mode information updated by the distortion minimizing unit 47, and outputs the speech code. Multiplexing means).

FIG. 2 is a block diagram showing a speech decoding apparatus according to Embodiment 1 of the present invention. In the figure, reference numeral 61 demultiplexes an LSP code, an index, and mode information multiplexed by the speech encoding apparatus. Separation unit (separation means), 6
Reference numerals 2 and 63 denote a sound source decoding unit that has a codebook that outputs a codeword corresponding to the index separated by the separation unit 61 and generates a sound source from the codeword.

Reference numeral 64 stores a past excitation of a predetermined length, and upon receiving an adaptive excitation code from the separation unit 61, outputs an adaptive code vector which is a time series vector that periodically repeats the past excitation corresponding to the adaptive excitation code. Adaptive excitation codebook, 6
Reference numeral 5 denotes a driving excitation codebook that stores a driving code vector that is a plurality of non-noise time-series vectors and, when receiving a driving excitation code from the separation unit 61, outputs a driving code vector corresponding to the driving excitation code. Is a gain codebook that stores a code word (a word indicating a gain value) related to the gain and receives a gain code from the separation unit 61 and outputs a gain value corresponding to the gain code. A multiplier that multiplies the gain value by the adaptive code vector output by the adaptive excitation codebook 64; 68, a multiplier that multiplies the gain value output by the gain codebook 66 by the driving code vector output by the driving excitation codebook 65; Is an adder that adds the multiplication result of the multiplier 67 and the multiplication result of the multiplier 68 and outputs the addition result (temporary sound source).

Reference numeral 70 denotes a driving excitation codebook that stores a driving code vector, which is a plurality of noise-like time-series vectors, and outputs a driving code vector corresponding to the driving excitation code when receiving the driving excitation code from the separation unit 61. , 71 store a code word (a word indicating a gain value) relating to the gain, and
, A gain codebook that outputs a gain value corresponding to the gain code, 72 multiplies the gain value output by the gain codebook 71 with the driving code vector output by the driving excitation codebook 70, This is a multiplier that outputs a multiplication result (temporary sound source).

Upon receiving the mode information from the separating section 61, the 73 selects a temporary excitation output from the excitation decoding section 62 or a temporary excitation output from the excitation decoding section 63 according to the mode information. Synthesizing filter 75 for temporary sound source
, A switch 74 provided to the L
The SP code is decoded and the decoded result is
And a spectrum decoding unit that converts the filter coefficients to a synthesis filter 75 and a post-processing unit 76.

Reference numeral 75 denotes a synthesis filter that performs a filtering process on the temporary sound source selected by the changeover switch 73 using the filter coefficients output from the spectrum decoding unit 74 to generate a temporary synthesized sound, and 76 denotes a spectrum decoding filter. A post-processing unit that performs post-processing such as voice emphasis processing on the synthesized sound generated by the synthesis filter 75 based on the filter coefficients and the like output by the conversion unit 74, and outputs the reproduction result (output sound) of the input sound. . The sound source decoding unit 6
2, 63, changeover switch 73, spectrum decoding section 7
4. Decoding means is composed of the synthesis filter 75 and the post-processing unit 76. FIG. 3 is a flowchart showing a codeword arrangement method according to Embodiment 1 of the present invention.

Next, the operation will be described. The conventional speech encoding device and speech decoding device require about 5 to 50 ms for one.
The processing is executed for each frame as a frame.

First, the pre-processing unit 41 of the speech encoding device
Upon receiving the input voice, the CPU executes noise suppression processing for suppressing background noise superimposed on the input voice, and also executes low-frequency rejection filter processing for cutting a DC component of the input voice. When the pre-processing unit 41 performs pre-processing on the input voice, the spectrum analysis unit 42 analyzes the input voice after the pre-processing and obtains LSP, which is spectrum envelope information of the voice.

Then, spectrum encoding section 43 encodes the LSP obtained by spectrum analysis section 42 and outputs the LSP code to multiplexing section 60. Also,
The LSP is quantized, the LSP after the quantization is converted into a filter coefficient of the synthesis filter 44, and the filter coefficient is output to the synthesis filter 44 and the auditory weighting unit 46.

Upon receiving the filter coefficients from the spectrum encoding unit 43, the synthesis filter 44 performs a filtering process on the temporary sound source selected by the changeover switch 59 using the filter coefficients to generate a temporary synthesized sound. I do. The process of generating a temporary sound source will be described later. The subtractor 45
When the synthesis filter 44 generates a synthesized sound, a difference signal between the synthesized sound and the input voice after preprocessing by the preprocessing unit 41 is output.
Calculates an auditory weighting filter coefficient on the basis of the filter coefficient output by the above, and performs an auditory weighting filter process on the difference signal output by the subtractor 45 using the auditory weighting filter coefficient to output an auditory weighting difference signal.

The distortion minimizing section 47 attempts to minimize the power of the perceptual weighting difference signal output from the perceptual weighting section 46 by sequentially updating the index and the coding mode. That is, each time the index and the mode information are appropriately selected and output to the excitation decoding sections 48 and 49 and the changeover switch 59, the power of the perceptual weighting difference signal is calculated,
As a result of the calculation, a combination of an index and mode information with the smallest power is searched. Then, when the index and the mode information that minimize the power of the auditory weighting difference signal are obtained, the index and the mode information are output to the multiplexing unit 60. However, since excitation excitation decoding section 49 does not include an adaptive excitation codebook, it does not output an adaptive excitation code when outputting mode information indicating the second encoding mode.

Upon receiving the index from distortion minimizing section 47, excitation decoding sections 48 and 49 generate a temporary excitation according to the index. More specifically, first, adaptive excitation codebook 50 of excitation decoding section 48 stores a past excitation of a predetermined length, and receives an adaptive excitation code from distortion minimizing section 47, and stores the past excitation corresponding to the adaptive excitation code. Is output as an adaptive code vector. Note that adaptive excitation codebook 50 includes distortion minimizing section 47.
After selecting the index and mode information, the changeover switch 59
When the selected temporary sound source is output, the temporary sound source is stored as the final sound source.

Driven excitation codebook 51 of excitation decoding section 48
Stores a driving code vector, which is a plurality of non-noise time-series vectors, and upon receiving a driving excitation code from the distortion minimizing unit 47, outputs a driving code vector corresponding to the driving excitation code. However, the driving excitation codebook 51 has a
By providing an algebraic excitation table that expresses each time-series vector by a plurality of pulse positions and polarities, an algebraic excitation is generated based on the driving excitation code output by the distortion minimizing unit 47, and the algebraic excitation is driven. You may make it output as a code vector.

