JP2002221998A - Method, device and program for encoding and decoding acoustic parameter and voice - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize vector encoding/decoding having low quality degradation of LSP parameters in a silence interval or a steady state noise interval. SOLUTION: While conducting vector encoding/decoding of LSP parameters of moving average type voice, a vector of the spectrum equivalent to a steady noise interval or a vector from which beforehand obtained average value vector of a voice interval is subtracted, is stored as one vector C0 of a vector coding table 14A and a spectrum equivalent to a silence interval or a steady noise is outputted as one of the coded vectors.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a low bit rate audio signal encoding / decoding method and apparatus in a mobile communication system, the Internet, etc., for encoding and transmitting audio signals such as voice signals and music signals. The present invention relates to a method and apparatus for encoding and decoding acoustic parameters, and a program for executing these methods on a computer.

[0002]

2. Description of the Related Art In the field of digital mobile communication and voice storage, a voice coding apparatus for compressing voice information and coding it with high efficiency has been used for effective use of radio waves and storage media. Such a speech coding apparatus uses a method using a model suitable for representing a speech signal so that a speech signal of high quality can be represented even at a low bit rate. For example, CELP (Code Excited)
Linear Prediction (Code Excitation Linear Prediction Coding). For more information on CELP technology, see MRSchroeder an
d BSAtal: "Code-Excited Linear Prediction (CEL
P): High-quality Speech at Very Low BitRates ", Pro
c. ICASSP-85, 25.1.1, pp. 937-940, 1985 ".

[0003] The CELP type speech coding system is based on a speech synthesis model corresponding to a human vocal utterance mechanism. Synthesize the audio signal. More specifically, the digitized audio signal is divided for every certain frame length (about 5 ms to 50 ms), linear prediction of the audio signal is performed for each frame, and a prediction residual (excitation signal) by linear prediction is known. The coding is performed using an adaptive code vector composed of the following waveforms and a fixed code vector.
The adaptive code vector is stored in the adaptive codebook as a vector representing a drive excitation signal generated in the past, and is used to represent a periodic component of the audio signal. The fixed code vector is stored as a vector having a predetermined number of waveforms prepared in advance in the fixed code book, and is used to mainly express aperiodic components that cannot be expressed by the adaptive code book. The vectors stored in the fixed codebook include:
A vector composed of a random noise sequence, a vector expressed by a combination of several pulses, and the like are used.

An algebraic fixed codebook is one of the typical fixed codebooks that expresses the fixed code vector by a combination of several pulses. Specific contents of the algebraic fixed codebook are shown in "ITU-T Recommendation G.729" and the like. In conventional speech coding systems, speech linear prediction coefficients are converted into parameters such as partial autocorrelation (PARCOR) coefficients and line spectrum pairs (LSP: Line Spectrum Pairs, also called line spectrum frequency), and then quantized. After being converted into a digital code, it is stored or transmitted. Details of these methods are described in, for example, "Digital Speech Processing" by Sadahiro Furui (Tokai University Press).

In encoding the linear prediction coefficients, LSP
As a parameter encoding method, a weighted vector obtained by multiplying a code vector output from a vector codebook in one or more past frames by a weighting factor selected from a weighting factor codebook, or this vector is obtained in advance. The quantization parameter of the current frame is represented by a vector obtained by adding the average vector of the LSP parameters of the entire speech signal, and the distortion of this quantization parameter with respect to the LSP parameter obtained from the input speech, that is, the quantization distortion is A code vector to be output by the vector codebook and a weight coefficient set to be output by the weight coefficient codebook are selected so as to be minimum or sufficiently small, and are output as codes of LSP parameters.

[0006] Generally, if weighted vector quantization or weighting coefficients are considered as prediction coefficients from the past,
It is called moving average (MA) moving vector prediction vector quantization. On the decoding side, the received vector code and the weight coefficient code are used to multiply the code vector of the current frame and the past code vector by a weight coefficient, or to add an average vector of LSP parameters of the entire audio signal which is obtained in advance. The resulting vector is output as a quantization vector of the current frame.

A vector codebook that outputs a code vector of each frame includes a basic one-stage vector quantizer, a divided vector quantizer that divides the dimension of a vector,
Two or more multi-stage vector quantizers, or
A configuration such as a multi-stage / division vector quantizer in which a multi-stage and a division vector quantizer are combined is possible.

[0008]

In the above-mentioned conventional LSP parameter encoder / decoder, the number of frames is large in the silent section and the stationary noise section, and the encoding and decoding processes have a multi-stage configuration. A vector cannot be output so that the parameters synthesized corresponding to a silent section or a stationary noise section change smoothly.
This is because the vector codebook used for encoding is usually obtained by learning, but in this learning, a sufficient amount of a silent section or a stationary noise section is not included in the training speech. If the vector corresponding to the section cannot always be sufficiently reflected and learned, or if the number of bits given to the quantizer is small, a codebook sufficiently containing these quantized vectors corresponding to the non-speech section. Could not be designed.

[0009] In such an LSP parameter encoder / decoder, the quantization performance in a non-voice section cannot be sufficiently exhibited in the coding at the time of actual communication, and the quality of reproduced sound cannot be degraded. Such a problem has occurred not only in the encoding of acoustic parameters equivalent to linear prediction coefficients representing the spectral envelope of a speech signal, but also in the same encoding of a music signal. An object of the present invention has been made in view of the above point, and corresponds to a silent section and a stationary noise section in encoding / decoding of an acoustic parameter equivalent to a linear prediction coefficient representing a conventional spectral envelope of an acoustic signal. By facilitating the output of the vector, an audio parameter encoding / decoding method and apparatus with less deterioration in quality in these sections, an audio signal encoding / decoding method and apparatus using them, and these methods are described. A computer-implemented program is provided.

[0010]

SUMMARY OF THE INVENTION The present invention provides an acoustic parameter equivalent to a linear prediction coefficient representing a spectral envelope of an acoustic signal, that is, a parameter (hereinafter, simply referred to as an acoustic parameter) such as an LSP parameter, an α parameter, and a Percoll parameter. In encoding / decoding, an acoustic parameter vector representing a substantially flat spectrum envelope corresponding to a silent section or a stationary noise section, which is not originally obtained by learning the codebook, is added to the codebook by adding the code vector to the codebook.
One feature is that it is selectable. The prior art is that a vector including a component of an acoustic parameter vector representing a substantially flat spectrum envelope is obtained in advance by calculation, and stored as one vector of a vector codebook. The difference is that the split vector quantization configuration is configured to output the code vector.

[0011] The acoustic parameter encoding method according to the present invention includes: (a) calculating an acoustic parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic of an acoustic signal for each frame having a fixed time length; One or more sets of code vectors output from at least one frame in the closest past and a code vector selected in the current frame from a vector codebook that stores code vectors corresponding to indices representing them. Weighted coefficients of the set selected from the coefficient codebook stored in correspondence with the indices representing the weighted coefficients are multiplied and added to generate a weighted vector, and a vector including the components of the weighted vector is generated. Is determined as a candidate of a quantized acoustic parameter for the above acoustic parameter of the current frame. And (c) a set of code vectors of the vector codebook and weight coefficients of the coefficient codebook, using a criterion such that distortion of the candidates for the quantized acoustic parameters with respect to the calculated acoustic parameters is minimized. And determining and outputting an index representing the set of the determined code vector and the weight coefficient as the quantization code of the acoustic parameter, wherein the vector codebook stores one of the stored code vectors. First, it includes a vector including components of an acoustic parameter vector representing a substantially flat spectrum envelope.

A method of decoding an acoustic parameter according to the present invention comprises the steps of: (a) storing a plurality of code vectors of an acoustic parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic of an acoustic signal in correspondence with an index representing them; From a codebook and a coefficient codebook in which one or more sets of weighting factors are stored in correspondence with indexes representing those sets, a code vector corresponding to an index represented by a code input for each frame and one set of weights Outputting a coefficient; and (b) outputting the code vector output from the vector codebook in at least one frame in the closest past and the code vector output from the vector codebook in the current frame, respectively. Multiplies the weighted coefficients of a given set and adds them to generate a weighted vector Outputting the vector containing the components of the weighted vector as a decoded quantized vector of the current frame, wherein the vector codebook has a substantially flat spectrum as one of the stored code vectors. Includes a vector containing the components of the acoustic parameter vector representing the envelope.

An audio parameter encoding apparatus according to the present invention analyzes an input audio signal for each frame and calculates an audio parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic of the audio signal, and a plurality of code vectors. And a coefficient codebook that stores one or more sets of weighting coefficients corresponding to the indexes that represent those sets, and a vector codebook that stores one or more weighting coefficients corresponding to the indexes that represent those sets. The code vector for the current frame and the code vector output in at least one frame in the closest past are multiplied by each of the weight coefficients of the set selected from the coefficient code book, and the weighted vector is added. A vector including the components of the generated weighted vector is generated. A quantization parameter generating unit that outputs as a candidate of a quantized acoustic parameter for the acoustic parameter of the game, a distortion calculating unit that calculates a distortion of the quantized acoustic parameter with respect to the acoustic parameter calculated by the parameter calculating unit, Using a criterion such that the distortion is minimized, a code vector of the vector codebook and a weight coefficient of the set of coefficient codebooks are determined, and an index representing each of the determined code vector and weight coefficient set. As a code of the acoustic parameter, and the vector codebook is configured to include, as one code vector, a vector including a component of an acoustic parameter vector representing a substantially flat spectrum envelope. Have been.

An acoustic parameter decoding apparatus according to the present invention includes a vector codebook storing a plurality of code vectors of acoustic parameters equivalent to linear prediction coefficients representing a spectral envelope characteristic of an acoustic signal in correspondence with indices representing them. One code vector is output from the vector codebook according to a coefficient codebook in which one or more sets of weighting coefficients are stored in correspondence with indices representing them and an index represented by a code input for each frame; Outputting a set weighting coefficient from the coefficient codebook, and weighting the set output in the current frame to the code vector output in the current frame and the code vector output in at least one frame in the closest past. Generates a weighted vector by multiplying each coefficient and adding them together. Means for outputting a vector including a component of the weighted vector as a decoded quantized acoustic parameter of the current frame, and the vector codebook represents a substantially flat spectrum envelope. A vector including a component of the acoustic parameter vector is stored as one of the code vectors.

