WO2001024164A1

WO2001024164A1 - Voice encoder, voice decoder, and voice encoding and decoding method

Info

Publication number: WO2001024164A1
Application number: PCT/JP2000/006542
Authority: WO
Inventors: Tadashi Yonezaki
Original assignee: Matsushita Electric Industrial Co., Ltd.
Priority date: 1999-09-28
Filing date: 2000-09-25
Publication date: 2001-04-05
Also published as: EP1132891A1; AU7321200A; JP3360046B2; JP2001100798A

Abstract

A voice analyzer (101) in a voice encoder (100) extracts the fundamental frequency and spectral envelope information from an input voice signal. A fundamental frequency quantizer (102) quantizes the fundamental frequency. A matrix generator (103) derives a spectral envelope from the spectral envelope information, and a spectral envelope quantizer (104) quantizes the spectral envelope. A multiplexer (105) multiplexes the quantized spectrum envelope and the quantized fundamental frequency for transmission. In a voice decoder (200), a spectral envelope composer (202) restores the quantized spectral envelope from the spectral envelope information, and a voice synthesizer (203) extracts the spectral envelope based on the fundamental frequency information to synthesize the decoded voice. Thus, high-quality voice decoding can be achieved in the case of transmission at a low bit rate.

Description

Description Speech coding apparatus, speech decoding apparatus, and speech coding / decoding method

The present invention relates to a speech encoding device, a speech decoding device, and a speech coding / decoding method used for a communication device of a wireless communication system such as a mobile phone and a mobile phone. Background art

In recent years, in the field of wireless communication systems, where demand has been rapidly increasing, development of devices capable of encoding and decoding speech at a low bit rate and high quality is being promoted for effective use of radio wave resources.

FIG. 1 is a block diagram showing the configuration of a conventional speech encoding device and speech decoding device.

In speech encoding apparatus 1 of FIG. 1, spectrum envelope analysis section 11 estimates spectral envelope information of an input speech signal. The spectrum envelope quantization unit 12 quantizes the spectrum envelope information estimated by the spectrum envelope analysis unit 11. The inverse filter 13 filters the inverse characteristic of the frequency characteristic of the spectrum envelope information quantized by the spectrum envelope quantization unit 12 with respect to the input audio signal, and removes the spectrum envelope component. Thus, a signal having a flat frequency characteristic can be obtained. This signal is considered to mimic the sound source signal generated in the vocal cords during the vocalization process. Hereinafter, this signal is referred to as “sound source signal”.

Excitation codebook 14 stores signals having flat frequency characteristics. Source coding section 15 searches source codebook 14 for a signal closest to the source signal, and outputs the code (hereinafter referred to as “source code”).

The multiplexing unit 16 receives the spectrum envelope output from the spectrum envelope quantization unit 12. A code indicating the quantized value of the link information and the excitation code output from excitation encoding section 15 are multiplexed as a code sequence and transmitted to the communication channel.

In speech decoding apparatus 2 in FIG. 1, demultiplexing section 21 separates the received code string into a code indicating a quantized value of spectrum envelope information and an excitation code.

In excitation codebook 22, the same signals as in excitation codebook 14 are stored. Sound source selecting section 23 selects and extracts a signal corresponding to the received sound source code from sound source codebook 22.

The synthesis filter 24 filters the signal extracted by the sound source selection unit 23 so as to have the frequency characteristic of the received spectrum envelope information, and outputs a decoded voice.

As described above, in the conventional speech coding apparatus and speech decoding apparatus, the spectral envelope information having different dynamic ranges and quantization characteristics of the signal is separated from the source signal, and the quantizer corresponding to each characteristic is separated. Thus, high-quality speech codec is realized.

However, in the above-described conventional speech coding apparatus and speech decoding apparatus, since filtering is performed based on the result of quantizing the spectrum envelope information, the spectrum envelope information is reduced by reducing the bit rate. If sufficient accuracy cannot be obtained in the quantization of the sound source, there is a problem that the excitation signal cannot be flattened, the quantization efficiency decreases, and the quality of the decoded speech deteriorates. Disclosure of the invention

An object of the present invention is to provide a speech coding apparatus, a speech decoding apparatus, and a speech codec method capable of implementing high-quality speech decoding even when transmitting information at a low bit rate. is there.