When gain codebook 52 outputs a gain value corresponding to the gain code, adaptive code vector output from adaptive excitation codebook 50 and drive code vector output from drive excitation codebook 51 are multiplied by a multiplier. The gain value is multiplied by 53 and 54, and the multiplier 5 is multiplied by an adder 55.
The 3,54 multiplication results are added to each other.

On the other hand, driving excitation codebook 56 of excitation decoding section 49 stores a driving code vector, which is a plurality of noise-like time-series vectors, and receives driving excitation code from distortion minimizing section 47 to drive the excitation codebook. A driving code vector corresponding to the excitation code is output. However, the driving excitation codebook 56 is provided with an algebraic excitation table in which each time-series vector is represented by a plurality of pulse positions and polarities in advance, so that an algebraic excitation code output from the distortion minimizing unit 47 is obtained. A sound source may be generated, and the algebraic sound source may be output as a driving code vector. When the gain codebook 57 outputs a gain value corresponding to the gain code, the driving code vector output from the driving excitation codebook 56 is multiplied by the gain value by the multiplier 58.

As described above, when the provisional excitation is output from adder 55 of excitation decoding section 48 and the temporary excitation is output from multiplier 58 of excitation decoding section 49, changeover switch 5
9 selects either the temporary excitation output by the excitation decoding section 48 or the temporary excitation output by the excitation decoding section 49 in accordance with the mode information output by the distortion minimizing section 47, and selects the selected temporary excitation. Is given to the synthesis filter 44.

The multiplexing section 60 includes a spectrum encoding section 43
Is multiplexed with the index and mode information (index and mode information that minimizes the power of the auditory weighting difference signal) updated by the distortion minimizing unit 47 to generate a speech code. Outputs speech codes.

Next, upon input of the speech code output from the speech encoding device, the separation unit 61 of the speech decoding device separates the LSP code, index, and mode information contained in the speech code.

Upon receiving the index from separation section 61, excitation decoding sections 62 and 63 generate a temporary excitation according to the index. Specifically, first, adaptive excitation codebook 64 of excitation decoding section 62 stores a past excitation at a predetermined length and, when receiving an adaptive excitation code from separation section 61, outputs a past excitation corresponding to the adaptive excitation code. Is output as an adaptive code vector. In addition,
When the changeover switch 73 selects and outputs a temporary excitation, the adaptive excitation codebook 64 stores the temporary excitation as the final excitation.

Driving excitation codebook 65 of excitation decoding section 62
Stores a driving code vector, which is a plurality of non-noise time-series vectors, and upon receiving a driving excitation code from the separation unit 61, outputs a driving code vector corresponding to the driving excitation code. However, the driving excitation codebook 65 is provided with an algebraic excitation table in which each time-series vector is represented by a plurality of pulse positions and polarities in advance, so that the algebraic excitation is output based on the driving excitation code output by the separation unit 61. Alternatively, the algebraic excitation may be generated and output as a driving code vector.

When gain codebook 66 outputs a gain value corresponding to the gain code, the adaptive code vector output from adaptive excitation codebook 64 and the driving code vector output from driving excitation codebook 65 are multiplied by a multiplier. The gain value is multiplied by 67 and 68, and the multiplier 6 is multiplied by an adder 69.
The 7,68 multiplication results are added to each other.

On the other hand, driving excitation codebook 70 of excitation decoding section 63 stores a driving excitation vector which is a plurality of noise-like time-series vectors, and receives driving excitation code from separation section 61, and receives the excitation excitation code. Is output. However, the driving excitation codebook 70 is provided with an algebraic excitation table in which each time-series vector is expressed by a plurality of pulse positions and polarities in advance, so that the algebraic excitation is output based on the driving excitation code output by the separation unit 61. Alternatively, the algebraic excitation may be generated and output as a driving code vector. When the gain codebook 71 outputs a gain value corresponding to the gain code, the driving code vector output from the driving excitation codebook 70 is multiplied by the gain value by the multiplier 72.

As described above, when the provisional excitation is output from adder 69 of excitation decoding section 62 and the temporary excitation is output from multiplier 72 of excitation decoding section 63, changeover switch 7
3 is a temporary excitation or excitation decoding unit 6 output by the excitation decoding unit 62 according to the mode information output by the separation unit 61.
3 selects one of the tentative sound sources output, and gives the selected tentative sound source to the synthesis filter 75.

When the separating section 61 outputs the LSP code, the spectrum decoding section 74 decodes the LSP code,
The decoding result is converted into a filter coefficient of the synthesis filter 75, and the filter coefficient is output to the synthesis filter 75 and the post-processing unit. Upon receiving the filter coefficient from the spectrum decoding unit 74, the synthesis filter 75 performs a filtering process on the temporary sound source selected by the changeover switch 73 using the filter coefficient, and generates a temporary synthetic sound. The post-processing unit 76 performs post-processing such as voice enhancement processing on the synthesized sound generated by the synthesis filter 75 based on the filter coefficients and the like output from the spectrum decoding unit 74, and reproduces the input sound (output sound). Is output.

FIG. 4 is an explanatory diagram showing an example of the gain codebook used by the speech encoding device and the speech decoding device. Particularly, FIG. 4A shows the gain codebooks 52 and 6.
FIG. 4B shows an example of gain codebooks 57 and 71.
An example is shown.

In this example, each gain codebook stores 128 gain codewords. However, the gain codewords stored in the gain codebooks 52 and 66 are composed of codewords indicating sets of two gain values that are multiplied by the adaptive code vector and the driving code vector, and the gain codebooks 57 and 71
Is composed of a codeword indicating one gain value by which the driving code vector is multiplied. Although the index and the evaluation value rank are not actually stored in each gain codebook, they are described for convenience of explanation. The index has a value of 0 to 127 in order from the upper code word. The evaluation value is the value of the power (sum of squares) of the gain codeword (the evaluation value is not limited to the power of the gain codeword, but may be the average amplitude of the gain codeword). For example, the power ranking of a codeword having an index of “1” is “102”.

As an operation of each gain codebook, when a certain gain code is input, a gain codeword stored at an index position corresponding to the gain code is output. The gain codeword stored in each gain codebook is created by learning so that distortion between the learning speech and the encoded speech is reduced. Then, in order to minimize the deterioration of the output voice due to a code error when transmitting the voice code, the gain codeword is appropriately rearranged.