An audio signal encoding apparatus for encoding an input audio signal according to the present invention includes means for encoding the spectral characteristics of the input audio signal using the audio parameter encoding method, and a periodic component of the input audio signal. An adaptive codebook holding an adaptive codebook representing a fixed codebook storing a plurality of fixed vectors, and an adaptive codebook generated from the adaptive codebook from the adaptive codebook and the fixed vector from the fixed codebook. Input the sound source vector as an excitation signal,
Filter means for synthesizing a synthesized audio signal using a filter coefficient based on the quantized audio parameter; and selecting from the fixed codebook and the adaptive codebook such that distortion of the synthesized audio signal with respect to the input audio signal is reduced. Means for determining an adaptive code vector and a fixed vector to perform, and outputting an adaptive code and a fixed code corresponding to the determined adaptive code vector and the fixed vector, respectively.

An audio signal decoding apparatus for decoding an input code and outputting an audio signal according to the present invention uses the above-described audio parameter decoding method and is equivalent to a linear prediction coefficient representing a spectral envelope characteristic from an input code. Means for decoding acoustic parameters, a fixed codebook storing a plurality of fixed vectors, an adaptive codebook holding an adaptive code vector indicating a periodic component of the synthesized audio signal, and an input adaptive code and fixed code, A corresponding fixed vector is taken out from the fixed code book, a corresponding adaptive code vector is taken out from the adaptive code book, a means for generating an excitation vector by combining these, and a filter coefficient based on the acoustic parameter are set, Filter means for reproducing an acoustic signal using the excitation vector.

An audio signal encoding method for encoding an input audio signal according to the present invention comprises the steps of: (A) encoding the spectral characteristics of the input audio signal using the audio parameter encoding method; As an excitation signal, an excitation vector generated based on an adaptive code vector from an adaptive code book holding an adaptive code vector indicating a periodic component of an acoustic signal and a fixed vector from a fixed code book storing a plurality of fixed vectors is used as an excitation signal. (C) generating a synthetic audio signal by performing a synthetic filter process with a filter coefficient based on the quantized audio parameter; And an adaptive code vector and a fixed vector to be selected from the adaptive codebook and the adaptive code vector. Step of outputting a corresponding adaptive code and a fixed code respectively torr, including capital.

An audio signal decoding method for decoding an input code and outputting an audio signal according to the present invention comprises the following steps: And (B) extracting adaptive code vectors from the adaptive codebook holding the adaptive code vectors representing the periodic components of the input audio signal by the input adaptive code and fixed code, and Extracting a corresponding fixed vector from the fixed codebook storing the fixed vector, generating an excitation vector by vector synthesis of the adaptive code vector and the fixed vector, and (C) using a filter coefficient based on the acoustic parameter. Reproducing the synthesized acoustic signal by performing a synthesis filter process on the excitation vector.

The above-described invention can be provided in the form of a computer-executable program. According to the present invention, a weighted vector quantizer (or M
A predictive vector quantizer) previously obtains and stores a vector including a component of an acoustic parameter vector representing a substantially flat spectrum envelope as a code vector of a vector codebook. It is possible to output a quantization vector corresponding to an acoustic parameter corresponding to a noise section.

According to another embodiment of the present invention,
When a multi-stage vector codebook is used as the configuration of the vector codebook of the acoustic parameter encoder / decoder, the component of the acoustic parameter vector representing a substantially flat spectrum envelope is added to the one-stage codebook. By storing a vector including the corresponding vector and storing a zero vector in the codebook of the other stage, a quantized vector corresponding to an acoustic parameter corresponding to a corresponding silent section or a stationary noise section is output. can do. The zero vector need not always be stored. When the zero vector is not stored, when a vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is selected from the codebook of one stage, the component of the acoustic parameter vector representing the substantially flat spectrum envelope is selected. Need only be output as code vector candidates for the current frame.

When the vector codebook is composed of divided vector codebooks, a plurality of divided vectors obtained by dividing the dimensions of a vector including components of an acoustic parameter vector representing a substantially flat spectrum envelope are used. Are distributed and stored one by one in a plurality of divided vector codebooks. In the search of each divided vector codebook, each divided vector is selected, and a vector obtained by integrating the divided vectors is assigned to a corresponding silent section or a stationary section. It can be output as a quantization vector corresponding to an acoustic parameter corresponding to a noise section.

Further, the vector quantizer has a multi-stage / division vector quantization configuration, and by combining the above-mentioned multi-stage vector quantization configuration and the technology of the division vector quantization configuration, a corresponding silence section or a stationary noise section is obtained. Can be output as a quantization vector corresponding to the acoustic parameter to be processed. When the codebook has a multi-stage configuration, a scaling coefficient for each of the second and subsequent codebooks is provided as a scaling coefficient codebook corresponding to each code vector of the first stage codebook. By reading the scaling coefficient corresponding to the selected code vector in the codebook from each of the scaling coefficient codebooks and multiplying each of the selected code vectors from the second-stage codebook, encoding with less quantization distortion can be realized. .

As described above, it is possible to provide an audio parameter encoding / decoding method and an apparatus thereof, which are objects of the present invention and have less quality deterioration in the section. In the acoustic signal encoding device of the present invention, any of the parameter encoding devices is used in quantization of the linear prediction coefficient in an acoustic parameter area equivalent to the linear prediction coefficient. According to this configuration, the same operation and effect as any of the above can be obtained. In the acoustic signal decoding device according to the present invention, in decoding the linear prediction coefficient, any one of the parameter decoding devices is used in an acoustic parameter area equivalent to the linear prediction coefficient. According to this configuration, the same function and effect as any of the above can be obtained.

[0024]

Embodiment 1 Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of an acoustic parameter encoding device according to one embodiment to which a linear prediction parameter encoding method according to the present invention is applied. This encoding device includes a linear prediction analysis unit 12, an LSP parameter calculation unit 13, and a codebook 14, a quantization parameter generation unit 15, a distortion calculation unit 16, and a codebook search control unit 17 that constitute the parameter encoding unit 10. It is composed of In the figure, an input terminal T
From 1, for example, a series of digitized audio signal samples is input. The linear prediction analysis unit 12 performs a linear prediction analysis on the audio signal samples for each frame stored in the internal buffer, and calculates a set of linear prediction coefficients. In the following description, underlined symbols represent vectors. Now, let the order of the linear prediction analysis be p
Then, the LSP parameter calculation unit 13 calculates the equivalent p-order LSP (line spectrum pair) from the p-th linear prediction coefficient.
Calculate the parameters. The details of these processing methods are described in the aforementioned book by Furui. The p LSP parameters are expressed as vectors as f (n) = (f ₁ (n), f ₂ (n), ..., f _p (n)) (1). Here, the integer n is
The number n of a certain frame is shown, and the frame at that time is called a frame n.

The codebook 14 contains the LSP data obtained by learning.
Stores N code vectors representing parameter vectors
Vector codebook 14A and K weighting factor sets
And a coefficient codebook 14B in which
Index Ix (n) specifying the signal vector and the weighting factor
By the index Iw (n) that specifies the sign,
The corresponding sign vectorx(n) and weight coefficient set (w₀, w₁,
 ..., w_m) Is output. The quantization parameter generation unit 15
M buffer units 15B connected in series₁, ..., 15B_m, M +
One multiplier 15A₀, 15A₁, ..., 15A_m, Register 15C,
It comprises a vector adder 15D. Vector codebook 14A
Of the current frame n selected as one of the candidates
Khutorx(n) and the buffer unit 15B₁~ 15B_mStored in
Code vector determined for past frames n-1, ..., n-m
Torrx(n-1), ...,x(n-m) and the sign vector respectively
Multiplier 15A₀, ..., 15A_mWeight coefficient w of the set selected in₀,
 ..., w _mAre multiplied, and the result of these multiplications is an adder
The audio signal is added at 15D and further obtained in advance.
Average vector of all LSP parametersy _aveIs register 1
5C is added to the adder 15D. In this way,
From the calculator 15D, the candidate of the quantization vector, that is, the LSP parameter
Data candidatesy(n) is generated. Mean vectory _aveAs
May use the average vector of voiced parts, or
The zero vector may be used as follows.

Vector codebook 1 for the current frame n
The code vector x (n) selected from 4A is defined as x (n) = (x ₁ (n), x ₂ (n), ..., x _p (n)) (2), and similarly, one frame Let x (n-1) be the previous determined code vector, and 2
x (n-2), the determined code vector m frames before is defined as x (n-
m), the quantization vector candidate y (n) = (y ₁ (n), y ₂ (n), ..., y _p (n)) (3) of the current frame is y (n ) = W ₀ × x (n) + Σ _{j = 1} ^m w _j × x (nj) + y _ave (4) Here, the larger the value of m, the higher the quantization efficiency, but the effect when a code error occurs extends to m frames later. Since it is necessary to go back to the past, m is appropriately selected as needed. For voice communication,
In the case of about 20 ms per frame, a value of m of 6 or less is sufficient, and a value of 1 to 3 may be used. This m is also called the order of the moving average prediction.

The thus obtained quantization vector candidatesy
(n) is sent to the distortion calculator 16 and calculated by the LPS parameter.
LSP parameters calculated by output unit 13fQuantization for (n)
Calculate the distortion. The distortion d is, for example, the following weight
Is defined by the attached Euclidean distance. d = Σ_{i = 1} ^pr_i(f_i(n) −y_i(n))^Two (5) where r_i, I = 1,…, p are LSP parametersf(n)
Weighting factor near the formant frequency of the spectrum
The performance is good when the weight is set so as to place importance on.

The codebook search unit 17 sequentially changes the set of indices Ix (n) and Iw (n) given to the codebook 14, and for each set of indices, calculates the distortion d of the equation (5) as described above. Is repeated from the code vector of the vector codebook 14A in the codebook 14 and the weight coefficient set of the coefficient codebook 14B to minimize or sufficiently reduce the distortion d output from the distortion calculator 16. A search is made for the index, and the indexes Ix (n) and Iw (n) are transmitted from the terminal T2 as the sign of the input LSP parameter. The codes Ix (n) and Iw (n) sent from the terminal T2 are sent to a decoder via a transmission line or stored in a storage device.