The purpose of this is to focus on the fact that a speech signal is generated by cutting out a temporally continuous spectrum envelope surface based on the fundamental frequency. And speech source information are completely separated to realize speech codec processing that is not affected by the quantization accuracy of the spectrum envelope information, and a high-efficiency quantization method of the spectrum envelope information that is effective in the analysis / synthesis model This is achieved by realizing a high-efficiency speech codec through. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a block diagram illustrating a configuration of a conventional speech encoding device and a conventional speech decoding device. FIG. 2 is a block diagram illustrating a configuration of the speech encoding device and the speech decoding device according to the first embodiment of the present invention. ,

FIG. 3 is a block diagram showing an internal configuration of a spectrum envelope quantization unit of the speech coding apparatus according to Embodiment 2 of the present invention.

FIG. 4 is a block diagram showing an internal configuration of a spectrum envelope quantization unit of the speech coding apparatus according to Embodiment 3 of the present invention.

FIG. 5 is a model diagram of a spectrum envelope curved surface according to Embodiment 3 of the present invention, and FIG. 6 is an internal configuration of a spectrum envelope quantization unit of a speech encoding device according to Embodiment 4 of the present invention. A block diagram showing the

FIG. 7 is a block diagram showing an internal configuration of a spectrum envelope quantization unit of a speech coding apparatus according to Embodiment 5 of the present invention.

FIG. 8 is a block diagram showing an internal configuration of a model applicator of a speech coding apparatus according to Embodiment 6 of the present invention.

FIG. 9 is a block diagram illustrating an internal configuration of a parameter quantizer of a speech coding apparatus according to Embodiment 7 of the present invention.

FIG. 10 is a block diagram showing an internal configuration of a parameter quantizer in a speech coding apparatus according to Embodiment 8 of the present invention.

FIG. 11 is a block diagram showing an internal configuration of a parameter overnight quantizer of a speech coding apparatus according to Embodiment 9 of the present invention, and FIG. 12 is a block diagram showing the internal configuration of the spectrum envelope configuration unit of the speech decoding device according to Embodiment 10 of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(Embodiment 1)

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

In speech encoding apparatus 100 in FIG. 2, speech analysis section 101 extracts a fundamental frequency and short-time spectrum envelope information from an input speech signal. The fundamental frequency quantization unit 102 quantizes the fundamental frequency extracted by the speech analysis unit 101.

As for the speech analysis that extracts the fundamental frequency and short-time spectrum envelope information from the input speech signal, see [Hideki Kawahara, Ikuyo Masuda, "Conversion of speech using interpolation in the time-frequency domain," Technical report ΕΑ96-28, ρρ.9-18, 1996], etc., discloses a method based on a STRAIGHT analysis / synthesis model. In this model, the sound source information is only the fundamental frequency and is completely independent of the spectrum envelope information.Therefore, the quantization error of the sound source information and the spectrum envelope information affects the quantization of each other's information. Has no effect.

The matrix generation unit 103 generates a spectrum envelope surface on a time-frequency plane by arranging the short-time spectrum envelope information extracted by the speech analysis unit 101 along the time axis. The spectrum envelope quantization unit 104 quantizes the spectrum envelope curved surface generated by the matrix generation unit 103.

The spectral envelope information is quantized as a continuous function on the time-frequency plane.If only the extracted spectral envelope is quantized, the spectral envelope information is quantized depending on the sound source information. This is because information that is the gist of the present invention cannot be separated in the quantization processing. The multiplexing unit 105 includes a code indicating the quantized value of the spectrum envelope curved surface output from the spectrum envelope quantization unit 104 and the basic value output from the fundamental frequency quantization unit 102. A code indicating a quantized value of the frequency is multiplexed and transmitted to the communication path.

In speech decoding apparatus 200 in FIG. 2, demultiplexing section 201 converts the received code string into a code indicating the quantized value of the spectrum envelope information and a code indicating the quantized value of the fundamental frequency. To separate.