For example, the expected value of the magnitude of the degradation that occurs when a one-bit error is actually given to the gain code is calculated,
Furthermore, an expected value of the magnitude of the degradation that occurs when two randomly selected gain codewords are exchanged is calculated, and when the expected value of the latter is reduced as compared with the expected value of the former, the gain codeword of the gain is actually calculated. Swap the storage order. This operation is repeated until the decrease in the expected value becomes very small.

However, with respect to the gain codewords stored in the gain codebooks 57 and 71, the power of each gain codeword is checked and stored in the gain codebooks 57 and 71 based on the power. Update the storage order of gain codewords. That is, when the powers of the gain codewords stored in the gain codebooks 57 and 71 are checked (step ST1), the gains stored in the gain codebooks 52 and 66 in which the rearrangement of the gain codewords has already been completed. A process of rearranging the storage order of the gain codewords stored in the gain codebooks 57 and 71 so as to be in the same order as the power order of the codewords is executed (step ST2).

Each gain codebook in FIG. 4 has already been rearranged. In the gain codebooks 52 and 66, for example, the power (evaluation value) rank of the gain codeword corresponding to the index “0” is “41”. The gain codeword of “41” is stored so as to correspond to the index of “0”. Similarly, the storage order is rearranged for gain codewords whose index is “1” or later.

FIG. 5 is an explanatory diagram showing an example of a speech code output from the multiplexing section 60. In the multiplexing unit 60, LS
P code, mode information, gain code, driving excitation code and adaptive excitation code are multiplexed (however, the adaptive excitation code is included in the multiplexing target only in the first coding mode) to generate a speech code. However, in the first embodiment,
Whether the encoding mode is the first encoding mode or the second encoding mode, the audio code is encoded so that the number of encoded bits of the gain code and the multiplexing position of the gain code do not change. Has been generated.

Here, the transmission error is superimposed on the speech code (see FIG. 5 (a)) generated by the speech coding apparatus in the first coding mode, so that the speech code is changed as shown in FIG. It is assumed that the state changes as shown in b). In this case, the speech decoding device erroneously recognizes that the encoding mode is the second encoding mode and performs the decoding process of the input speech, but as described above, the encoding mode is the first encoding mode. In the coding mode and the second coding mode, the speech code is generated so that the number of bits of the gain code and the multiplexing position of the gain code do not change. Therefore, even if a transmission error occurs, the speech decoding apparatus can accurately recognize the value of the gain code. In the example of FIG. 5, the value of the gain code is 2 irrespective of whether or not the mode information is erroneously recognized.

Therefore, even if a transmission error occurs in the mode information, the gain codebooks 66 and 71 in the speech decoding apparatus can output a gain codeword (gain value) corresponding to the same gain code. Also, gain codebook 6
As described above, the gain code words stored in 6, 6 and 71 are rearranged so that the power value order is the same. Therefore, even if a transmission error occurs in mode information, a gain code having the same value is used. If it can be input, the magnitude of the output gain value will not change drastically.

As is apparent from the above, the first embodiment
According to the above, the gain codebooks 52, 66 (or 57, 7
The storage order of the gain codeword of 1) is different from that of the other gain codebooks 5.
Since the rearrangement is performed in accordance with the order of the evaluation values of the gain codewords of 7, 71 (or 52, 66), the speech encoding apparatus generates a transmission error and performs speech decoding. This makes it possible to generate a speech code capable of suppressing the deterioration of the reproduction quality of the speech even if the coding device misidentifies the mode information. On the other hand, in the audio decoding device, even if the mode information is erroneously recognized, it is possible to suppress the deterioration of the audio reproduction quality.

According to the first embodiment, the power or the average amplitude of the gain codeword is used as the evaluation value for the gain codeword. As a result, even if the speech decoding device mistakenly recognizes the mode information, it is possible to generate a speech code that does not significantly degrade the power and amplitude of the speech reproduced by the speech decoding device. On the other hand, in the audio decoding device, even if the mode information is erroneously recognized, the power and amplitude of the audio are not significantly deteriorated.
It has the effect of being able to reproduce audio.

Further, according to the first embodiment, gain codebooks 52, 66, 57, and 71 are configured to be codebooks that output excitation gains (gain values). Even if a transmission error occurs and the speech decoding device misidentifies the mode information, it is possible to generate a speech code that does not significantly degrade the gain value of the speech reproduced by the speech decoding device. Play. On the other hand, in the audio decoding device, even if the mode information is erroneously recognized, there is an effect that the audio can be reproduced without causing significant deterioration of the audio gain value.

Embodiment 2 FIG. 6 is a block diagram showing a speech encoding apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as in FIG. 1 denote the same or corresponding parts, and a description thereof will be omitted. Reference numeral 81 denotes an excitation mode selection unit that determines mode information from the LSP quantized by the spectrum encoding unit 43, and 82 receives mode information from the excitation mode selection unit 81, and the driving excitation codebook 51 outputs according to the mode information. While selecting the driving code vector to be output or the driving code vector output by the driving excitation codebook 56,
A changeover switch for selecting a gain value output by the gain codebook 52 or a gain value output by the gain codebook 57.

A multiplier 83 multiplies the adaptive code vector output from the adaptive excitation codebook 50 by the gain value selected by the changeover switch 82, and a multiplier 84 selects the gain value selected by the changeover switch 82 by the changeover switch 82. A multiplier for multiplying the driving code vector obtained by the driving,
This is an adder that adds the multiplication result of No. 3 and the multiplication result of the multiplier 84 and outputs the addition result (temporary sound source). Note that a sound source mode selection unit 81, a changeover switch 82, a multiplier 83,
The 84 and the adder 85 constitute an encoding unit.

FIG. 7 is a block diagram showing a speech decoding apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as in FIG. 2 denote the same or corresponding parts, and a description thereof will be omitted. Reference numeral 91 denotes a sound source mode selection unit for determining mode information from the LSP quantized by the spectrum decoding unit 74;
2 receives the mode information from the sound source mode selection unit 91,
According to the mode information, the drive code vector output from the drive excitation codebook 65 or the drive code vector output from the drive excitation codebook 70 is selected, and the gain value output from the gain codebook 66 or the gain codebook 71 is output. A changeover switch for selecting a gain value.