[0029] When the output code vector x of the current frame (n) is determined, the past frame buffer section 15B _j (nj)
The code vector x (nj), j = 1, ..., m-1 of the current frame n is sequentially sent to the next buffer unit 15B _{j + 1.}
x (n) is input to the buffer unit 15B _1. The feature of the present invention is that one code vector stored in the vector codebook 14A used in the above-described weighted vector quantization of the LSP parameter or the encoding by the moving average vector quantization is described above. Mean vector y _ave
Y but if it is zero, the LSP parameter vector F corresponding to the silent interval or the stationary noise interval, or, if y _ave is not zero, from the LSP parameter vector F
_This is to store the vector C _{0 from} which _ave has been subtracted. In other words, when y _ave is not zero, if the LSP parameter vector corresponding to a silent section or a stationary noise section is F = (F ₁ , F ₂ ,..., F _p ), the vector codebook 14A in FIG. code vector C ₀ to be stored in the, C ₀ = F - calculated as y _ave. Assuming that C ₀ is continuously selected as a code vector over m frames in coding by a moving average prediction in a silent section or a stationary noise section, a quantization vector y (n) becomes y (n) = w ₀・x (n) + Σ _{j = 1} ^m w _j・x (nj) + y _ave = w ₀・C ₀ + Σ _{j = 1} ^m w _j・C ₀ + y _ave = (w ₀ + Σ _{j = 1} ^m w _j ) · C ₀ + y _ave . Here, the sum of the weighting factors from w ₀ to w _m 1 or it When a value close, y (n) is, C ₀ +
y _ave , that is, calculated from the LSP parameter of the silent section
F or a vector close to F 2 can be output as a quantization vector, and coding performance in a silent section or a stationary noise section can be improved. With the above configuration, the vector codebook 14A stores a vector including the component of the vector F as one code vector. When the quantization parameter generation unit 15 generates the quantization vector y (n) including the component of the average vector y _ave as the code vector including the component of the vector F , the average vector y _ave is subtracted from the vector F. When a quantized vector y (n) that does not include a component of the average vector y _ave is generated by using the vector F , the vector F itself is used.

FIG. 2 is a block diagram of a decoding system according to an embodiment of the present invention.
Is a configuration example of a quantization apparatus, and includes a codebook 24 and a quantization parameter.
And a generation unit 25. These codebooks 24
The quantization parameter generation unit 25 is provided in the encoding device of FIG.
Codebook 14 and quantization parameter generator 15
It is configured similarly. Sent from the encoding device of FIG.
Indexes Ix (n) and Iw (n) as parameter codes
Code vector corresponding to the index Ix (n)x
(n) is output from the vector codebook 24A.
Weight coefficient set w corresponding to dex Iw (n)₀, w₁, ..., w_m
Is output from the coefficient codebook 24B. Vector codebook
Code vector output for each frame from 24Ax(n)
Is the buffer unit 25B connected in series₁, ..., 25B_mInput sequentially
Is done. Code vector of current frame nx(n) and the buffer
Part 25B₁, ..., 25B_m1, ..., m frame past code vector
x(n-1), ...,x(n-m) and weighting factor w₀, w₁, ..., w_mThe multiplier 2
5A₀, 25A₁, ..., 25A_mAnd multiply these multiplication results by an adder
25D, and further stored in the register 25C in advance.
Vector of LSP parameters of the entire speech signaly _{av e}
Is added to the adder 25D.
Torry(n) is output as the decoding LSP parameter.y
_aveIs the mean vector or zero vector of voiced partszWhen
It is also possible to keep.

According to the present invention, even in this decoding device,
As in the encoding device shown in FIG. 1, by storing the vector C ₀ as one code vector in the vector codebook 24A, the LSP parameter vector F obtained in the silent section or the stationary noise section of the acoustic signal can be obtained. Can be output. Adder 15D in FIG. 1, adder 25D in FIG.
_Does not add the average vector y _ave (set to zero vector)
If, LSP parameter vector F corresponding to the silent interval or the stationary noise region in the vector codebook 14A, 24A are stored as one of the code vectors in place of the vector C _0. In the following description, each vector codebook 14A,
LSP parameter vector F or vector stored in 24A
Denoted as a vector C ₀ on behalf of C _0.

FIG. 3 shows the vector codebook 14A in FIG.
Alternatively, a configuration example of the vector codebook 24A in FIG. 2 is shown as a vector codebook 4A. In this example, a single-stage vector codebook 41 is used. In the vector codebook 41, N code vectors x ₁ ,..., X _N are stored as they are, and the input index Ix One of the N code vectors is selected and output according to (n). In the present invention, the code vector C ₀ is used as one of the code vectors x . Vector codebook 41
While the N code vectors produced by learning like the example prior art, either in this invention, the most similar (low distortion) one vector to the vector C ₀ of which are replaced by C _0, or Simply added.

VectorC ₀There are several ways to ask
is there. One is usually a silent section or a stationary noise section.
In between, the spectral envelope of the input audio signal is flat
Therefore, for example, a p-order LSP parameter vectorFPlace
Π / (1 + p), 2π / (1 + p), ..., π
P values of almost equal size such as / (1 + p)
It may be used as a meter vector. Or silence
Actual LSP parameter vector in the interval and stationary noise intervalFOr
LaC ₀=F-y _aveAsk by Alternatively, white noise or Hot
h Input noise and find its LSP parameter,
vectorFUsed asC ₀=F-y _aveMay be required. What
Average vector of LSP parameters of the entire audio signaly
_aveIs generally the code vector of the vector codebook 41xTo
When learning, as the average vector of all learning vectors
Ask for it. P = 10th order LSP parameter as acoustic parameter
LSP parameters in a silent section or a stationary noise section
10-dimensional vector with data normalized to values between 0 and π
C ₀,y _ave as well asFIs shown in Table 1 below.

[0034]

[Table 1] A vector F is an example of a code vector of an LSP parameter representing a silent section and a stationary noise section written in the codebook according to the present invention. The values of the elements of this vector increase at substantially constant intervals, which means that the frequency spectrum is almost flat. Embodiment 2 FIG. 4 shows another example of the configuration of the vector codebook 14A of the LSP parameter encoder of FIG. 1 or the vector codebook 24A of the LSP parameter decoding device of FIG. This is the case where a book is used. The first-stage codebook 41 stores N p-dimensional code vectors x ₁₁ ,..., X _1N, and the second-stage codebook 42 stores N ′ p-dimensional code vectors x ₂₁ , ..., x _{2N '} are stored.

First, an index for designating a code vector
When the code Ix (n) is input, the code analysis unit 43
The Ix (n) is analyzed and the first-stage code vector is specified.
The index Ix (n) to determine₁And the second stage code vector
Index Ix (n) _TwoGet. And of each stage
Index Ix (n)₁, Ix (n)_TwoI-th corresponding to
And the i'th code vectorx _1i,x _{2i '}To the first stage codebook 4
Read from the first and second stage codebooks 42,
And add both code vectors, and
xOutput as (n).

In the case of the vector codebook having the two-stage configuration, the code vector search is performed by using only the first-stage codebook 41 up to a predetermined number of candidate code vectors in ascending order of quantization distortion. This search is performed in the coefficient codebook 1 shown in FIG.
This is performed in combination with the 4B weight coefficient set. Next, for the combination of the first-stage code vector of each candidate and the respective code vectors of the second-stage codebook, a combination of code vectors that minimizes quantization distortion is searched. As described above, when searching for a code vector by giving priority to the first-stage codebook 41, the code vector C ₀ (or, as one code vector in the first-stage codebook 41 of the multi-stage vector codebook 4A)
F ) is stored in advance, and a zero vector z is stored in advance as one code vector in the codebook 42 in the second stage. As a result, the code vector C ₀
If but a selected, codebook 42 the zero vector z is selected from, as a result, as the output of codebook 4A from the adder 44, and the silent interval, the code vector C ₀ in the case of corresponding to the stationary noise region output A configuration that can be implemented is realized. The zero vector z may not be stored, and when the code vector C ₀ is selected from the code book 41, the selection and addition from the code book 42 may not be performed.

Each code vector of the first-stage codebook 41 and 2
When searching for all combinations of each code vector in the second codebook, the code vector C ₀ and the zero vector z
May be stored in either codebook as long as they are different from each other. The code vector C ₀ and the zero vector z are likely to be selected at the same time in a silent section or a stationary noise section, but they may not always be selected at the same time due to calculation errors and other factors. In the codebook of each stage, the code vector C ₀ and the zero vector z are to be selected as one code vector, similarly to the selection of other code vectors.

The zero vector may not be stored in the second-stage codebook 42. In that case, the vector from the first stage codebook 41
When C ₀ is selected, without selecting from the code book 42 code vectors may be directly outputs the code C ₀ in codebook 41 from the adder 44. By configuring the codebook 4A with a multi-stage codebook as shown in FIG. 4, it is effectively the same as providing code vectors by the number of selectable code vector combinations. There is an advantage that the size of the codebook (here, the total number of code vectors) can be reduced as compared with the case of only the codebooks of the stage. FIG.
Shows a case in which the vector codebooks are composed of two stages of vector codebooks 41 and 42. However, when the number of stages is three or more, the codebooks are simply added by the number of additional stages, and the codebooks of the respective stages are encoded by the index for each stage. Since only the vectors are selected and the vectors are synthesized, they can be easily extended. Embodiment 3 FIG. 5 shows the vector codebook 4A of the embodiment of FIG.
Each code vector of the first-stage codebook 41 is multiplied by a predetermined scaling coefficient to a code vector selected from the second-stage codebook 42, and added to the code vector from the first-stage codebook 41. Output. Scaling factor codebook 45 are provided, each of the code vectors x ₁₁ of the first-stage codebook 41, ..., C _0, ..., for example, about 0.5 to 2 was determined by pre-learning in response to x _1N scaling factor s ₁ of, ..., s _n is stored, is accessed by the first-stage codebook 41 and common index Ix (n) _1.

