JPH02280200A - Voice coding and decoding system - Google Patents

Abstract

PURPOSE:To obtain the good synthesized voice having a high degree of approximation to a sound source signal at a bit rate of about 4.8kb/s by expressing the sound source signal of one pitch section (key section) by using a small number of pulses to give an amplitude and phase and a code book to indicate the characteristics of the sound source signal. CONSTITUTION:The voice signal of a frame is divided to every pitch section corresponding to the pitch period determined from a pitch parameter. The sound source signal of one pitch section in the pitch sections is then expressed by a small number of the pulses to give the amplitude and phase and the code book 175 to express the characteristics of the sound source signal and the amplitude and phase of the pulse are so determined as to decrease the error between the voice signal and the synthesis signal obtd. by the code book 175. One code word is then selected from the code book 175 and the pitch parameter and spectrum parameter, the amplitude and phase of the pulse, and the index indicating the pitch parameter are outputted. The coding and decoding of the voice signal with high quality by a small computing quantity at the low bit rate, more particularly about 4.8kb/s are possible in his way.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention provides a method for converting audio signals to low bit rates, especially 4.8K.
The present invention relates to a speech encoding/decoding method for high-quality encoding and decoding with a relatively small amount of calculation at approximately b/s.

[Conventional technology]

As a method for encoding audio signals at a low focus rate of about 4.8 kb/s, for example, Japanese Patent Application No. 59-27243
The pitch interpolation multi-pulse method described in Japanese Patent Application No. 5 (Reference 1) and Japanese Patent Application No. 60-178911 (Reference 2) is known. According to this method, on the transmitting side, a spectral parameter representing the spectral characteristics of the audio signal and a pinch parameter representing the pitch are extracted from the audio signal for each frame, and in the voiced section of the audio signal, the sound source signal of one frame is One pitch section (representative section) out of a plurality of pitch sections (representative section) obtained by dividing one frame into pitch sections is represented by a multi-pulse, and the amplitude, one phase, spectrum, and pitch parameter of the multi-pulse in the representative section are transmitted. In the silent section, the sound source of one frame is represented by a small number of multipulses and a noise signal, and one phase of the amplitude of the multipulse and the gain and index of the noise signal are transmitted.

On the receiving side, in a voiced section, the multipulse in the representative section of the current frame and the multipulse in the representative section of the adjacent frame are used to interpolate the amplitude and phase of the multipulses, and the multipulse in the pitch section other than the representative section is interpolated. Restore the pulse and restore the frame's driving sound source signal. Furthermore, in the unvoiced section, the driving sound source signal of the frame is restored using the index and gain of the multi-pulse and the noise signal. Furthermore, the restored drive sound source signal is input to a synthesis filter using spectral parameters to output a synthesized speech signal.

(Problems to be Solved by the Invention) According to the conventional method described above, in a voiced section, a sound source signal is expressed by interpolating a small number of multipulses in a representative section and multipulses in a representative section of adjacent frames. .

However, in order to encode the two-phase amplitude of a multi-pulse,
A total of about 10 bits is required per pulse. Therefore, in order to apply to a focus rate of about 4.8 kb/s, the number of multipulses in the representative section is usually set to 4.
It is necessary to reduce the number to about 1. Therefore, with such a small number, the approximation of the sound source signal is not sufficient, and there is a problem in that the sound quality deteriorates, especially for male speakers with long pitch periods. Furthermore, in the conventional method, for pitch sections other than the representative section, the sound source signal was restored by linearly interpolating the pulses in the representative section, but the pulse has two types of parameters, amplitude and phase, and they are related to each other. Therefore, there is a problem in that the characteristics deteriorate when these are interpolated independently.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems and provide a speech encoding/decoding method that achieves good sound quality at approximately 4.8 kb/s with a relatively small amount of calculation.

[Means to solve the problem]

In the audio encoding/decoding method, which is the first invention, on the transmitting side, a spectral parameter representing a spectral envelope and a pitch parameter representing a pinch are extracted from a human-generated discrete audio signal every frame of a predetermined time length. The F voice ff signal of the frame is divided into pitch sections corresponding to the pitch period obtained from the pitch parameter, and the sound source signal of one pitch section among the pitch sections is combined with a small number of pulses and the sound source signal. A codebook representing the characteristics of and outputs the pinch parameter, the spectrum parameter, the amplitude 1 phase of the pulse, and the index representing the codeword, and on the receiving side,
generating a sound source signal in the pitch section using one phase of the amplitude of the pulse and the index; further restoring sound source signals in other pitch sections by interpolation; and driving a synthesis filter configured using the spectral parameter. It is characterized by determining and outputting synthesized speech.