The spectrum envelope configuration unit 202 reconstructs a quantized spectrum envelope surface from the received spectrum envelope information. The speech synthesis unit 203 synthesizes and outputs a decoded speech by cutting out the spectrum envelope curved surface reconstructed by the spectrum envelope construction unit 202 based on the fundamental frequency information.

Next, the flow of the information processing operation of the speech coding apparatus and the speech decoding apparatus according to the present embodiment shown in FIG. 2 will be described.

First, a fundamental frequency and short-time spectrum envelope information are extracted from an input speech signal input by a speech analysis unit 101 of the speech coding apparatus 100. The extracted fundamental frequency is quantized by the fundamental frequency quantization unit 102.

On the other hand, the extracted short-time spectrum envelope information is arranged along the time axis in a matrix generation unit 103, and a spectrum envelope curved surface on a time-frequency plane is generated. The spectrum envelope curved surface is quantized by a spectrum envelope quantization unit 104.

The quantized fundamental frequency and spectrum envelope curved surface are multiplexed by the multiplexing unit 105 and transmitted to the communication path. Then, the fundamental frequency and the spectrum envelope surface are received by the demultiplexing unit 201 of the speech decoding apparatus 200, and are separated into a quantized value of the spectrum envelope information and a quantized value of the fundamental frequency. .

The quantized value of the spectrum envelope information is input to the spectrum envelope construction unit 202, and the spectrum envelope curved surface is reconstructed in the spectrum envelope construction unit 202. Then, the speech synthesis unit 203 uses the reconstructed spectrum envelope surface as a basis. By extracting based on this frequency information, the decoded speech is synthesized and output.

In this way, by independently quantizing the sound source information and the spectrum envelope information, the quantization efficiency of one of the information is reduced and the quantization efficiency of the other information is reduced. Therefore, even when information is transmitted at a low bit rate, high-quality speech decoding can be realized.

(Embodiment 2)

FIG. 3 is a block diagram showing an internal configuration of the spectrum envelope quantization unit of the speech encoding device according to Embodiment 2 of the present invention.

Note that the configuration of the speech coding apparatus according to the present embodiment is the same as the configuration of the speech coding apparatus shown in FIG. 2 of Embodiment 1, and a description thereof will be omitted.

In the spectrum envelope quantization unit 104 in FIG. 3, the two-dimensional orthogonal transformer 310 performs two-dimensional orthogonal transformation on the spectrum envelope curved surface in the time axis direction and the frequency axis direction. The parameter overnight quantizer 302 quantizes the transform coefficients obtained by the two-dimensional orthogonal transform process in the two-dimensional orthogonal transformer 301.

Here, in general, the difference between the high-frequency components of the spectrum envelope curved surface is hard to be recognized by hearing. Therefore, even if speech is synthesized on the decoding side using only the low frequency component coefficient information obtained by the orthogonal transform, the speech quality does not significantly deteriorate. Therefore, the parameter overnight quantizer 302 quantizes only the coefficient information of the low frequency component.

As described above, by using the orthogonal transform, information that is not perceptually important can be deleted, and high-quality speech decoding can be realized even when information is transmitted at a low bit rate.

(Embodiment 3)

FIG. 4 is a block diagram showing the internal configuration of the spectrum envelope quantization unit of the speech encoding device according to Embodiment 3 of the present invention.

The configuration of the speech coding apparatus according to the present embodiment is shown in FIG. 2 of Embodiment 1. Since the configuration is the same as the configuration of the speech encoding device shown, the description is omitted.

In the spectral envelope quantization unit 104 of FIG. 4, the model applicator 311 models a spectrum envelope curved surface and extracts model parameters.

This model is a model of a spectrum envelope surface in the time-frequency space.For example, as shown in Fig. 5, the application of the all-pole model to both sections on the time axis of the spectrum envelope surface and its application It can be modeled by interpolation.

The parameter overnight quantizer 302 quantizes the model parameter overnight extracted by the model applicator 311.

In this way, by modeling the spectral envelope surface, the quantization efficiency of the spectral envelope surface can be improved, and high quality speech decoding can be achieved even when transmitting information at a low bit rate. can do.

(Embodiment 4)

FIG. 6 is a block diagram showing the internal configuration of the spectrum envelope quantization unit of the speech encoding device according to Embodiment 4 of the present invention.