First, when an index Ix (n) designating a code vector is input, the code analysis unit 43 analyzes the index Ix (n), and an index Ix (n) designating a first-stage code vector. n) ₁ and an index Ix (n) ₂ specifying the second-stage code vector are obtained. Index Ix
(n) reading the code vector x _1i corresponding to _one from the first-stage codebook 41. Also, from the scaling coefficient codebook 45,
The scaling factor s _i corresponding to the index Ix (n) ₁
Is read. Next, the code vector x _{2i ′} corresponding to the index Ix (n) ₂ is read from the second-stage codebook 42, and the scaling coefficient s _i is calculated in the multiplier 46 by the code vector x _{2i ′} from the second-stage codebook 42. Multiply by A vector obtained by multiplying the code vector x _1i from the first-stage codebook 41 are added in the addition unit 44, and outputs the addition result as the code vector x (n) from the codebook 4A.

In this embodiment as well, the search for the code vector firstly searches for a predetermined number of candidate code vectors in order from the one with the smallest quantization distortion using only the first-stage codebook 41. Next, each candidate code vector and the second-stage codebook 42
Is searched for a combination that minimizes the quantization distortion for each combination with the respective code vectors. In this case, for the multistage vector codebook 4A with the scaling coefficient, the vector C _{0 is used} as one code vector in the first stage codebook 41.
Is stored in advance, and a zero vector z is stored in advance as one code vector of the codebook 42 in the second stage. As in the case of FIG. 4, if the search is performed for all combinations between the code vectors of the two code books 41 and 42, if the code vector C ₀ and the zero vector z are stored in separate code books, Either may be stored. Alternatively, the zero vector z need not be stored in the codebook as in the above-described embodiment. If not stored, sign vector
When C ₀ is selected, selection addition from the codebook 42 is not performed.

In this way, a code vector corresponding to a silent section or a stationary noise section can be output. The code vector C ₀ and the zero vector z are likely to be selected simultaneously in a silent section or a stationary noise section, but may not always be selected simultaneously due to a calculation error or other relations. In the codebook at each stage, the code vector C ₀ and the zero vector z are to be selected as one code vector similarly to the selection of other code vectors. Using the scaling coefficient codebook 45 as in the embodiment of FIG. 5 is effectively the same as providing the second-stage codebook by the number N of the scaling coefficients. There is an advantage that small coding can be realized. Embodiment 4 FIG. 6 shows a case where the vector codebook 14A of the parameter encoding device of FIG. 1 or the vector codebook 24A of the parameter decoding device of FIG. 2 is configured as a divided vector codebook 4A and the present invention is applied. Show. Although FIG. 6 is composed of a two-part vector codebook, the case where the number of divisions is three or more can be similarly extended.

In this codebook 4A, a low-order vector codebook 41 storing N low-order code vectors x _L1 ,..., X _LN is stored.
Comprising a _L, N 'pieces of high-order code vectors x _H1, ..., x _HN' high-order vector codebook 41 _H that stores. Assuming that the output code vector is x (n), the low-order and high-order vector codebooks 41 _L ,
41 _H, of the order p, low Next 1~k until next, up to k + 1 to p next as higher order, constitute a codebook consisting of the vector for each rank. That is, i of the low-order vector codebook 41 _L
Th vector, x _Li = is represented by _{(x Li1, x Li2, ...} , x Lik) (9), ' the second vector, x _Hi' i of the higher order vector codebook 41 _H = (x _{Hi'k + 1} , x _{Hi'k + 2} , ..., x _Hi'p ) (10) The input index Ix (n) is analyzed by the analysis unit 4
3 is divided into Ix (n) _L and Ix (n) _H, these Ix (n) _L and Ix (n) _H
, The lower-order and higher-order divided vectors x _Li , x _{Hi ′} are selected from the respective codebooks 41 _L , 41 _H , and the integrating unit 47 integrates these divided vectors x _Li , x _{Hi ′} , Generate an output code vector x (n). That is, when a code vector outputted from the integrating unit 47 and x (n), x (n ) = (x Li1, x Li2, ..., x Lik | x Hi'k + 1, x Hi'k + ₂ ,
..., x _Hi'p ).

In this embodiment, the low-order vector codebook 4
1 stores the low-order vector C _0L of the vector C ₀ as a vector of _L, and, as one vector codebook 41 _H of the high-order vector, higher vector of the vector C ₀
To store the C _0H. In this way, a configuration is realized in which C ₀ = ( C _0L | C _0H ) can be output as a code vector corresponding to a silent section or a stationary noise section. Further, depending on the case, it may be output as a combination of C _0L and another higher-order vector or a combination of another lower-order vector and C _0H . Since the provision of the divided vector codebooks 41 _L and 41 _H as shown in FIG. 6 is equivalent to the provision of code vectors by the number of combinations of two divided vectors, it is necessary to reduce the size of each divided vector codebook. There is an advantage that can be. Fifth Embodiment FIG. 7 shows still another configuration example of the vector codebook 14A of the acoustic parameter encoding device of FIG. 1 or the vector codebook 24A of the acoustic parameter decoding device of FIG. -It is a case where it is configured as a divided vector codebook. The codebook 4A is a codebook of the second stage in the codebook 4A of FIG.

The first codebook 41 has N code vectors.
x ₁₁ , ..., x _1N are stored, and the second-stage low-order codebook 42 _L
N 'pieces of divided vectors x _2L1, ..., x _2LN' in is stored, the second stage high-order codebook 42 _H N "pieces of split vector x
_2H1, ..., x _{2HN "is} stored. Inputted index Ix (n) is at code analyzer _431, the index Ix (n) ₁ specifying the code vector at the first stage, 2 It is analyzed into an index Ix (n) ₂ that specifies the code vector of the first stage.
The i-th code vector x _1i corresponding to the index Ix (n) ₁ at the first stage is read from the first-stage codebook 41. Further, Ix is the second stage index Ix (n) ₂ of the analysis unit 43 ₂ (n) _2L and Ix
(n) is analyzed _2H, each i 'th these Ix (n) _2L, Ix (n) 2-stage and low-order split vector codebook 42 _L by _2H, 2-stage high-order split vector codebook 42 _H And the “i” th division vector x _{2Li ′} and x _{2Hi ”,} and the selected division vectors are vector-integrated by the integration unit 47 to generate a second-stage code vector x _2i′i” . In part 44, 1
Code vector x _1i of the second stage and integrated vector x _2i'i of the second stage
Are added and output as a code vector x (n).

In this embodiment, as in the embodiments of FIGS. 4 and 5, the vector C ₀ is stored as one code vector of the first-stage codebook 41, and the second-stage divided vector codebook is stored. The divided zero vectors z _L and z _H are stored as one divided vector of each of the low-order divided vector codebook 42 _L and the high-order divided vector codebook 42 _H of the 42. By doing so, a configuration for outputting a code vector corresponding to a silent section or a stationary noise section is realized. The number of codebook stages may be three or more. Further, the divided vector codebook may be used for an arbitrary stage, and the number of divided vector codebooks per stage is not limited to two. The number of stages to be divided may be one or more. Further, if search is performed for all sets of code vectors between the first-stage codebook 41 and the second-stage codebooks 42 _L and 42 _H , the vector C ₀ and the divided zero vectors z _L and z _H are mutually It may be stored in any of the codebooks in different stages. Alternatively, the divided zero vector need not be stored in the codebook as in the second and third embodiments. Otherwise, when the vector C ₀ is selected, the codebooks 42 _L , 42 _H
Does not perform selective addition from. Example 6 FIG. 8 is to low-order codebook 42 _L and higher codebook 42 _H of the split vector codebook 42 in the vector codebook 4A of the embodiment of FIG. 7, the scaling factor codebook in the embodiment of FIG. 5 Scaling coefficient codebook 45 _L and 4 similar to 45
This is an example in which the present invention is applied to a multi-stage divided vector codebook 4A with scaling coefficients provided with 5 _H. The coefficients for multiplying the lower-order and higher-order split vectors respectively are as follows:
The low-order scaling coefficient codebook 45 _L and the high-order scaling coefficient codebook 45 _H each have N values of, for example, about 0.5.
About 2 coefficients are stored.

The input index Ix (n) is the analyzer 43 _1, the index Ix (n) ₁ specifying the code vector of the first stage, an index specifying the code vector of the second stage Ix (n) It is parsed into ₂ . First, the index Ix (n) ₁
Obtaining a corresponding code vector x _1i in the first stage codebook 41. Further, the low-order scaling coefficient s _Li and the high-order scaling coefficient s _Hi are read from the low-order scaling coefficient codebook 45 _L and the high-order scaling coefficient codebook 45 _H , respectively, corresponding to the index Ix (n) ₁ . Then, the index Ix (n) ₂ is the analysis unit 43 _2, and an index Ix (n) _2L
Is parsed into Ix (n) _2H,, Ix (n) _2L and Ix (n) 2-stage by _2H low-order split vector codebook 42 _L and each of the divided vector x 2-stage high-order split vector 42 _{H 2Li '} , x _{2Hi "}
Select For the selected divided vectors, low-order and high-order scaling coefficients are provided in multipliers 46 _L and 46 _H.
The vector obtained by multiplying s _Li and s _Hi is integrated by the integration unit 47 to generate a second-stage code vector x _{2i′i ”} . The addition unit 44 generates the first-stage code vector x _1i. When adding the second-stage integrated vector _x 2i'i _", the addition result is outputted as the code vector x (n).