The second invention, the audio encoding/decoding method, is such that on the transmitting side, a spectral parameter representing a spectral envelope and a pitch parameter representing a pitch are determined from an input discrete audio signal every frame of a predetermined time length. The audio signal of the frame is divided into pitch sections according to the pitch period determined from the pitch parameter, and the sound source signal of one of the pitch sections is combined with a small number of pulses and the characteristics of the sound source signal. Further, in other pitch sections other than the pitch section, a correction coefficient for correcting at least one of one phase of the amplitude of the pulse is determined, and the pitch parameter, the spectrum parameter, and the amplitude of the pulse are determined.

outputting the phase, an index representing the selected codeword of the codebook, and the correction coefficient, and on the receiving side,
A sound source signal in the pitch section is generated using one phase of the amplitude of the pulse and an index of the code word, and a sound source signal is restored in another pitch section using the sound source signal in the pitch section and the correction coefficient. The method is characterized in that the reconstructed sound source signal drives a synthesis filter configured using the spectral parameters to obtain and output synthesized speech.

[Effect]

In the voice encoding/decoding method according to the present invention, as shown in the block diagram of FIG. A pulse generating section 700 that generates a pulse, and a sound source signal forming section 710 that selects one from a codebook 720 of filter impulse responses or filter coefficients representing the spectral envelope of the sound source signal to form a sound source signal. It is characterized by Furthermore, the synthesis filter 73 uses the sound source signal expressed in this way.
0 to obtain playback audio.

Now, for example, let the number of pulses be 1.

A codebook consisting of a set of impulse responses of filters representing the spectral envelope of the sound source signal is used as the codebook. Let this be hi(n) (J=12M). This impulse response can be determined by various methods. For example, a predetermined number of samples per frame of a prediction residual signal obtained by LPG analysis of an audio signal is subjected to FFT (
Fast Fourier transform) is performed to determine the absolute value spectrum, and further inverse FFT is performed to determine the impulse response. Another method is to perform well-known LPC analysis on the prediction residual signal to determine the coefficients of a synthesis filter, and to determine the impulse response of this filter. In addition to the above, known methods can be used.

The amplitude g1 phase m of the pulse and the codebook are determined as follows. The audio waveform of the frame shown in FIG. 4(a) is shown. Divide the frame into pinch sections of each pitch period T,
Focusing on one pitch section (representative section) (Fig. 4 (b)
)). Let xk(n) be the audio signal in this section. The amplitude g1 phase m of the pulse in this section and the selection of the optimal code word from the codebook are performed so as to minimize the weighted error power expressed by the following equation. Weighted error power E
, is Ek-Σ((xk(n)-xk(n-m):1*w(
n) ) ” ・ ・ ・(1) However, Xk (n − m) = g − tz (n − m) *
hs(n)・・・(2) Here, w(n) represents the impulse response of the auditory weighting filter. A specific configuration example is “A New Model of LPC Excita” by Mr. Ata1 et al.
tion for producing Natura
l Sounding 5peach at low
BitRates”, Proc, ICASSP
, pp. 614-617.1982゜Reference 3). However, this filter is not necessary. x-(n) represents a sound source signal using pulses and the j-th codeword selected from the codebook, and represents the reproduced sound obtained by passing this through a synthesis filter and reproducing it. Further, h and (n) indicate impulse responses of a synthesis filter for synthesizing speech. Substituting equation (2) into equation (1), partial differentiation with respect to g, and setting O to obtain the following equation.

g −Σ X wk (n) X' wk
(n m)/ΣX' wk (n m)
' iik (n m)...(3) Here, x, k(n)=xk(n)*w (n)x '' wk
(n m) -h, (n-m)*h, (n)*w (
n)・・・・(4) Optimal g, m, that minimizes equation (1)
The set of hJ is obtained as follows. Equation (3) is calculated using the code word formed as the impulse response series h, and g and m are determined so as to minimize equation (1).

This includes g'ΣXwk (n) x, ni' (n-m
)/ΣX' wk (n m) X' ww (n
All you have to do is find g and m that maximize m). The above processing is performed for all j, g・ΣXwk (n ) X' wk (n m
)/ΣX' wll(n m)X' wk(n
The set of g, m, and j with the largest value of m) is the set to be sought.

Through the above processing, the pulse amplitude, one phase, and the code word are found in the pitch section of interest.

FIGS. 4C and 4D show the obtained pulse and the synthesized waveform obtained by driving the synthesis filter with the sound source signal of the representative section generated by the obtained pulse and the selected code word, respectively. The above processing may be performed for all pinch sections within a frame, or may be performed only for one pitch section (representative section).