Reference numeral 93 denotes a multiplier for multiplying the adaptive code vector output from the adaptive excitation codebook 64 by the gain value selected by the changeover switch 92, and 94 denotes a gain value selected by the changeover switch 92 by the changeover switch 92. Multiplier for multiplying the driving code vector by
This is an adder that adds the multiplication result of No. 3 and the multiplication result of the multiplier 94 and outputs the addition result (temporary sound source). Note that a tone generator mode selector 91, a changeover switch 92, a multiplier 93,
The adder 94 and the adder 95 constitute a decoding unit.

Next, the operation will be described. In the first embodiment, the changeover switch 59 (or 73) selects the temporary sound source output from the sound source decoding unit 48 (or 62) or the temporary sound source output from the sound source decoding unit 49 (or 63). , The selected temporary sound source is output to the synthesis filter 44 (or 75).
As shown in (5), the changeover switch 82 (or 92) selects the driving code vector of the driving excitation codebook 51 (or 65) or the driving code vector of the driving excitation codebook 56 (or 70) and the multiplier 83 (or 93). ) And the gain value of the gain codebook 52 (or 66) or the gain value of the gain codebook 57 (or 71) is selected and output to the multiplier 84 (or 94), and the adder 85 (or 95) is selected. ) May add the multiplication result of the multiplier 83 (or 93) and the multiplication result of the multiplier 84 (or 94), and output the addition result to the synthesis filter 44 (or 75) as a temporary sound source. . In this case, the same effect as in the first embodiment can be obtained.

However, the gain codewords stored in the gain codebooks 57 and 71 are the same as the gain codewords stored in the gain codebooks 52 and 66. Is assumed to be composed of a code word indicating a set of gain values of.

In the first embodiment, the order in which the gain codewords are stored in each gain codebook is rearranged. However, the present invention is not limited to this, and vector code such as a power codebook or an LSP codebook may be used. As for the book, if the configuration uses a different codebook for each mode, it is also possible to use a codebook that is rearranged so that its order matches the power or amplitude of each codeword as an evaluation value. is there.

The evaluation value ranking of the two codebooks does not need to completely match as long as the difference between the rankings is small, and the same effect can be obtained. Also, after matching the evaluation value ranks of the two codebooks by the above method, the codewords in the two codebooks are rearranged simultaneously,
The rearrangement can be performed by various methods, such as minimizing deterioration when a bit error is superimposed on the gain code.

Embodiment 3 In the first embodiment, the storage order of the gain codewords stored in the gain codebooks 52 and 66 and the gain codebooks 57 and 71 is rearranged as shown in FIG. 4, but is shown in FIG. May be rearranged as follows.

More specifically, gain codebooks 52 and 66 store 128 gain codewords, and gain codebooks 57 and
The 71 stores 256 gain codewords. The gain codewords stored in the gain codebooks 52 and 66 are composed of codewords indicating sets of two gain values that are multiplied by the adaptive code vector and the driving code vector.
The gain codeword stored in 71 is composed of a codeword indicating one gain value by which the drive code vector is multiplied.

Although the index, the value of the upper 7 bits of the index, and the evaluation value rank are not actually stored in each gain codebook, they are described for convenience of explanation. The index is 0 from the codeword above
To 127 or a value from 0 to 255. The value of the upper 7 bits of the index is, for example, two such that the index is “0” when the index is “0” or “1” and “1” when the index is “2” or “3”. Each have the same value.

The evaluation value is the power (sum of squares) of the gain codeword.
(The evaluation value is not limited to the power of the gain codeword, but may be the average amplitude of the gain codeword.) For the gain codebooks 52 and 66, the evaluation value ranking of each gain codeword is It is shown. Gain codebook 57,
Regarding 71, the rank regarding the average of the evaluation values in the two gain codewords in which the upper 7 bits of the index have the same value is shown as the evaluation value average rank.

As an operation of each gain codebook, when a certain gain code is input, a gain codeword stored at an index position corresponding to the gain code is output. The gain codeword stored in each gain codebook is created by learning so that distortion between the learning speech and the encoded speech is reduced. Then, for the gain codebooks 52 and 66, the gain codewords are appropriately rearranged in order to minimize the deterioration of the output voice due to a code error when transmitting the voice code.

For example, the expected value of the magnitude of the degradation that occurs when a one-bit error is actually given to the gain code is calculated,
Furthermore, an expected value of the magnitude of the degradation that occurs when two randomly selected gain codewords are exchanged is calculated, and when the expected value of the latter is reduced as compared with the expected value of the former, the gain codeword of the gain is actually calculated. Swap the storage order. This operation is repeated until the decrease in the expected value becomes very small.

First, the gain codebooks 57 and 71 are appropriately gained in the same manner as the gain codebooks 52 and 66 in order to minimize the deterioration of the output voice due to a code error when transmitting the voice code. Reorder codewords. Next, two gain codewords whose upper 7 bits of the index have the same value at that time are paired. Then, the average value of the power of each gain codeword pair is obtained, and the gain codebook 57,
By examining the rank of the average value of the power in the gain codebook 71, the gain codebooks 52,
The gain codeword pairs in the gain codebooks 57 and 71 are rearranged so as to be in the same order as the power order of the 66 gain codewords.

Each gain codebook in FIG. 8 has already been rearranged. In the gain codebooks 52 and 66, for example, the power (evaluation value) rank of the gain codeword corresponding to the index “0” is “41”. Is "41"
Are stored so as to correspond to the index of “0”. Similarly, the storage order is rearranged for gain codewords whose index is “1” or later.

FIG. 9 is an explanatory diagram showing an example of a speech code output from the multiplexing section 60. In the multiplexing unit 60, LS
P code, mode information, gain code, driving excitation code and adaptive excitation code are multiplexed (however, the adaptive excitation code is included in the multiplexing target only in the first coding mode) to generate a speech code. However, in the third embodiment,
When the encoding mode is the first encoding mode, the number of encoded bits of the gain code is “7”, and when the encoding mode is the second encoding mode, the number of encoded bits of the gain code is “8”.
However, the gain code 7 bits in the first coding mode and the upper 7 bits of the gain code in the second coding mode
The speech code is generated such that the bit multiplexing positions match.

Here, a transmission error is superimposed on a speech code (see FIG. 9A) generated by the speech coding apparatus in the first coding mode, so that the speech code becomes It is assumed that the state changes as shown in b). In this case, the speech decoding apparatus erroneously recognizes that the encoding mode is the second encoding mode, and performs the decoding process on the input speech. However, as described above, the gain in the first encoding mode is Since the audio code is generated such that the multiplexing position of the 7-bit code and the upper 7 bits of the gain code in the second encoding mode match, even if a transmission error occurs in the mode information, the audio decoding device Can accurately recognize the value of the gain code.