The multi-stage with a scaling coefficient of this embodiment
In the divided vector codebook 4A, 1
One of storing the vector C ₀ as the code vector and the second-stage split vector codebook of low-order split vector codebook 42 _L, split zero vectors z _L as a split vector order split vector codebook 42 _H, z _H a store, respectively. By doing so, a configuration for outputting a code vector corresponding to a silent section or a stationary noise section is realized. The number of codebook stages may be three or more. In that case, two or more stages after the second stage may be respectively configured by the divided vector codebook. In any case, the number of divided vector codebooks per stage is not limited. Embodiment 7 FIG. 9 shows still another configuration example of the vector codebook 4A of the acoustic parameter encoding device of FIG. 1 or the vector codebook 24A of the acoustic parameter decoding device of FIG.
In this embodiment, the first-stage codebook 41 is also configured by the same divided vector codebook as the embodiment of FIG. In this embodiment, the first stage low for the next codebook 41 _L N
Number of low-order split vector x _1L1, ..., x _1LN is stored, the first-stage high-order codebook 41 _H N 'pieces of high-order split vector
x _1H1, ..., x _{HN 'is} stored, the second stage in the low-order codebook 42 _L is
N "low-order divided vectors x _2L1 , ..., x _2LN" are stored and 2
The stage higher-order codebook 42 _H has N ′ ″ higher-order divided vectors.
x _2H1 , ..., x _{2HN '"} are stored.

The input index Ix (n) is converted by the code analysis unit 43 into an index for designating the first-stage vector.
Ix (n) ₁ and index Ix specifying the second stage vector
(n) is analyzed as ₂ . The vector corresponding to the first-stage index Ix (n) ₁ is stored in the first-stage low-order divided vector codebook 41.
_L, and 1 each i-th stage higher order split vector codebook 41 _H and i 'th split vector x _1Li, x _1Hi' is selected, integrated by the integration of the first stage them in integration unit 47 ₁ to generate a vector x _{1ii '.} Also, the second-stage index Ix
(n) ₂ is also similar to the first stage, a second stage low-order split vector codebook 42 _L, each 2-stage high-order split vector codebook 42 _H i "th and i '" th split vectors x _{2Li ",} x _{2HI '"} is selected, these are integrated in integrating section 47 ₂ 2-stage integrated vector x _{2i "i'} for generating _a". At adder 44, the first-stage integrated vector x _{1ii 'and} integrated vector x _{2i "i} of the second _stage' by _adding", and outputs the addition result as the code vector x (n).

In this embodiment, in the first stage, FIG.
The Like the split vector codebook arrangement, and stores the low-order split vector C _0L of the vector C ₀ as one code vector of the codebook 41 _L of low-order vector of the first stage and the first stage high The higher-order divided vector C _0H of the vector C ₀ is stored as one divided vector of the code book 41 _H of the next-order vector, and the lower-order divided vector code book 42 _L of the second-stage divided vector code book 42, The divided zero vectors z _L and z _H are stored as one vector in each of the second-order higher-order divided vector codebooks 42 _H. This configuration realizes a configuration that can output a code vector corresponding to a silent section or a stationary noise section. Again, the number of stages is 2
The number of divided vector codebooks per stage is not limited to two. Embodiment 8 FIG. 10 is a block diagram showing a configuration of an audio signal transmitting apparatus and a receiving apparatus to which the present invention is applied.

The audio signal 101 is converted into an electric signal by the input device 102 and output to the A / D converter 103. The A / D conversion device 103 converts the (analog) signal output from the input device 102 into a digital signal and outputs the digital signal to the speech encoding device 104. The audio encoding device 104 encodes the digital audio signal output from the A / D conversion device 103 using an audio encoding method described later, and outputs encoded information to the RF modulation device 105. The RF modulation device 105 converts the audio encoded information output from the audio encoding device 104 into a signal to be transmitted on a propagation medium such as a radio wave and outputs the signal to the transmission antenna 106. The transmission antenna 106 transmits an output signal output from the RF modulation device 105 as a radio wave (RF signal) 107. The above is the configuration and operation of the audio signal transmitting device.

The transmitted radio wave (RF signal) 108 is received by the receiving antenna 109 and output to the RF demodulator 110. Note that a radio wave (RF signal) 108 in the drawing is a radio wave (RF signal) 107 as viewed from the receiving side, and unless there is signal attenuation or noise superimposition on the propagation path, the radio wave (RF signal)
It is exactly the same as 107. The RF demodulation device 110 demodulates audio encoded information from the RF signal output from the receiving antenna 109 and outputs the demodulated information to the audio decoding device 111. The audio decoding device 111 decodes an audio signal from the audio coding information output from the RF demodulation device 110 using an audio decoding method described later, and outputs the audio signal to the D / A conversion device 112. The D / A conversion device 112 converts the digital audio signal output from the audio decoding device 111 into an analog electric signal and outputs it to the output device 113.
The output device 113 converts the electric signal into the vibration of air and outputs the sound wave 114
Output as audible to human ears. The above is the configuration and operation of the audio signal receiving device.

By providing at least one of the above-described audio signal transmitting device and receiving device, a base station device and a mobile terminal device in a mobile communication system can be configured. The speech signal transmitting apparatus has the features of the speech encoding apparatus 104. FIG. 11 shows a speech encoding device 1.
FIG. 4 is a block diagram showing the configuration of 04. Figure 1 shows the input audio signal
0 is a signal output from the A / D converter 103 and input to the preprocessing unit 200. The pre-processing unit 200 performs a waveform shaping process or a pre-emphasis process that leads to an improvement in the performance of a high-pass filter process for removing a DC component and a subsequent encoding process. And to the parameter determination unit 212. LP
C analysis section 201 performs a linear prediction analysis on Xin, and outputs the analysis result (linear prediction coefficient) to LPC quantization section 202. LP
The C quantization unit 202 includes an LSP parameter calculation unit 13, a parameter encoding unit 10, a decoding unit 18, and a parameter conversion unit 19. Parameter encoding unit 10
Has the same configuration as the parameter encoding unit 10 in FIG. 1 to which the vector codebook of the present invention according to any of the embodiments in FIGS. The decoding unit 18 has the same configuration as the decoding device in FIG. 2 to which any one of the codebooks in FIGS.

The linear prediction coefficient (LPC) output from the LPC analysis unit 201 is converted into an LSP parameter by the LSP parameter calculation unit 13, and the obtained LSP parameter is explained by the parameter encoding unit 10 with reference to FIG. Is encoded as described above. Codes Ix (n) and Iw (n) obtained by encoding, that is, quantized LP
The code L representing C is output to the multiplexing unit 213, and
The codes Ix (n) and Iw (n) are decoded by the decoding unit 18 to obtain quantized LSP parameters, which are again converted to LPC parameters by the parameter conversion unit 19, and the obtained quantization The LPC parameters are given to the synthesis filter 203. The synthesis filter 203 uses the quantized LPC as a filter coefficient, synthesizes an acoustic signal by a filter process with the driving sound source signal output from the adder 210, and outputs the synthesized signal to the adder 204.

The adder 204 calculates an error signal ε between the Xin and the synthesized signal, and outputs it to the auditory weighting unit 211.
The auditory weighting unit 211 performs auditory weighting on the error signal ε output from the adder 204, calculates the distortion of the synthesized signal with respect to the Xin in the auditory weighting area, and outputs the calculated distortion to the parameter determining unit 212. . The parameter determining unit 212 performs adaptive codebook 205 and fixed codebook 2 so that the coding distortion output from the auditory weighting unit 211 is minimized.
07 and a signal to be generated from the quantization gain generation unit 206 are determined. It should be noted that not only the coding distortion output from the auditory weighting unit 211 is minimized, but also a signal to be generated from the three units is determined by using another coding distortion minimizing method using Xin. This can further improve the encoding performance.

The adaptive codebook 205 buffers the excitation signal of the immediately preceding frame n-1 output by the adder 210 in the past when the distortion is minimized, and is output from the parameter determination unit 212. The excitation vector is cut out from the position specified by the adaptive vector code A, and is repeatedly connected until it becomes one frame length to generate an adaptive vector including a desired periodic component. The fixed codebook 207 stores a plurality of one-frame-length fixed vectors corresponding to the fixed vector codes, and outputs the fixed vector code F
Multiplier with a fixed vector having the shape specified by
Output to 209.

The quantization gain generation section 206 multiplies the quantization adaptive vector gain g _A and the quantization fixed vector gain g _F for the adaptive vector and the fixed vector specified by the gain code G output from the parameter determination section 212, respectively. Container 20
Give to 8 and 209. Multiplier 208, the quantized output from the quantizer gain generating unit 206 adaptive vector gain g _A, adaptive codebook
Multiplied by the adaptive vector output from 205 and output to adder 210. Multiplier 209, the quantized fixed vector gain g _F outputted from the quantization gain generating section 206, multiplied by the fixed vector outputted from the fixed vector codebook 207, adder 210
Output to

The adder 210 performs vector addition of the adaptive vector after the gain multiplication and the fixed vector, and
And output to adaptive codebook 205. Finally, the multiplexing unit 213
Is a code L representing the quantized LPC from the LPC quantization unit 202, an adaptive vector code A representing the adaptive vector and a fixed vector code F representing the fixed vector from the parameter determination unit 212.
And a gain code G representing a quantization gain, respectively, and multiplexes these codes to output as coded information to a transmission path. FIG. 12 is a block diagram of the speech decoding apparatus 111 shown in FIG.
FIG. 3 is a block diagram showing the configuration of FIG.

In the figure, the coded information output from the RF demodulation unit 110 is obtained by converting the coded information multiplexed by the demultiplexing unit 1301 into individual codes L. A, F, and G are separated. The separated LPC code L is provided to an LPC decoding unit 1302, the separated adaptive vector code A is provided to an adaptive codebook 1305, and the separated gain code G is
06, and the separated fixed vector code F is provided to the fixed codebook 1307. The LPC decoding section 1302 includes a decoding section 1302A configured in the same manner as in FIG. 2 and a parameter conversion section 1302B. The code L = (Ix (n), Iw (n)) given from the demultiplexing / demultiplexing unit 1301 is decoded by the decoding unit 1302A in the LSP parameter area as shown in FIG. Is output to the synthesis filter 1303.