On the other hand, in the unvoiced section, the entire sound source signal of one frame can be represented using conventional multi-pulse or random number codebook signals. Regarding the latter method, for example, "Code-excited" by Schrader, Attal et al.
1inear prediction (CELP)
: H4ghquality 5peech at v
937-940.198 Milk Literature 4
), etc.

〔Example〕

FIGS. 1(a) and 1(b) respectively show a speech encoding device and a speech decoding device that implement the speech encoding/decoding method according to the first invention.

First, the audio encoding method on the transmitting side will be explained. In FIG. 1(a), an audio signal is input from the input terminal 100, and the audio signal x for one frame (for example, 20 m5) is input.
(n) is stored in the buffer memo January 10.

The K parameter calculation circuit 140 calculates K as a spectral parameter representing the spectral characteristics of the audio signal of the frame.
The parameters are determined from the audio signal of the frame by well-known LPG.
The analysis is performed and only a predetermined order M is calculated. Regarding this specific calculation method, reference can be made to the above-mentioned document 1.2 regarding the parameter calculation circuit. Note that the K parameter is the same as the PARCOR coefficient.

The K-parameter encoding circuit 160 outputs a code lk obtained by quantizing the K-parameter with a predetermined number of quantization bits to the multiplexer 260, and also decodes the code lk to obtain a linear prediction coefficient a.

(i=1 to M) and weighting circuit 200. It is output to the impulse response calculation circuit 170. Regarding the method of encoding K parameters and converting K parameters to linear prediction coefficients, see "t,1ne" by J. Malhou1 et al.
ar Prediction of 5peech”
(Reference 5) and the above-mentioned publications 1.2 and the like can be referred to.

The pitch calculation circuit 130 calculates the average pitch period T from the audio signal of the frame. As this method, for example, a method based on an autocorrelation method is known, and for details, refer to the pitch extraction circuit in Document 1.2. In addition to this method, other well-known methods (for example, cepstrum method, 5TFT method, phase change open method, etc.) can be used.

The pitch encoding circuit 150 outputs a code obtained by quantizing the average pitch period BH'' with a predetermined number of bits to the multiplexer 260°, and also performs pinch classification on the decoded pitch period T' obtained by decoding the code. Circuit 205. Outputs to the sound source signal calculation circuit 220.

The codebook 175 includes a series h (n) of impulse responses of a filter representing the spectral envelope of the residual signal (n=1
-L) sets (codebooks) of 2M types are stored. Here, the codebook is created in advance by learning from filter impulse response data representing the spectral envelope of the residual signal, which is analyzed from the predicted residual signals of a large amount of speech signals. As a method for this learning, vector quantization learning method is known, for example, "
Vec, tor mouth uantization in
5peech Coding, (Proc,
IEEE.

Vol,? 3. LL 15511588.1985.
Reference 6) etc. can be referred to. Moreover, various well-known methods can be used to obtain the filter characteristics representing the spectral envelope of the residual signal. For example, LPG analysis, covariance analysis, modified cepstral analysis, etc. can be used for the residual signal. For LPG analysis and covariance analysis, Rabiner and Schaf
“Digital Processin” by Mr. er et al.
of Speech Signals” (by Prentice-Hall, 1978, 1 ref. 7).
) can be referenced. Regarding the improved cepstral analysis, please refer to "Extraction of spectral envelope by improved cepstral method" by Imai et al. (IEICE Transactions, J62-A,
21? -233 pages, 1979, 1 Reference 8), etc. The codebook 175 stores the impulse response calculation circuit 1 one by one from j=1 to j=2M for 2'4 impulse response sequences hj(n) (j=1-2').
Output to 70.

The impulse response calculation circuit 170 uses the linear prediction coefficients a,' from the K-parameter encoding circuit 160 to calculate the impulse response hw(n
) and further calculate the output hJ from codebook 175
Impulse responses X', , k (nm) obtained by performing convolution calculation according to equations (n) and (4) are output to the autocorrelation function calculation circuit 180.

Autocorrelation function calculation circuit 180, impulse response xwk
'(n-m) autocorrelation function Rhh(n) is calculated and outputted up to a predetermined delay time. Regarding the operation of the autocorrelation function calculation circuit 180, reference can be made to the above-mentioned documents 1 and 2.

The subtracter 190 subtracts the output of the synthesis filter 281 by one frame from the frame audio signal X(n), and outputs the subtraction result to the weighting circuit 200.

The weighting circuit 200 passes the subtraction result through an auditory weighting filter whose impulse response is represented by w (n),
A weighted signal x,(n) is obtained and output. For the weighting method, reference can be made to the above-mentioned documents 1 and 2.