That is, since the encoding mode is the first encoding mode, as shown in FIG. 9A, if the value of the gain code is “1” and there is no transmission error in the mode information, the gain The decoding process is performed using a gain codeword whose index in the codebook 66 is “1” (a codeword whose evaluation value rank is “102”). However, if a transmission error occurs in the mode information, the coding mode is erroneously recognized as the second coding mode. In the third embodiment, regardless of the presence or absence of the erroneous recognition,
The decoding process is performed using the gain codeword whose upper 7 bits of the index of the gain codebook 71 is “1”. Specifically, as shown in FIG. 9B, since the next bit of the gain code is “0”, the gain codeword whose index is “2” (= 1 × 2 + 0) (evaluation value rank is The decoding process is performed using the code word “102”).

Therefore, even if a transmission error occurs in the mode information, the gain codebooks 66 and 71 in the speech decoding apparatus store the gain codeword corresponding to the gain code whose evaluation value rank and the evaluation value average rank match or almost match. (Gain value) can be output, so that even if a transmission error occurs in the mode information, the magnitude of the output gain value does not extremely change.

As a result, the same effects as in the first embodiment can be obtained. Although the third embodiment has been described with reference to the case where the present invention is applied to the speech encoding apparatus of FIG. 1 and the speech decoding apparatus of FIG. 2, as in the second embodiment, the speech encoding apparatus of FIG. The present invention may be applied to the audio decoding device of FIG.

Embodiment 4 FIG. 10 is a configuration diagram showing a codeword arrangement apparatus to which a codeword arrangement method according to Embodiment 4 of the present invention is applied. In the figure, reference numeral 101 denotes a driving excitation codebook corresponding to the driving excitation codebooks 51 and 65; Are driving excitation codebooks equivalent to driving excitation codebooks 56 and 70,
03 and 104 are synthesis filters, 105 is a distance calculation unit, 1
Reference numeral 06 denotes the excitation codebook 10 until the calculation result of the distance calculation unit 105 (the total value of the deviations of the evaluation values) decreases and is minimized.
2 is a codeword exchanging unit that updates the storage order of the drive code vector, which is the codeword stored in 2.

Next, the operation will be described. As for the excitation codebook 101, as in the case of the gain codebook 52 in the first embodiment, in order to minimize the deterioration of the output speech due to a code error when transmitting the speech code, an appropriate codeword Perform the rearrangement. Further, using the driving excitation codebooks 101 and 102, a speech encoding process is performed using a large number of learning speech signals as inputs, and a filter coefficient for the synthesis filters 103 and 104 is calculated for each frame. The excitation codeword and gain value are used as learning data.
Accumulate separately.

First, distance calculation section 105 outputs the driving excitation code for each frame included in the learning data to driving excitation codebooks 101 and 102, and outputs the filter coefficients to synthesis filters 103 and 104.

Driving excitation codebook 101 includes distance calculation section 10
5 receives a driving excitation code, outputs a driving code vector corresponding to the driving excitation code,
Uses the filter coefficient output from the distance calculation unit 105 to perform synthesis filtering on the driving code vector to generate a first synthesized sound. Upon receiving the driving excitation code from distance calculating section 105, driving excitation codebook 102 outputs a driving code vector corresponding to the driving excitation code, and synthesis filter 104 uses the filter coefficient output from distance calculating section 105. Then, synthesis filtering is performed on the driving code vector to generate a second synthesized sound.

The distance calculation unit 105 includes the synthesis filter 103
Is calculated for each frame, the distance between the first synthesized sound generated by the above and the second synthesized sound generated by the synthesis filter 104 is summed, and the total distance is calculated by the code word replacing unit. Output to 106.

The code word exchange unit 106 is provided with the distance calculation unit 1
When 05 outputs the total distance, the total distance is stored. Up to this point is the initialization processing of the code word array device of FIG. Subsequent repetitive processing is as follows.

The codeword exchanging section 106 generates the driving excitation codebook 1 corresponding to the two driving excitation codes selected at random.
The codeword of 02 is exchanged. Distance calculator 105
Outputs the driving excitation code for each frame included in the learning data to the driving excitation codebook 101 and the driving excitation codebook 102 in which codewords have been exchanged again, and converts the filter coefficients into synthesis filters 103 and 104. Output to

The distance calculation section 105 receives the first synthesized sound and the second synthesized sound generated in the same manner from the synthesis filters 103 and 104, and outputs the first synthesized sound and the second synthesized sound. Is calculated for each frame, the distance values of all the frames are summed, and the total distance is output to the codeword replacing unit 106.

[0128] The code word exchanging unit 106 includes the distance calculating unit 1
When the total distance is received from 05, the total distance is compared with the total distance stored in advance. If the total distance has decreased, the total distance inputted this time is newly stored, and if the total distance has not decreased, processing for restoring the previous replacement of the code word is performed. Then, the process returns to the beginning of the repetition processing. The repetition processing up to this point is repeated until the decrease in the total distance is reduced, and driving excitation codebook 1
The rearrangement of the code word 02 is completed.

As is clear from the above, this embodiment 4
According to the configuration described above, the codewords of the driving excitation codebook 102 are exchanged until the total distance calculated by the distance calculation unit 105 is reduced and minimized. Even if an error occurs and the speech decoding device mistakenly recognizes the mode information, the expected value of the degradation related to the predetermined evaluation value is reduced, and as a result, the speech code capable of suppressing the degradation of the speech reproduction quality. Is produced. On the other hand, in the audio decoding device, even if the mode information is erroneously recognized, the expected value of the deterioration with respect to the predetermined evaluation value is reduced, and as a result, there is an effect that the deterioration of the sound reproduction quality can be suppressed. Further, even when different excitation codebooks are used at the time of encoding and at the time of decoding, a vector codebook that can provide a decoding result with a low expected value of deterioration related to the predetermined evaluation value is obtained.

In the fourth embodiment, as the distance in distance calculating section 105, the sum of squares of the difference between the values of the two synthesized sounds and the weight of the per-sample of the two synthesized sounds subjected to auditory weighting are used. Various things such as the sum of squares of the difference between the values and the power difference between two synthesized sounds can be applied. Also, here, the rearrangement of the driving excitation codebooks 101 and 102 has been described.
When a plurality of other codebooks such as the SP codebook are provided and the mode is switched, codewords may be rearranged by a similar sequential exchange process.