The adaptive codebook 1305 extracts an adaptive vector from the position specified by the adaptive vector code A output from the demultiplexing / demultiplexing unit 1301, and outputs the adaptive vector to the multiplier 1308. The fixed codebook 1307 generates a fixed vector specified by the fixed vector code F output from the demultiplexing unit 1301, and
Output to 1309. Quantization gain generating section 1306 outputs to the multiplexing designated by the outputted gain code G from the separation unit 1301 adaptive vector gain g _A and decodes the fixed vector gain g _F multipliers 1308 and 1309. The multiplier 1308 multiplies the adaptive code vector by the adaptive code vector gain g _A and outputs the result to the adder 1310. Multiplier 1309, the multiplied fixed code vector gain g _F in the fixed code vector,
Output to adder 1310. Adder 1310 adds the adaptive vector after the gain multiplication output from adders 1308 and 1309 and the fixed vector, and outputs the result to synthesis filter 1303.
The synthesis filter 1303 performs filter synthesis using the vector output from the adder 1310 as a drive excitation signal, using the filter coefficient decoded by the LPC decoding unit 1302,
The combined signal is output to post-processing section 1304. Post-processing unit 1304
Performs a process for improving the subjective quality of speech, such as formant emphasis and pitch emphasis, and a process for improving the subjective quality of stationary noise, and outputs the final decoded speech signal.

In the above description, the LSP parameter is used as a parameter equivalent to the linear prediction coefficient representing the spectrum envelope of the speech signal.
A Percoll coefficient or the like may be used. Also when these are used, the spectrum envelope becomes flat in a silent section or a stationary noise section, so that the calculation of the parameters in this section can be easily performed. 1.0 and 1 to p order may be set to 0.0. Even when other acoustic parameters are used, any vector of acoustic parameters determined to represent a substantially flat spectrum envelope may be used. The LSP parameter is practical because of its good quantization efficiency.

In the above description, when the vector codebook has a multi-stage configuration, the vector C ₀ is represented by, for example, C ₀ = C ₀₁ + C ₀₂ and two combined vectors, and C ₀₁ and C ₀₂ are codebooks of different stages. May be stored. Further, the present invention can be applied not only to encoding and decoding of audio signals but also to encoding and decoding of general acoustic signals such as music signals. Further, the device of the present invention can also execute encoding and decoding of an audio signal by causing a computer to execute a program. FIG. 13 shows the acoustic parameter encoding device and the decoding device of FIGS. 1 and 2 using the codebook according to the present invention of any of FIGS. 3 to 9, and FIGS. 11 and 11 to which the encoding method and the decoding method are applied. 12 shows an embodiment in which twelve acoustic signal encoding devices and decoding devices are executed by a computer.

A computer embodying the present invention includes a modem 410 connected to a communication network, an input / output interface 420 for inputting / outputting audio signals, a buffer memory 430 for temporarily storing digital audio signals or audio signal codes, And a random access memory (RAM) 440 for executing decoding processing there, a central processing unit (CPU) 450 for controlling input / output of data and execution of programs, and a hard disk 46 storing encoding and decoding programs.
0, comprises a driving device 470 for driving the recording medium 470M,
These are connected to each other by a common bus 480.

As the recording medium 470M, a compact disk CD, a digital video disk DVD, a magneto-optical disk MO, a memory card, or any other type of recording medium may be used. The heart disk 460 stores a program representing the encoding method and the decoding method implemented in the audio signal encoding apparatus and the decoding apparatus in FIGS.
The program includes, as a subroutine, a program for executing the acoustic parameter encoding and decoding of FIGS.

When encoding the input audio signal, the CPU
The 450 reads the audio signal encoding program from the hard disk 460 into the RAM 440, and encodes the audio signal fetched into the buffer memory 430 via the input / output interface 420 by performing processing according to the encoding program in the RAM 440 for each frame. The obtained code is transmitted to the communication network via the modem 410, for example, as encoded audio signal data. Alternatively, it is temporarily stored in the hard disk 460. Alternatively, the recording medium 470 is
Write to M.

When decoding the input encoded audio signal data, the CPU 450 reads the decoding program from the hard disk 460 into the RAM 440. The acoustic code data is downloaded from the communication network to the buffer memory 430 via the modem 410, or is read from the recording medium 470M into the buffer memory 430 by the driving device 470. , And the obtained audio signal data is output from the input / output interface 420.

[0066]

The effect of the present invention is shown in Table 1 of FIG.
In the case of embedded vector C ₀ and the zero vector z silence section codebook this invention, showing the quantization performance of the acoustic parameter coding device when conventional not to fill vectors C ₀ to codebook as. In Table 1, the vertical axis represents cepstrum distortion, which corresponds to logarithmic spectrum distortion, and is expressed in decibels (dB). The smaller the cepstrum distortion, the better the quantization performance. The average distortion is calculated for the voice section for calculating the distortion in the average of all sections (Total), in sections other than the silent section and the voice stationary section (Mode 0), and in the voice steady section (Mode 1). Was. There is a silent section in Mode 0, and the distortion there is 0.11 dB lower in the proposed codebook, indicating that there is an effect of inserting a silent section and a zero vector.
In addition, the cepstrum distortion in Total is lower when the proposed codebook is used, and there is no deterioration even in a stationary speech section. Therefore, the effectiveness of the codebook according to the present invention is clear.

As described above, according to the present invention, the weighted sum of the code vector of the current frame and the code vector output in the past or the vector obtained by adding the average vector obtained in advance to the code vector is equivalent to the linear prediction coefficient. In the encoding of quantizing parameters, a vector stored in a vector codebook, a parameter vector corresponding to a silent section or a stationary noise section, or a vector obtained by subtracting the average vector from the parameter vector as a code vector. Since it is possible to select and output the code, it is possible to provide an encoding / decoding method and an apparatus therefor with less quality deterioration in these sections.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional configuration of an acoustic parameter encoding device to which a codebook according to the present invention is applied.

FIG. 2 is a block diagram showing a functional configuration of an acoustic parameter decoding device to which a codebook according to the present invention is applied.

FIG. 3 is a diagram showing a configuration example of a vector codebook according to the present invention for encoding and decoding LSP parameters.

FIG. 4 is a diagram showing a configuration example of a vector codebook according to the present invention in a multi-stage configuration.

FIG. 5 is a diagram showing a configuration example of a vector codebook according to the present invention when configured with a divided vector codebook.

FIG. 6 is a diagram showing a configuration example of a vector codebook according to the present invention when a scaling coefficient is applied to a multistage vector codebook.

FIG. 7 is a diagram showing a configuration example of a vector codebook according to the present invention when a second-stage codebook is configured by a divided vector codebook.

FIG. 8 is a diagram showing a configuration example of a vector codebook in a case where a scaling coefficient is applied to each of two divided vector codebooks in the codebook of FIG. 7;

9 is a diagram showing a configuration example of a vector codebook in a case where each stage of the multistage vector codebook in FIG. 4 is a divided vector codebook.

FIG. 10A is a block diagram illustrating a configuration example of an audio signal transmission device to which the encoding method according to the present invention is applied, and FIG. 10B is a block diagram illustrating a configuration example of an audio signal receiving device to which the decoding method according to the present invention is applied; FIG.

FIG. 11 is a diagram showing a functional configuration of an audio signal encoding device to which the encoding method according to the present invention is applied.

FIG. 12 is a diagram showing a functional configuration of an audio signal decoding device to which the decoding method according to the present invention is applied.

FIG. 13 is a diagram showing an example of a configuration in a case where an encoding device and a decoding device according to the present invention are implemented by a computer.

FIG. 14 is a graph for explaining the effect of the present invention.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Hiyori ▲ Saki ▼ Yusuke 2-3-1 Otemachi, Chiyoda-ku, Tokyo Nippon Telegraph and Telephone Corporation (72) Inventor Hiroyuki Ehara Kohoku-ku, Yokohama-shi, Kanagawa Matsushita Communication Industrial Co., Ltd. (3-1) Tsunashima Higashi 4-chome (72) Inventor Kazutoshi Yasunaga 1006 Kazuma Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. F term (reference) 5D045 CA01 CC07 DA11 5J064 AA01 BA13 BB03 BC01 BC08 BC09 BC21 BD02

Claims (41)