In the spectrum envelope quantization unit 104 of FIG. 6, the time axis orthogonal transformer 3221 performs orthogonal transformation in the time axis direction on the spectrum envelope curved surface. The model applicator 3 1 1 applies a model according to the order on the time axis to the orthogonally transformed time axis conversion coefficient, and extracts parameters. The parameter overnight quantizer 3 02 quantizes the model parameters extracted by the model applicator 3 1 1.

In this way, by applying a model according to the order of the time-base transform coefficient, the quantization efficiency by modeling can be improved, and even when information is transmitted at a low bit rate, high-quality speech can be obtained. Decoding can be achieved.

(Embodiment 5) FIG. 7 is a block diagram showing the internal configuration of the spectrum envelope quantization unit of the speech encoding device according to Embodiment 5 of the present invention.

In the spectral envelope quantization unit 104 of FIG. 7, the time-axis orthogonal transformer 331 1 performs orthogonal transformation on the spectrum envelope curved surface in the time-axis direction, and models the orthogonally transformed time-axis transformation coefficient. Classify into those that do and those that do not. As this classification method, for example, there is a method of applying the all-pole model because the coefficient of the 0th order on the time axis is a spectrum envelope obtained by averaging the spectrum envelope surface, and not applying the model to the other coefficients.

The model applicator 3 1 1 applies a model corresponding to the order on the time axis to a part of the orthogonally transformed time axis transform coefficients, and extracts parameters. The frequency axis orthogonal transformer 333 performs orthogonal transformation in the frequency axis direction for the time axis conversion coefficient to which no model is applied. The parameter quantizer 302 quantizes the model parameters extracted by the model applicator 311 and the transform coefficients output from the frequency-axis orthogonal transformer 332.

In this way, by applying the model only to the time-domain transform coefficients for which the quantization efficiency is improved by the modeling, the modeling distortion is reduced while the quantization efficiency is improved by the modeling, and at a low bit rate. Even when information is transmitted, high-quality speech decoding can be realized.

(Embodiment 6)

FIG. 8 is a block diagram showing an internal configuration of a model applicator of a speech encoding device according to Embodiment 6 of the present invention.

Note that the model applicator 3 11 according to the present embodiment is the one shown in any of the above-described third to fifth embodiments.

The model parameter overnight estimator 4 01 applies the model to the input signal and Extract the evening.

For example, in the case of speech coding, the input signal is modeled using an all-pole model in consideration of the speech generation process. However, when the order of the model is low, the model cannot represent the zero included in the signal, and the model causes analysis distortion.

Therefore, the model error estimator 402 estimates the analysis distortion generated when applying the model and outputs it to the parameter quantizer.

In this way, by quantizing the modeling distortion, the modeling distortion is reduced while improving the quantization efficiency by modeling, and high-quality speech decoding is performed even when information is transmitted at a low bit rate. Can be realized.

(Embodiment 7)

FIG. 9 is a block diagram showing an internal configuration of a parameter quantizer in a speech coding apparatus according to Embodiment 7 of the present invention.

Note that the parameter quantizer 302 according to the present embodiment is as described in any of Embodiments 2 to 5 above.

The weight calculator 501 determines the quantization sensitivity for each quantization target value using the fundamental frequency information. Hereinafter, an example of a method of determining the quantization sensitivity in the weight calculator 501 will be described.

In the speech decoding process, the spectrum envelope surface is cut out according to the fundamental frequency and connected on the time axis to generate a decoded speech. At this time, the harmonic amplitude value of the fundamental frequency is important information as compared with other spectral amplitude values. Therefore, a weight coefficient surface is generated by weighting the harmonic amplitude value at the extracted spectrum envelope position.

Next, the quantization sensitivity is determined for each quantization target value by performing conversion using the same method as that for obtaining the quantization target value and calculating a weight coefficient in the quantization target parameter space.

The weight calculator 502 uses the spectral envelope information for each quantization target value. Determine the quantization sensitivity. Hereinafter, an example of a method of determining the quantization sensitivity in the weight calculator 502 will be described.