Embodiment 5 FIG. FIG. 11 is a configuration diagram showing a speech encoding apparatus according to Embodiment 5 of the present invention.
FIG. 2 is a configuration diagram showing a speech decoding apparatus according to Embodiment 5 of the present invention. In the drawings, the same reference numerals as those in FIGS. 1 and 2 denote the same or corresponding parts, and a description thereof will not be repeated. 11
1 maps the gain code after updating by the distortion minimizing section 47 and maps the mapped gain code to the gain codebook 5.
7 is a mapping unit that maps the gain code separated by the separation unit 61 and outputs the mapped gain code to the gain codebook 71. Note that the mapping units 111 and 112 constitute mapping means.

Next, the operation will be described. First, upon receiving the gain code from the distortion minimizing unit 47, the mapping unit 111 of the speech coding apparatus performs a mapping process according to a predetermined rule, and performs a mapping gain code corresponding to the gain code (a gain code after mapping). ) To the gain codebook 57
Output to However, in the gain codebook 57 according to the fifth embodiment, it is assumed that the rearrangement of the gain codewords based on the evaluation order as in the gain codebook 57 according to the first embodiment is not performed. That is, it is assumed that the storage order of the gain codebook 57 is as shown in FIG.

The gain codebook 57 includes a mapping unit 111
When receiving the mapping gain code from, it outputs the gain codeword stored at the index position corresponding to the mapping gain code. However, in the fifth embodiment, mapping section 111 has a mapping table as shown in FIG. 13 so that the same reordering result as the gain codeword in the first embodiment can be obtained.

For example, in the case of the mapping table of FIG. 13, when the mapping unit 111 inputs a gain code of “0”, it outputs a mapping gain code of “1”. As a result, the gain codebook 57 outputs a gain value whose evaluation value rank corresponding to the mapping gain code of “1” is “41” (see FIG. 17B). Therefore, the same gain value as the gain value output by gain codebook 57 in the first embodiment is obtained. Note that the other operations of the speech encoding apparatus are the same as those in the first embodiment, and a description thereof will not be repeated.

Next, the mapping unit 11 of the speech decoding apparatus
2 receives a gain code from the separation unit 61, performs mapping processing according to a predetermined rule, and maps a gain code corresponding to the gain code (gain code after mapping).
Is output to the gain codebook 71. However, in the gain codebook 71 according to the fifth embodiment, it is assumed that the rearrangement of the gain codeword based on the evaluation order as in the gain codebook 71 according to the first embodiment is not performed. That is, the storage order of the gain codebook 71 is as shown in FIG.
Is assumed to be as shown in FIG.

The gain codebook 71 has a mapping unit 112
When receiving the mapping gain code from, it outputs the gain codeword stored at the index position corresponding to the mapping gain code. However, in the fifth embodiment, mapping section 112 has a mapping table as shown in FIG. 13 so that the same rearrangement result as the gain codeword in the first embodiment can be obtained.

For example, in the case of the mapping table of FIG. 13, when the mapping unit 112 inputs a gain code of “0”, it outputs a mapping gain code of “1”. As a result, the gain codebook 71 outputs a gain value whose evaluation value rank corresponding to the mapping gain code of “1” is “41” (see FIG. 17B). Therefore, the same gain value as the gain value output by gain codebook 71 in the first embodiment is obtained. Note that the other operations of the speech decoding device are the same as those in the first embodiment, and a description thereof will not be repeated.

As is clear from the above, the fifth embodiment
According to the configuration, the mapping units 111 and 112 for mapping the gain codes and outputting the gain codes after the mapping to the gain codebooks 57 and 71 are provided. This makes it possible to construct a state equivalent to the storage order after the update without updating.

When a plurality of mappings of the mapping units 111 and 112 are prepared and an optimum mapping is used for an error condition to be superimposed on a speech code, a wide range of error conditions can be used without greatly increasing the memory amount. There is an effect that an audio encoding device and an audio decoding device with low quality degradation can be obtained below.

In the fifth embodiment, mapping sections 111 and 112 are provided only in the preceding stage of gain codebooks 57 and 71.
Is shown, but the gain codebooks 52, 6
The mapping units 111 and 112 may also be provided at a stage preceding the step # 6. Further, mapping sections 111 and 112 may be provided only in the preceding stage of gain codebooks 52 and 66.

It is also possible to adopt a configuration in which mapping sections 111 and 112 are introduced at the preceding stage of a codebook other than the gain codebook, and the gain codebook in the speech encoding apparatus shown in FIG. 6 and the speech decoding apparatus shown in FIG. The mapping unit 111, 1
A configuration in which 12 is introduced is also possible.

Further, the mapping unit 11 introduced here
A configuration is also possible in which the mapping of 1,112 is not fixed, and a plurality of mappings are switched and used according to the conditions of an error correction code externally applied to the speech code. For example, if the mode information is strongly protected, the gain codebook 5
If mappings designed to be strong against bit errors are applied independently to each other, if the protection of the mode information is weak, the mapping is performed so as to suppress the deterioration when the mode information is erroneous by the method described above. Can be designed.

FIG. 14 is an explanatory diagram showing two mapping tables when two mappings are switched and used. The first mapping table (see FIG. 14A) is used when the mode information is strongly protected, and is designed so that the gain codebooks 57 and 71 are already alone and resistant to bit errors. By doing so, the sign is not changed by the mapping. The second mapping table (see FIG. 14B) is used when the protection of the mode information is weak, and is the same as the mapping table in FIG. The first mapping table may be omitted, and a method of switching whether or not to perform mapping may be used.

[0144]

As described above, according to the present invention, the storage order of codewords in a plurality of codebooks is rearranged in accordance with the order of evaluation values for codewords in other codebooks. Thus, even if a transmission error occurs and the speech decoding device mistakenly recognizes the mode information, there is an effect that a speech code capable of suppressing deterioration of speech reproduction quality can be generated.

According to the present invention, since the power or the average amplitude of the code word is used as the evaluation value for the code word, a transmission error occurs, and the speech decoding apparatus misidentifies the mode information. Also, there is an effect that it is possible to generate a speech code without significantly deteriorating the power and amplitude of the speech reproduced by the speech decoding device.

According to the present invention, since the plurality of codebooks are configured to be codebooks for outputting excitation gains, even if a transmission error occurs and the speech decoding device misidentifies mode information, the There is an effect that it is possible to generate a speech code that does not greatly degrade the gain value of the speech reproduced by the decoding device.