[Claims] 1. An audio parameter encoding method, comprising: (a) calculating an audio parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic of an audio signal for each frame having a fixed time length; One or more sets of code vectors output in at least one frame in the closest past from a codebook stored in correspondence with indices representing the code vectors and the code vector selected in the current frame. Weighted coefficients of the set selected from the coefficient codebook stored in correspondence with the indices representing the weighted coefficients are multiplied and added to generate a weighted vector, and a vector including the components of the weighted vector is generated. Obtaining a quantized acoustic parameter candidate for the acoustic parameter of the current frame (C) determining a set of code vectors of the vector codebook and weighting coefficients of the coefficient codebook using a criterion such that distortion of the candidate of the quantized acoustic parameter with respect to the calculated acoustic parameter is minimized. And determining and outputting an index representing the set of the determined code vector and weight coefficient as the quantization code of the acoustic parameter. The vector codebook stores one of the stored code vectors.
First, it includes a vector including components of an acoustic parameter vector representing a substantially flat spectrum envelope. 2. The encoding method according to claim 1, wherein said vector codebook comprises a plurality of stages of codebooks each of which stores a plurality of vectors corresponding to an index representing them. In the codebook of one stage of the codebook, a vector including the component of the acoustic parameter vector representing the above-mentioned substantially flat spectrum envelope is stored as one vector, and in the codebook of the other stage, one zero vector is stored as one vector. Stored as a vector, as described in step (b) above.
Includes selecting vectors from the plurality of codebooks, adding the vectors, and outputting the resultant as the selected code vector of the current frame. 3. The encoding method according to claim 1, wherein said vector codebook comprises a plurality of stages of codebooks each of which stores a plurality of vectors corresponding to an index representing them. In the codebook of one stage of the codebook, a vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is stored as one vector, and the step (b) includes When a code vector other than the vector including the component of the parameter vector is selected from the codebook, vectors are respectively selected from the codebooks of the plurality of stages, and the vectors are added to each other as the selected code vector of the current frame. A vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope from the one-stage codebook. And selecting a vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope as the selected code vector of the current frame. 4. The encoding method according to claim 2, wherein at least one of the plurality of stages of the codebook is distributed as a plurality of divided vectors in which the dimension of the code vector is divided into a plurality. It includes a plurality of stored divided vector codebooks and an integrating unit that integrates the divided vectors output from the plurality of divided vector codebooks and outputs the integrated vector as an output vector of the codebook at that stage. 5. The encoding method according to claim 2, wherein the vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is obtained in advance from a parameter vector equivalent to the linear prediction coefficient. This is a vector generated by subtracting an average vector of parameters equivalent to the linear prediction coefficients of the entire acoustic signal. 6. The encoding method according to claim 1, wherein said vector codebook is provided for a plurality of stages of codebooks each storing a plurality of code vectors, and for each of the second and subsequent stages. A scaling coefficient codebook, wherein each of the scaling coefficient codebooks stores a predetermined scaling coefficient corresponding to each code vector of the first-stage codebook. In the codebook of one stage, a vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is stored, and in the codebooks of the remaining stages, one zero vector is stored for each. The step (b) reads the corresponding scaling coefficient from the scaling codebook for the codebooks of the second and subsequent stages for the code vector selected in the first stage, Multiplying the code vectors selected from the first and second codebooks, and outputting the multiplication result as a vector of each stage; and adding the output vector of each stage and the vector of the first stage, Outputting the addition result as a code vector from the vector codebook. 7. The encoding method according to claim 2, wherein steps (b) and (c) are jointly performed,
First, a step of searching for a predetermined number of code vectors having the smallest distortion by the code vector selected from the one-stage codebook, and then, from each of the predetermined number of codevectors and the codebooks of the remaining stages, The method includes a step of obtaining the distortion for each of all combinations with the code vector selected one by one, and determining a set of code vectors that minimizes the distortion. 8. The encoding method according to claim 6, wherein at least one of the second and subsequent codebooks of the plurality of codebooks has a plurality of code vector dimensions divided into a plurality. A plurality of divided vector codebooks distributed and stored as divided vectors, wherein the scaling coefficient codebook corresponding to the at least one stage codebook includes a plurality of divided vectors provided for the plurality of divided vector codebooks. Each of the divided vector scaling coefficient codebooks includes a vector scaling coefficient codebook, and each of the divided vector scaling coefficient codebooks has a predetermined divided vector scaling coefficient corresponding to each code vector of the first stage codebook. Wherein said step (b) comprises selecting each of said plurality of divided vector codebooks of said at least one stage. Of the divided vector
Reading the divided vector scaling coefficient corresponding to the index of the vector selected in the codebook of the second stage from each of the divided vector scaling coefficient codebooks and multiplying the divided vector scaling coefficients by: Integrating and outputting as an output vector of the codebook of the stage. 9. The encoding method according to claim 1, wherein said vector codebook comprises a plurality of divided vector codebooks obtained by dividing a code vector into a plurality of dimensions, and a divided vector codebook output from said divided vector codebook. And an integrated unit that outputs a single code vector, and the vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is divided into divided vectors, and 1 And divided and stored as one divided vector. 10. The encoding method according to claim 1, wherein the vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is obtained by previously calculating the average vector from the acoustic parameter vector representing the substantially flat spectrum envelope. The step (b) is performed by adding an average vector of a parameter equivalent to the linear prediction coefficient of the entire sound signal, which is obtained in advance, to the weighted vector, and calculating the component of the weighted vector. Generating a vector containing 11. The encoding method according to claim 1, wherein a parameter equivalent to the linear prediction coefficient is an LSP parameter. 12. A method for decoding acoustic parameters, comprising: (a) a vector storing a plurality of code vectors of acoustic parameters equivalent to linear prediction coefficients representing a spectral envelope characteristic of an acoustic signal in correspondence with indices representing them; From a codebook and a coefficient codebook in which one or more sets of weighting factors are stored in correspondence with indexes representing those sets, a code vector corresponding to an index represented by a code input for each frame and one set of weights Outputting a coefficient; and (b) outputting the code vector output from the vector codebook in at least one frame in the closest past and the code vector output from the vector codebook in the current frame, respectively. The weighted coefficients of the set are multiplied and added to generate a weighted vector. Outputting a vector containing the components of the weighted vector as a decoded quantized vector of the current frame, wherein the vector codebook contains one of the stored code vectors.
First, it includes a vector including components of an acoustic parameter vector representing a substantially flat spectrum envelope. 13. The decoding method according to claim 12, wherein
The vector codebook includes a plurality of stages of codebooks each of which stores a plurality of vectors corresponding to indices representing the vectors. The vector including the components of the acoustic parameter vector representing the spectral envelope is stored as one vector, and the codebooks at the other stages store the zero vector as one vector.
(b) outputting a vector specified by the index represented by the input code from each of the plurality of stages of codebooks, adding the vectors, and outputting the addition result as a code vector in the current frame. including. 14. The decoding method according to claim 12, wherein
The vector codebook includes a plurality of stages of codebooks each of which stores a plurality of vectors corresponding to indices representing the vectors, and the substantially flat codebook includes one of the plurality of stages of codebooks. The vector including the components of the acoustic parameter vector representing the spectral envelope is stored as one vector, and the step (b) comprises the steps of: When a code vector other than a vector including a component is output, vectors are respectively selected from the codebooks of the plurality of stages, the vectors are added, and the resultant is output as the selected code vector of the current frame. Vector containing acoustic parameter vector components representing the above-mentioned almost flat spectrum envelope from the codebook If so, the method includes a step of outputting a vector including a component of the acoustic parameter vector representing the substantially flat spectrum envelope as a code vector of the current frame. 15. The decoding method according to claim 13, wherein at least one of the codebooks in the plurality of codebooks is distributed as a plurality of divided vectors in which the dimension of the code vector is divided into a plurality. It includes a plurality of stored divided vector codebooks, and an integration unit that integrates the divided vectors output from the plurality of divided vector codebooks and outputs the resulting vector as an output vector of the codebook at that stage. 16. A decoding method according to claim 13, wherein the vector including a component of the parameter vector equivalent to the linear prediction coefficient is a sound vector previously determined from the parameter vector equivalent to the linear prediction coefficient. This is a vector generated by subtracting an average vector of parameters equivalent to the linear prediction coefficient of the entire signal. 17. The decoding method according to claim 12, wherein:
The vector codebook includes a plurality of codebooks each storing a plurality of code vectors, and a scaling coefficient codebook provided for each of the second and subsequent codebooks. Stores a predetermined scaling coefficient corresponding to each code vector of the first-stage codebook. The substantially flat spectrum envelope is stored in the one-stage codebook of the plurality of codebooks. The vector including the component of the acoustic parameter vector to be represented is stored, and the codebooks of the remaining stages each store a zero vector, one at a time. For the code vector, read the corresponding scaling coefficient from the scaled codebook for the second and subsequent codebooks, and select the code vector respectively selected from the second and subsequent codebooks And multiplying the result of the multiplication as a vector of each stage, adding the output vector of each stage and the vector of the first stage by vector, and adding the addition result to a code vector from the vector codebook. And outputting as 18. The decoding method according to claim 17, wherein
At least one of the codebooks in the second and subsequent stages of the codebooks in the plurality of stages
The codebook of one stage is composed of a plurality of divided vector codebooks distributed and stored as a plurality of divided vectors obtained by dividing a code vector dimension into a plurality of dimensions, and the scaling coefficient code corresponding to the at least one stage codebook is provided. The book includes a plurality of division vector scaling coefficient codebooks provided for the plurality of division vector codebooks, and each of the division vector scaling coefficient codebooks has a code vector of the first stage codebook. And a plurality of scaling coefficients for divided vectors are stored corresponding to the divided vector codebooks selected from the plurality of divided vector codebooks of the at least one stage.
Reading the divided vector scaling coefficient corresponding to the index of the vector selected in the codebook of the second stage from each of the divided vector scaling coefficient codebooks and multiplying the divided vector scaling coefficients by: Integrating and outputting as an output vector of the codebook of the stage. 19. A decoding method according to claim 12, wherein:
The vector codebook includes a plurality of divided vector codebooks in which the dimensions of a code vector are divided into a plurality of parts, and an integration unit that integrates the divided vectors output from the divided vector codebook and outputs the resultant as one code vector. The vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is divided into divided vectors, and each divided vector codebook distributes and stores one as a divided vector. 20. The decoding method according to claim 12, wherein
The vector including the component of the acoustic parameter vector representing the substantially flat spectrum envelope is a vector generated by subtracting the average vector from the acoustic parameter vector representing the substantially flat spectrum envelope in advance,