The pitch dividing circuit 205 divides the frame audio signal into T' units using the decoded pitch period T'.

The cross-correlation function calculation circuit 210 calculates the weighting signals x, , (
n) and impulse response X'wk (nm) are input, and the cross-correlation number φ is calculated and outputted up to a predetermined delay time. For this calculation method, reference can be made to the above-mentioned documents 1 and 2.

The sound source signal calculation circuit 220 calculates representative l in the frame.
For the pitch section (representative section), codebook output h
= Amplitude g and phase m of one pulse when using (n)
seek. At this time, the above equation (3) is used to calculate g and m. Next, as described in the section on the effect, J (n) is output from the codebook 175 for 2'' types, and the above processing is repeated to (1) minimize the error power of 2 g, m;

Find the set of J(n). Then, a code indicating the index of the selected codebook is output to the multiplexer 260, and g and m are output to the encoder 230.

The encoder 230 encodes the amplitude g1 phase m of the pulse in the representative section using a predetermined number of bits and outputs the encoded signal. Also,
Information P indicating the subframe position of the representative section is encoded with a predetermined number of bits and output to the multiplexer 260. Furthermore, these are decoded and output to the drive signal restoration circuit 283.

The drive signal restoration circuit 283 generates a sound source signal in the representative section using the amplitude 1 phase of the pulse found in the representative section and the selected code word.

In other pitch sections, the amplitudes of the pulses in the representative sections of the previous and subsequent frames are used to linearly interpolate the amplitudes to obtain pulses in the other pitch sections.

Furthermore, for the selected codeword, the codewords in the representative interval are linearly interpolated to obtain an impulse response representing the spectral envelope of the residual signal in other pitch intervals. Through the above processing, the frame sound source signal is restored and generated.

The synthesis filter 281 inputs the restored sound source signal, inputs the linear prediction coefficients a,' from the K-parameter encoding circuit 160 via the weighting circuit 200, and generates a synthesized speech signal for one frame. At the same time, the influence signal for the next frame is calculated for one frame, and this is output to the subtracter 190. The calculation method of the influence signal is described in the patent application 1986-23.
Reference can be made to the specification of No. 1605 (Document 9).

The multiplexer 260 combines and outputs a code representing nine phases of the amplitude of the pulse in the representative section, a code representing the position of the representative section, a code representing the K parameter, a code representing the pitch period, and a code representing the selected code word.

Next, the audio decoding method on the receiving side will be explained. In FIG. 1(b), on the receiving side, the demultiplexer 3
00 separates the received signal and outputs it.

The decoder 310 calculates the amplitude of the pulse in the representative section.

Decode and output the phase.

The parameter decoder 320 decodes the code representing the K parameter, and converts the decoded parameter into a linear prediction coefficient a.
Convert to i' and output. It also decodes the code representing the pinch period and outputs the decoded pitch period T'.

The drive signal restoration circuit 340 performs the same operation as the drive signal restoration circuit 283 on the transmission side, and restores the sound source signal of the frame. Further, the codebook 350 stores the same codewords as the codebook 175 on the transmitting side.

The synthesis filter 360 inputs the frame sound source signal and outputs synthesized speech from a terminal 380°.

As described above, according to this embodiment, on the transmitting side, from the input discrete audio signal, a spectral parameter representing the spectral envelope and a pitch parameter representing the pitch are obtained for each frame of a predetermined time length, The audio signal of the frame is divided into pitch sections according to the pitch period determined from the pitch parameter, and the sound source signal of one pitch section among the pitch sections is divided into a small number of pulses and a code representing the characteristics of the sound source signal. calculate the amplitude and phase of the pulse so as to reduce the error between the speech signal, the pulse, and the composite signal obtained by the codebook, select one codeword from the codebook, and select the codeword from the codebook. A pitch parameter, the spectrum parameter, the amplitude two phases of the pulse, and an index representing the code word are output, and the receiving side outputs the amplitude of the pulse.

Generating a sound source signal in the pitch interval using the phase and the index, restoring the sound source signal in the other pinch interval by interpolation, and driving a synthesis filter configured using the spectral parameter to obtain synthesized speech. Output.

FIGS. 2(a) and 2(b) respectively show a speech encoding device and a speech decoding device that implement the speech encoding/decoding method according to the second invention. In Figure 1 (a), (b)
) Components with the same reference numbers as in Figure 1(a),
Since the operation is similar to that in (b), the explanation will be omitted.