According to the present invention, the storage order of the codewords of the plurality of codebooks is rearranged so that the total value of the deviations of the evaluation values of the corresponding codewords among the plurality of codebooks is minimized. Transmission error occurs,
Even if the speech decoding device mistakenly recognizes the mode information, the expected value of the degradation related to the predetermined evaluation value becomes small, and as a result, it is possible to generate a speech code capable of suppressing the degradation of the speech reproduction quality. effective.

According to the present invention, when a sound source is generated from a code word and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value. There is an effect that a speech code capable of suppressing deterioration can be generated.

According to the present invention, there is provided a mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, thereby storing codewords of a plurality of codebooks. A configuration equivalent to the storage order after the update is constructed without updating the order in advance based on the rank of the evaluation value, so that it is not necessary to update the storage order of the gain codewords in advance. effective.

According to the present invention, there is provided a mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, thereby storing codewords of a plurality of codebooks. Since the order is configured so that a state equivalent to the storage order after the update is constructed without updating the sum of the deviations of the evaluation values in advance, the storage order of the gain codeword is updated in advance. There is an effect that processing becomes unnecessary.

According to the present invention, since the plurality of codebooks are arranged such that the storage order of the codewords is rearranged in accordance with the order of the evaluation values regarding the codewords of the other codebooks,
Even if the mode information is erroneously recognized, there is an effect that the deterioration of the sound reproduction quality can be suppressed.

According to the present invention, since the power or the average amplitude of the code word is used as the evaluation value relating to the code word, even if the mode information is erroneously recognized, the power and the amplitude of the voice are greatly deteriorated. There is an effect that the sound can be reproduced without the need.

According to the present invention, since the plurality of codebooks are configured to be codebooks for outputting the excitation gain, even if the mode information is erroneously recognized, the voice gain value is not significantly deteriorated without causing significant deterioration of the voice gain value. There is an effect that can be reproduced.

According to the present invention, the storage order of the codewords of the plurality of codebooks is rearranged so that the total value of the deviations of the evaluation values of the corresponding codewords among the plurality of codebooks is minimized. Configuration, even if the mode information is erroneously recognized, the expected value of the deterioration with respect to the predetermined evaluation value decreases,
As a result, there is an effect that deterioration of the sound reproduction quality can be suppressed.

According to the present invention, when a sound source is generated from a codeword and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value. There is an effect that deterioration can be suppressed.

According to the present invention, there is provided a mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, thereby storing codewords of a plurality of codebooks. A configuration equivalent to the storage order after the update is constructed without updating the order in advance based on the rank of the evaluation value, so that it is not necessary to update the storage order of the gain codewords in advance. effective.

According to the present invention, there is provided a mapping means for mapping an index, and at least one or more codebooks output a codeword corresponding to the mapped index, thereby storing codewords of a plurality of codebooks. Since the order is configured so that a state equivalent to the storage order after the update is constructed without updating the sum of the deviations of the evaluation values in advance, the storage order of the gain codeword is updated in advance. There is an effect that processing becomes unnecessary.

According to the present invention, the evaluation values for the codewords of each codebook are examined, and at least one or more codewords of the codebook are stored in accordance with the order of the evaluation values for the codewords of the other codebooks. Since the order is rearranged, even if a transmission error occurs and the speech decoding device mistakenly recognizes the mode information, it is possible to obtain a codebook capable of suppressing deterioration of speech reproduction quality. .

According to the present invention, since the power or the average amplitude of the code word is used as the evaluation value relating to the code word, it is possible to reproduce the sound without causing significant deterioration in the power or amplitude of the sound. This has the effect of obtaining a codebook that can be used.

According to the present invention, since the plurality of codebooks are configured to be codebooks that output excitation gains, a codebook capable of reproducing a sound without causing significant deterioration of a sound gain value. The effect is obtained.

According to the present invention, the total value of the deviations of the evaluation values for the corresponding codewords among a plurality of codebooks is calculated,
Since the storage order of at least one or more codebooks is updated until the total value decreases and is minimized, even if the mode information is erroneously recognized, the expected value of deterioration with respect to a predetermined evaluation value is obtained. Is reduced, and as a result, there is an effect of obtaining a codebook capable of suppressing deterioration of the reproduction quality of audio.

According to the present invention, when a sound source is generated from a code word and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value. There is an effect of obtaining a codebook capable of suppressing deterioration.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a speech encoding device according to Embodiment 1 of the present invention.

FIG. 2 is a configuration diagram showing a speech decoding device according to Embodiment 1 of the present invention.

FIG. 3 is a flowchart showing a codeword arrangement method according to the first embodiment of the present invention.

FIG. 4 is an explanatory diagram illustrating an example of a gain codebook used by a speech encoding device and a speech decoding device.

FIG. 5 is an explanatory diagram showing an example of a speech code output from a multiplexing unit.

FIG. 6 is a configuration diagram illustrating a speech encoding device according to a second embodiment of the present invention.

FIG. 7 is a configuration diagram showing a speech decoding apparatus according to Embodiment 2 of the present invention.

FIG. 8 is an explanatory diagram showing an example of a gain codebook.

FIG. 9 is an explanatory diagram illustrating an example of a speech code output from a multiplexing unit.

FIG. 10 is a configuration diagram showing a codeword arrangement device to which a codeword arrangement method according to a fourth embodiment of the present invention is applied.

FIG. 11 is a configuration diagram illustrating a speech encoding device according to a fifth embodiment of the present invention.

FIG. 12 is a configuration diagram showing a speech decoding apparatus according to Embodiment 5 of the present invention.

FIG. 13 is an explanatory diagram showing a mapping table.

FIG. 14 is an explanatory diagram showing a mapping table.

FIG. 15 is a configuration diagram showing a conventional speech encoding device.

FIG. 16 is a configuration diagram showing a conventional speech decoding device.

FIG. 17 is an explanatory diagram showing an example of a gain codebook used by a conventional speech encoding device and speech decoding device.