The step (b) includes a step of adding a vector including a component of the weighted vector by adding, to the weighted vector, an average vector of a parameter equivalent to the linear prediction coefficient of the entire acoustic signal, which is obtained in advance. Including. 21. The decoding method according to claim 12, wherein a parameter equivalent to the linear prediction coefficient is an LSP parameter. 22. An audio parameter encoding apparatus, comprising: a parameter calculation means for analyzing an input audio signal for each frame and calculating an audio parameter equivalent to a linear prediction coefficient representing a spectrum envelope characteristic of the audio signal; And a coefficient codebook in which one or more sets of weighting factors are stored in correspondence with indices representing those sets, and a vector codebook in which one or more sets of weighting factors are stored in correspondence with the indices representing those sets. The code vector for the current frame and the code vector output in at least one frame in the closest past are multiplied by each of the weighting coefficients of the set selected from the coefficient codebook and added to form a weighted vector. Generate and include a vector including the generated weighted vector component in the current frame. A quantization parameter generation unit that outputs as a candidate of a quantized acoustic parameter for the acoustic parameter, a distortion calculating unit that calculates a distortion of the quantized acoustic parameter with respect to the acoustic parameter calculated by the parameter calculating unit, Using a criterion such that is small, a code vector of the vector codebook and a weight coefficient of the set of coefficient codebooks are determined, and an index representing each of the determined code vector and the set of weight coefficients is defined as A codebook search control unit that outputs the code of the acoustic parameter,
The vector codebook includes a vector including a component of an acoustic parameter vector representing a substantially flat spectrum envelope.
Included as one sign vector. 23. The encoding apparatus according to claim 22, wherein:
The vector codebook outputs a code vector by adding a plurality of stages of codebooks in which a plurality of vectors are stored in correspondence with indices representing them, and vectors output from the plurality of stages of codebooks. And a vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope is stored as one vector in the codebook of one of the plurality of codebooks. , The zero vector is stored as one code vector. 24. The encoding apparatus according to claim 23, wherein:
The codebook of at least one of the plurality of codebooks is a plurality of divided vectors distributed and stored as a plurality of divided vectors obtained by dividing the dimension of the code vector into a plurality of divided vectors. A codebook, and an integrating unit that integrates the divided vectors output from the plurality of divided vector codebooks and outputs the resultant as the output vector of the codebook at that stage. 25. An encoding apparatus according to claim 22, wherein:
The vector codebook includes a plurality of codebooks in each of which a plurality of code vectors are stored corresponding to indices representing the codebooks, and a codebook in the first stage provided for each of the second and subsequent codebooks. And a scaling coefficient codebook in which predetermined scaling coefficients corresponding to the respective code vectors are stored in correspondence with the indices representing them. A multiplication unit that reads a corresponding scaling coefficient from the codebook scaling for the codebook, multiplies the code vectors respectively selected from the second and subsequent codebooks, and outputs a multiplication result as a vector of each stage; , And the vector of the first stage are vector-added, and the addition result is obtained from the vector codebook. An adder that outputs a code vector, and a codebook of one of the plurality of codebooks stores a vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope, Zero vectors are stored in the codebooks of the remaining stages. 26. An encoding apparatus according to claim 25, wherein:
At least one of the codebooks in the second and subsequent stages of the codebooks in the plurality of stages
The codebook of one stage is composed of a plurality of divided vector codebooks distributed and stored as a plurality of divided vectors obtained by dividing a code vector dimension into a plurality of dimensions, and the scaling coefficient code corresponding to the at least one stage codebook is provided. The book includes: a plurality of division vector scaling coefficients codebooks in which a plurality of division vector scaling coefficients are stored corresponding to the first stage code vectors, respectively; For the divided vectors output from the plurality of divided vector codebooks in one stage,
A multiplying means for reading out the divided vector scaling coefficient corresponding to the index of the vector selected in the codebook of the stage from each of the divided vector scaling coefficient codebooks and multiplying them, And an integration unit that outputs the codebook as an output vector of the codebook. 27. The encoding apparatus according to claim 22, wherein:
The vector codebook integrates a plurality of divided vector codebooks distributed and stored as a plurality of divided vectors in which the dimension of the code vector is divided into a plurality of divided vectors, and integrates the divided vectors output from these divided vector codebooks into one. A vector including an acoustic parameter vector component representing a substantially flat spectrum envelope is divided into divided vectors and distributed and stored as one divided vector in each of the plurality of divided vector codebooks. Have been. 28. An audio parameter decoding apparatus, comprising: a vector codebook storing a plurality of code vectors of an acoustic parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic of an acoustic signal in correspondence with an index representing them. A coefficient codebook storing one or more sets of weighting coefficients corresponding to indices representing them, and outputting one code vector from the vector codebook according to an index represented by a code input for each frame; Outputting a set weighting coefficient from the coefficient codebook, and weighting the set output in the current frame to the code vector output in the current frame and the code vector output in at least one frame in the closest past. Generates a weighted vector by multiplying each coefficient and adding them, and generates the weight And a quantization parameter generating means for outputting a vector including a component of the current vector as a decoded quantized acoustic parameter of the current frame, wherein the vector codebook includes a component of an acoustic parameter vector representing a substantially flat spectrum envelope. Is stored as one of the code vectors. 29. The decoding apparatus according to claim 28, wherein
The vector codebook outputs a code vector by adding a plurality of stages of codebooks in which a plurality of vectors are stored in correspondence with indices representing them, and vectors output from the plurality of stages of codebooks. And a vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope is stored as one vector in the codebook of one of the plurality of codebooks. , The zero vector is stored as one code vector. 30. A decoding device according to claim 29, wherein:
The codebook of at least one of the plurality of codebooks is a plurality of divided vectors distributed and stored as a plurality of divided vectors obtained by dividing the dimension of the code vector into a plurality of divided vectors. A codebook, and an integrating unit that integrates the divided vectors output from the plurality of divided vector codebooks and outputs the resultant as the output vector of the codebook at that stage. 31. The decoding apparatus according to claim 28, wherein:
The vector codebook includes a plurality of codebooks in each of which a plurality of code vectors are stored corresponding to indices representing the codebooks, and a codebook in the first stage provided for each of the second and subsequent codebooks. And a scaling coefficient codebook in which predetermined scaling coefficients corresponding to the respective code vectors are stored in correspondence with the indices representing them. A multiplication unit that reads a corresponding scaling coefficient from the codebook scaling for the codebook, multiplies the code vectors respectively selected from the second and subsequent codebooks, and outputs a multiplication result as a vector of each stage; , And the vector of the first stage are vector-added, and the addition result is obtained from the vector codebook. An adder that outputs a code vector, and a codebook of one of the plurality of codebooks stores a vector including a component of an acoustic parameter vector representing the substantially flat spectrum envelope, Zero vectors are stored in the codebooks of the remaining stages. 32. The decoding device according to claim 31, wherein
At least one of the codebooks in the second and subsequent stages of the codebooks in the plurality of stages
The codebook of one stage is composed of a plurality of divided vector codebooks distributed and stored as a plurality of divided vectors obtained by dividing a code vector dimension into a plurality of dimensions, and the scaling coefficient code corresponding to the at least one stage codebook is provided. The book includes a plurality of division vector scaling coefficient codebooks in which a plurality of division vector scaling coefficients are stored corresponding to the first stage code vectors, respectively, in correspondence with the plurality of division vector codebooks, For the divided vectors output from the plurality of divided vector codebooks in one stage,
A multiplying means for reading out the divided vector scaling coefficient corresponding to the index of the vector selected in the codebook of the stage from each of the divided vector scaling coefficient codebooks and multiplying the readout result, And an integration unit that outputs the codebook as an output vector of the codebook. 33. The decoding apparatus according to claim 28, wherein
The vector codebook integrates a plurality of divided vector codebooks distributed and stored as a plurality of divided vectors in which the dimension of the code vector is divided into a plurality of divided vectors, and integrates the divided vectors output from these divided vector codebooks into one. A vector including an acoustic parameter vector component representing a substantially flat spectrum envelope is divided into divided vectors and distributed and stored as one divided vector in each of the plurality of divided vector codebooks. Have been. 34. An audio signal encoding apparatus for encoding an input audio signal, means for encoding a spectral characteristic of the input audio signal using the audio parameter encoding method according to claim 1, and said input audio signal. An adaptive codebook holding an adaptive codevector representing a periodic component of a signal, a fixed codebook storing a plurality of fixed vectors, and an adaptive codevector from the adaptive codebook and a fixed vector from the fixed codebook. Filter means for inputting the generated sound source vector as an excitation signal and synthesizing a synthetic audio signal using a filter coefficient based on the quantized audio parameter; and a distortion of the synthetic audio signal with respect to the input audio signal is reduced. The adaptive code vector and the fixed vector to be selected from the fixed codebook and the adaptive codebook are determined, and the determined adaptive code is determined. Vector and means for outputting an adaptive code and a fixed code respectively to the fixed vector corresponding, including capital. 35. An audio signal decoding device for decoding an input code and outputting an audio signal, wherein the audio parameter decoding method according to claim 12 is
A means for decoding an acoustic parameter equivalent to a linear prediction coefficient representing a spectral envelope characteristic from an input code, a fixed codebook storing a plurality of fixed vectors, and an adaptive code vector representing a periodic component of a synthesized acoustic signal are stored. Based on the adaptive codebook and the input adaptive code and fixed code, a corresponding fixed vector is extracted from the fixed codebook, a corresponding adaptive code vector is extracted from the adaptive codebook, and a vector is synthesized to generate an excitation vector. And a filter means for setting a filter coefficient based on the acoustic parameter and reproducing an acoustic signal using the excitation vector. 36. An audio signal encoding method for encoding an input audio signal, comprising: (A) encoding the spectral characteristics of the input audio signal using the audio parameter encoding method according to claim 1; (B) generated based on an adaptive code vector from an adaptive code book holding an adaptive code vector representing a periodic component of an input audio signal and a fixed vector from a fixed code book storing a plurality of fixed vectors. Using a sound source vector as an excitation signal, performing a synthetic filter process using a filter coefficient based on the quantized acoustic parameter to generate a synthetic acoustic signal, and (C) distortion of the synthetic acoustic signal with respect to the input acoustic signal. Determine the adaptive code vector and fixed vector to be selected from the fixed codebook and the adaptive codebook so as to be smaller,
Outputting an adaptive code and a fixed code corresponding to the determined adaptive code vector and fixed vector, respectively. 37. A sound signal decoding method for decoding an input code and outputting an audio signal, wherein (A) using the sound parameter decoding method according to claim 12, a spectrum envelope characteristic is obtained from the input code. Decoding an acoustic parameter equivalent to the linear prediction coefficient to be represented;
The corresponding adaptive code vector is extracted from the adaptive code book holding the adaptive code vector indicating the periodic component of the input audio signal, and the corresponding fixed vector is extracted from the fixed code book in which a plurality of fixed vectors are stored. Generating an excitation vector by synthesizing the excitation vector with a fixed vector, and (C) reproducing a synthesized audio signal by performing synthesis filtering on the excitation vector using a filter coefficient based on the audio parameter. . 38. A program for executing the acoustic parameter encoding method according to claim 1 on a computer. 39. A program for executing the acoustic parameter decoding method according to any one of claims 12 to 21 on a computer. 40. An audio input device for converting an audio signal into an electrical signal, an A / D converter for converting a signal output from the audio input signal device into a digital signal, and an output from the A / D converter. 35. The audio signal encoding apparatus according to claim 34, which encodes a digital signal to be encoded, an RF modulation apparatus that performs modulation processing or the like on encoded information output from the audio signal encoding apparatus, A transmitting antenna that converts a signal output from the device into a radio wave and transmits the radio wave. 41. A receiving antenna for receiving a received radio wave, and an RF for demodulating a signal received by the receiving antenna.
A demodulation device, an audio signal decoding device for decoding information obtained by the RF demodulation device, and a D / A conversion of a digital audio signal decoded by the audio signal decoding device. D / A conversion device, and D
And an audio signal output device that converts an electrical signal output from the / A conversion device into an audio signal.