When the same amount of noise is added to a signal, a signal with a small dynamic range is more noticeable in hearing than a large signal. Therefore, a weighted coefficient surface is generated which is heavily weighted as the amplitude is smaller in the spectrum envelope surface.

Next, the quantization sensitivity is determined for each quantization target value by performing conversion using the same method as that used to obtain the quantization target value and calculating the weighting coefficient in the parameter space to be quantized. . Since the adaptation method of the quantizer in the weight calculator 502 is also required in the decoding process, it is necessary to use the spectrum envelope information quantized in the previous frame in order to synchronize with the decoding process. desirable.

The statistic accumulator 503 stores a statistic for each quantization target value obtained in advance. The quantization generator 504 includes the quantization sensitivity for the quantization target value output from the weight calculator 501 and the weight calculator 502, and the statistics stored in the statistic storage 503. Design quantizer from quantity.

For example, when a scalar quantizer is used, the variance of the quantization target value is accumulated as a statistic, and the quantization step width is determined based on the variance and the quantization sensitivity. When the dispersion is the same, the quantization sensitivity is large. That is, the quantization step width is set to be smaller than the quantization target value that is easily affected by the quantization error. The quantizer 505 quantizes the value to be quantized based on the design result of the quantization generator 504.

Thus, by applying the quantizer to the fundamental frequency and the spectrum envelope information, it is possible to reduce objective quantization distortion and auditory distortion of the synthesized speech signal.

In the present embodiment, the quantization sensitivity for each quantization target value is determined using two pieces of information of the fundamental frequency information and the spectrum envelope information. The quantization sensitivity may be determined using one of the two pieces of information to design a quantizer.

(Embodiment 8)

FIG. 10 is a block diagram showing the internal configuration of the parameter quantizer in the speech coding apparatus according to Embodiment 8 of the present invention.

The error scale determiner 511 adaptively determines the quantization error scale on the spectrum envelope using the fundamental frequency information. The error scale determiner 512 adaptively determines a quantization error scale on the spectrum envelope using the spectrum envelope information. The error scale synthesizer 5 13 synthesizes the error scales obtained by the error scale determiner 5 11 and the error scale determiner 5 12 into one error scale.

The codebook 5 14 stores quantized values. The spectrum envelope constructor 515 converts the quantized values stored in the codebook 514 into a spectrum envelope curved surface. The spectrum envelope constructor 516 converts the quantization target value into a spectrum envelope curved surface.

The error calculator 517 is based on the error scale output from the error scale synthesizer 513, and the spectral envelope surface and the spectral envelope formed by the spectral envelope composer 515 are used. Calculate the error with respect to the spectrum envelope curved surface composed by the constructors 516.

The code selector 518 selects a code corresponding to the quantized value with the smallest error from the codebook 514 and outputs the selected code.

As described above, the error of the spectrum envelope curved surface is calculated on the time-frequency plane using the error scale at the time of quantization adapted to the fundamental frequency and the spectrum envelope information. Objective quantization distortion and auditory distortion can be reduced.

In this embodiment, both the fundamental frequency and the spectrum envelope information are used. Although the quantization error scale on the spectrum envelope is determined by using the above method, the quantization error scale may be determined for any one of them, and the error may be calculated.

(Embodiment 9)

FIG. 11 is a block diagram showing an internal configuration of a parameter quantizer in the speech coding apparatus according to Embodiment 9 of the present invention.

The error function determiner 521 adaptively determines a quantization error weighting function on the spectrum envelope using the fundamental frequency information. The error function determiner 522 adaptively determines a quantization error weight function on the spectrum envelope using the spectrum envelope information. The error function synthesizer 5 2 3 synthesizes the quantization error weight function obtained by the error function determiner 5 2 1 and the error function determiner 5 2 2 into one error function. The error function converter 524 defines an error measure over a quantization parameter for converting the quantization error weight function output from the error function synthesizer 523.

The codebook 525 stores quantized values. The error calculator 526 calculates an error between the quantization target value and the quantized value stored in the codebook 525 based on the error measure output from the error function converter 524.

The code selector 527 selects a code corresponding to the quantized value that minimizes the error from the codebook 525 and outputs the selected code.