[Explanation of symbols]

41 preprocessing unit (encoding unit), 42 spectrum analysis unit (encoding unit), 43 spectrum encoding unit (encoding unit), 44 synthesis filter (encoding unit), 45 subtractor (encoding unit), 46 Auditory weighting unit (encoding unit), 47 distortion minimizing unit (encoding unit), 48 sound source decoding unit (encoding unit), 49 sound source decoding unit (encoding unit), 50 adaptive excitation codebook, 51 driving Excitation codebook, 52 gain codebook, 53 multiplier, 54 multiplier,
55 adder, 56 driving excitation codebook, 57 gain codebook, 58 multiplier, 59 changeover switch (encoding means), 60 multiplexing section (multiplexing means), 61 demultiplexing section (separating means), 62 excitation decoding Part (decoding means), 6
3 Excitation decoder (decoding means), 64 adaptive excitation codebook, 65 driving excitation codebook, 66 gain codebook, 67
Multiplier, 68 multiplier, 69 adder, 70 excitation codebook, 71 gain codebook, 72 multiplier, 73
Changeover switch (decoding means), 74 spectrum decoding section (decoding means), 75 synthesis filter (decoding means), 76 post-processing section (decoding means), 81 sound source mode selection section (coding means), 82 Changeover switch (encoding means), 83 multiplier (encoding means), 84 multiplier (encoding means), 85 adder (encoding means), 91 sound source mode selection section (decoding means), 92 changeover switch (encoding means) Decoding means), 93 multiplier (decoding means), 94 multiplier (decoding means), 95 adder (decoding means), 101
Driving excitation codebook, 102 Driving excitation codebook, 103 synthesis filter, 104 synthesis filter, 105 distance calculation unit, 106 codeword replacement unit, 111 mapping unit (mapping unit), 112 mapping unit (mapping unit).

Claims (19)

[Claims] 1. A codebook corresponding to mode information is selected from a plurality of codebooks that output a codeword corresponding to an index, and input speech is converted for each frame using a codeword output by the codebook. In a speech encoding apparatus provided with an encoding unit for encoding and a multiplexing unit for multiplexing an encoding result of the encoding unit into a bit string, the plurality of codebooks are related to codewords of another codebook. A speech coding apparatus, wherein the storage order of codewords is rearranged in accordance with the order of evaluation values. 2. The speech encoding apparatus according to claim 1, wherein the power or the average amplitude of the code word is used as the evaluation value for the code word. 3. The codebook according to claim 1, wherein the plurality of codebooks are codebooks that output excitation gains.
A speech encoding device according to claim 1. 4. A codebook corresponding to mode information is selected from a plurality of codebooks that output a codeword corresponding to an index, and input speech is converted for each frame using a codeword output by the codebook. In a speech coding apparatus comprising coding means for coding, and multiplexing means for multiplexing the coding result of the coding means into a bit string, an evaluation value for each code word corresponding to the plurality of codebooks Wherein the storage order of the codewords of the plurality of codebooks is rearranged such that the total value of the deviations of the codebooks becomes minimum. 5. A method according to claim 1, wherein when a sound source is generated from a code word and a synthesized sound is generated from the sound source, an expected value relating to the synthesized sound is treated as an evaluation value. The speech encoding device according to claim 1. 6. A mapping means for mapping an index, wherein at least one or more codebooks output a codeword corresponding to the mapped index, thereby preliminarily evaluating the storage order of the codewords of the plurality of codebooks. 2. The speech coding apparatus according to claim 1, wherein a state equivalent to the storage order after the update is constructed without updating based on the order of the values. 7. A storage unit for mapping an index, wherein at least one or more codebooks output codewords corresponding to the index after the mapping, thereby preliminarily evaluating the storage order of the codewords of the plurality of codebooks. The speech encoding apparatus according to claim 4, wherein a state equivalent to the storage order after the update is constructed without updating the sum of the deviations of the values to be a minimum. 8. A separating means for separating an index from an encoding result multiplexed into a bit string, and an arbitrary codebook among a plurality of codebooks for outputting a codeword corresponding to the index separated by the separating means. And a decoding means for decoding the encoding result using the codeword output by the codebook, wherein the plurality of codebooks are codes of another codebook. A speech decoding apparatus, wherein the storage order of codewords is rearranged in accordance with the order of evaluation values for words. 9. The speech decoding apparatus according to claim 8, wherein the power or the average amplitude of the code word is used as the evaluation value for the code word. 10. The speech decoding apparatus according to claim 8, wherein the plurality of codebooks are codebooks that output excitation gains. 11. A separating means for separating an index from an encoding result multiplexed into a bit string, and an arbitrary codebook among a plurality of codebooks for outputting a codeword corresponding to the index separated by the separating means. And a decoding unit that decodes the encoded result using the codeword output by the codebook.
The speech order, wherein the storage order of the codewords of the plurality of codebooks is rearranged so that the total value of the deviations of the evaluation values for the corresponding codewords among the plurality of codebooks is minimized. Decryption device. 12. A method according to claim 8, wherein when a sound source is generated from a code word and a synthesized sound is generated from the sound source, an expected value related to the synthesized sound is treated as an evaluation value. The speech decoding device according to claim 1. 13. A mapping means for mapping an index, wherein at least one or more codebooks output codewords corresponding to the mapped index, thereby preliminarily evaluating the storage order of the codewords of the plurality of codebooks. 9. The speech decoding apparatus according to claim 8, wherein a state equivalent to the storage order after the update is constructed without updating based on the order of the values. 14. A method for mapping indices, wherein at least one or more codebooks output codewords corresponding to the indices after mapping, thereby pre-evaluating the storage order of codewords in a plurality of codebooks. 12. The speech decoding apparatus according to claim 11, wherein a state equivalent to the storage order after the update is constructed without updating the sum of the deviations of the values to be a minimum. 15. When a plurality of codebooks for outputting a codeword corresponding to an index are mounted on a speech encoding device or a speech decoding device, an evaluation value relating to a codeword of each codebook is examined, and another codebook is examined. A codeword arrangement method for rearranging the storage order of codewords of at least one or more codebooks in accordance with the order of evaluation values of the codewords. 16. The codeword arrangement method according to claim 15, wherein the power or the average amplitude of the codeword is used as the evaluation value for the codeword. 17. The codeword arrangement method according to claim 15, wherein the plurality of codebooks are codebooks that output excitation gains. 18. When a plurality of codebooks for outputting codewords corresponding to an index are mounted on a speech coding apparatus or a speech decoding apparatus, deviations of evaluation values of the codebooks corresponding to the plurality of codebooks are provided. A code word arrangement method for calculating the total value of the code words and updating the storage order of the code words of at least one or more code books until the total value decreases and minimizes. 19. The code according to claim 15, wherein, when a sound source is generated from the code word and a synthesized sound is generated from the sound source, an expected value related to the synthesized sound is treated as an evaluation value. Word alignment method.