As described above, by calculating the error between the quantization parameters using the error scale at the time of quantization adapted to the fundamental frequency and the spectral envelope information, the objective of the synthesized speech signal can be reduced with a small amount of processing. Quantization distortion and audible distortion can be reduced.

In the present embodiment, the quantization error weight function on the spectrum envelope is determined for both the fundamental frequency and the spectrum envelope information, but the quantization error weight function is determined for one of them, and the error May be calculated. (Embodiment 10)

FIG. 12 is a block diagram showing the internal configuration of the spectrum envelope configuration unit of the speech decoding device according to Embodiment 10 of the present invention.

Note that the configuration of the speech decoding apparatus according to the present embodiment is the same as the configuration of the speech decoding apparatus shown in FIG.

Here, as described in the second embodiment, in the speech codec using the orthogonal transform, information is compressed by not transmitting a high-frequency component that is not perceptually important on the code side. Therefore, in the present embodiment, an envelope surface is generated on the decoding side by complementing a parameter that has not been received by using a parameter value that has been statistically obtained in advance.

In the spectral envelope composing unit 202 shown in FIG. 12, the parameter storing unit 601 stores the parameter value statistically obtained in advance corresponding to each parameter not to be quantized. I have. The spectrum envelope generator 602 generates a spectrum envelope curved surface based on the input spectrum envelope information.

As described above, by using a value statistically obtained as a parameter that is not subject to quantization, a more accurate spectrum envelope surface can be restored as compared with a case where an arbitrary value is used.

As described above, according to the speech coding apparatus, the speech decoding apparatus, and the speech coding / decoding method of the present invention, the spectrum envelope information and the sound source information are completely separated from each other, and the quantum of the spectrum envelope information is obtained. It is possible to realize speech codec processing that is not affected by quantization accuracy, and to realize highly efficient speech codec processing through a high-efficiency quantization method of spectrum envelope information that is effective in an analysis-synthesis model. Therefore, even when information is transmitted at a low bit rate, high-quality speech decoding can be realized. The present specification is based on Japanese Patent Application No. 11-27551 19 filed on Sep. 28, 1999. This content is included here. INDUSTRIAL APPLICABILITY The present invention is suitable for use in a communication terminal apparatus, which is a base station apparatus of a wireless communication system that performs wireless communication of voice data.

Claims

The scope of the claims

1. Speech analysis means for extracting the fundamental frequency and spectrum envelope information from the input speech signal, fundamental frequency quantization means for quantizing the extracted fundamental frequency, and spectrum analysis from the extracted spectrum envelope information. A matrix generating means for generating a vector envelope surface, a spectral envelope quantization means for quantizing the generated spectral envelope surface, and a multiplexing of the quantized value of the spectral envelope surface and the quantized value of the fundamental frequency. And a multiplexing means for multiplexing and transmitting.

2. The spectrum envelope quantization means performs two-dimensional orthogonal transformation on the spectrum envelope curved surface in the time axis direction and the frequency axis direction, and quantizes the transform coefficient output from the two-dimensional orthogonal transformation means. 2. The speech encoding apparatus according to claim 1, further comprising a parameter quantization unit that performs the parameter quantization.

3. The spectral envelope quantizing means comprises: a model applying means for modeling a spectral envelope surface to extract model parameters, and a parameter quantizing means for quantizing the extracted model parameters. A speech coding apparatus according to claim 1.

4. The spectral envelope quantization means is a time axis orthogonal transform means for performing orthogonal transform in the time axis direction on the spectrum envelope curved surface, and a model corresponding to the order on the time axis for the orthogonal transformed time axis transform coefficient. 2. The speech encoding apparatus according to claim 1, comprising: a model application unit that applies parameters to extract parameter data; and a parameter parameter quantization device that quantizes the extracted model parameters.

5. A frequency axis orthogonal transform means for performing orthogonal transform in the frequency axis direction on a time axis transform coefficient to which no model is applied, wherein the parameter quantizing means includes an extracted model parameter set and the frequency axis orthogonal transform. 5. The speech encoding apparatus according to claim 4, wherein the transform coefficient output from the means is quantized.

6. The model applying means is a model parameter estimating means for applying a model to the input signal and extracting parameters, and an analysis distortion generated when the model is applied by the model parameter estimating means. And a model error estimating means for estimating The speech encoding device according to range 3.

7. Parameter quantization means includes weight calculation means for determining quantization sensitivity for each quantization target value using at least one of the fundamental frequency information and spectrum envelope information, and a pre-determined quantization target value for each quantization target value. A statistic accumulation means for accumulating the statistic; a quantization sensitivity for the quantization target value output from the weight calculating means and a quantization for designing a quantizer based on the statistic accumulated in the statistic accumulation means. 3. The speech encoding apparatus according to claim 2, comprising: a generating unit; and a quantizing unit that quantizes a quantization target value based on a design result of the quantization generating unit.

8. The parameter quantization means includes: an error scale determination means for adaptively determining a quantization error scale on the spectrum envelope using at least one of the fundamental frequency information and the spectrum envelope information; and a codebook. First spectral envelope forming means for converting the quantized value accumulated in the spectrum into a spectral envelope surface, second spectral envelope forming means for converting the quantization target value into a spectral envelope surface, An error between the spectrum envelope curved surface constituted by the first spectrum envelope composing means and the spectrum envelope curved surface constituted by the second spectrum envelope composing means is calculated based on an error scale. 3. The speech encoding apparatus according to claim 2, comprising: an error calculating unit that performs the calculation; and a code selecting unit that selects a code corresponding to the quantization value that minimizes the error from a codebook.

9. The parameter quantizing means includes: an error function determining means for adaptively determining a quantization error weight function on the spectrum envelope using at least one of the fundamental frequency information and the spectral envelope information; Error function conversion means for defining an error measure over the quantization parameter for converting the quantization error weight function, and an error between the quantization target value and the quantized value stored in the codebook based on the error measure. 3. The speech encoding apparatus according to claim 2, comprising: an error calculation unit that calculates the error;

10. A code string transmitted from the speech coding apparatus according to claim 1 is separated into a code indicating a quantized value of spectral envelope information and a code indicating a quantized value of a fundamental frequency. Demultiplexing means, a spectrum envelope constructing means for reconstructing a quantized spectrum envelope surface from the received spectrum envelope information, and a fundamental frequency information representing the reconstructed spectrum envelope surface. And a speech synthesizing means for synthesizing the decoded speech by extracting the speech based on the speech.

11. The spectral envelope composing means includes a parameter storing means for storing parameter values statistically obtained in advance corresponding to each parameter not to be quantized, and an input spectrum. 10. The speech decoding device according to claim 10, further comprising: a spectrum envelope generating unit configured to generate a spectrum envelope curved surface based on the envelope information.

1 2. On the code side, extract the fundamental frequency and spectrum envelope information from the input speech signal, quantize the extracted fundamental frequency, generate a spectrum envelope surface from the extracted spectrum envelope information, quantize it, The quantized value of the spectrum envelope surface and the quantized value of the fundamental frequency are multiplexed and transmitted. On the decoding side, the received code string is converted into a code representing the quantized value of the spectrum envelope information and the fundamental frequency. It is separated into a code indicating the quantization value, the quantized spectrum envelope surface is reconstructed from the received spectrum envelope information, and the reconstructed spectrum envelope surface is cut out based on the fundamental frequency information. A speech codec decoding method for synthesizing decoded speech.

1 3. The procedure for causing the computer to extract the fundamental frequency and the spectrum envelope information from the input audio signal, the procedure for quantizing the extracted fundamental frequency, and the step of extracting the spectrum envelope surface from the extracted spectrum envelope information. A speech code for executing a generating step, a step of quantizing the generated spectral envelope surface, and a step of multiplexing the quantized value of the spectral envelope surface and the quantized value of the fundamental frequency. A machine-readable recording medium that records a program.

14. The computer separates the received code sequence into a code indicating the quantized value of the spectral envelope information and a code indicating the quantized value of the fundamental frequency, and further quantizes the code sequence from the received spectral envelope information. A procedure for reconstructing the spectrum envelope surface that has been reconstructed, and extracting and reconstructing the reconstructed spectrum envelope surface based on the fundamental frequency information. And a machine-readable recording medium on which a speech decoding program for executing a procedure for synthesizing a signal speech is recorded.