In FIG. 2(a), 225 is an amplitude/phase correction calculation circuit. The amplitude/phase correction calculation circuit 225 calculates a correction coefficient for correcting the amplitude 1 phase of the pulse in the representative section in each of the pitch sections other than the representative section within the same frame.

Specifically, it is calculated as follows. Let the input voice and amplitude correction coefficient 1 phase correction coefficient in the i-th Pisochi section be X, (n), Ct, and d, respectively.

The perceptually weighted error power between the input voice in the i-th pitch section and the reproduced signal x = (n), which is reproduced by correcting the amplitude 1 phase of the pulse in the representative section and passing through the synthesis filter, can be written as follows. .

E,,,=ΣNX, (n)-c, i6 (n-T'-d,))*w (n)3"・・
-(5) Amplitude 1 phase correction coefficient c;, d= can be obtained by minimizing the above equation. The above equation is partially differentiated by the amplitude correction coefficient C6 and set to 0 to obtain the following equation.

Ci=ΣXwi (n) Mwt (n T' d
H) 72Mwt (n T' di ) Mwt
(n T' dl (6) The above equation is calculated for various phase correction coefficients d, and the set of c, d that maximizes the above equation is found.

The above processing is performed for all pitch sections other than the representative section within the frame, and the amplitude/phase correction coefficients for each section are output to the encoder 230.

The drive signal restoration circuit 285 generates a sound source signal using the pulse amplitude 8 phases and the selected code word in a representative section of the frame. In addition, in the i-th pitch section other than the representative section within the same frame, the sound source signal v (n) of the representative section is changed to the amplitude 1 phase correction coefficient c4. The sound source signal d,(n) is generated by correcting the sound source signal d,(n) using the following equation.

d, (n)=c, ・v (n-T'-d,)...
(7) However, v (n) = g-h; (n-m)...
(8) where hJ(n), g+m are the codeword of the codebook, the amplitude of the pulse, and the phase of the pulse.

In the reception side decoding device shown in FIG. 2(b), the drive signal restoration circuit 342 is replaced by the transmission side drive signal restoration circuit 285.
Perform the same movement as

As described above, according to this embodiment, on the transmitting side, from the input discrete audio signal, a spectral parameter representing the spectral envelope and a pitch parameter representing the pitch are obtained for each frame of a predetermined time length, The audio signal of the frame is divided into pitch sections according to the pitch period determined from the pitch parameter, and the sound source signal of one pitch section among the pitch sections is divided into a small number of pulses and a code representing the characteristics of the sound source signal. Furthermore, in other pitch sections other than the pitch section, the amplitude of the pulse is 1.
A correction coefficient for correcting the phase is obtained, and the pitch parameter, the spectrum parameter, the amplitude 1 phase of the pulse, an index representing the selected code word of the codebook, and the correction coefficient are output, and the receiving side outputs the correction coefficient. generating a sound source signal in the pitch section using the amplitude 1 phase and the index of the codeword; The reconstructed sound source signal drives a synthesis filter configured using the spectral parameters to obtain and output synthesized speech.

The embodiments described above are merely examples of the present invention, and various modifications thereof are possible.

For example, the calculation of the amplitude and nine phases of the pulse and the selection of the code word may be performed not only in the representative section but also in all pitch sections within the frame.

With such a configuration, the amount of information required to transmit the sound source information increases, but the characteristics are improved.

Furthermore, the representative section may be fixedly determined within the frame, such as at the center of the frame, or may be determined by searching for the best section. Regarding the latter specific method, reference can be made to the above-mentioned document 1.

Furthermore, although the number of pulses in the representative section may be two or more, the amount of transmitted information increases.

Furthermore, regarding codewords, linear interpolation may or may not be performed in pitch sections other than the representative section.

Furthermore, although the codebook is an impulse response of a filter representing the spectral envelope of the prediction residual signal of the audio signal, it may also be a coefficient of the filter.

In such a configuration, it is necessary to convert the filter coefficients into an impulse response. Specifically, as the coefficient, well-known coefficients such as a linear prediction coefficient, a parameter, a logarithmic cross-sectional area ratio, a cepstrum, and a mel-cepstrum can be used.

In addition, in this example, the parameters were encoded as spectral parameters, and LPG analysis was used as the analysis method. However, as the spectral parameters, other well-known parameters such as LSP, LPC cepstrum, cepstrum, improved cepstrum, Generalized cepstrum, mel cepstrum, etc. can also be used. Furthermore, it is possible to use the optimal analysis method for each parameter.

Further, in order to reduce the amount of calculation, calculation of the influence signal can be omitted. As a result, the drive signal restoration circuit 2
Since the 831 synthesis filter and the subtracter 190 are not necessary, it is possible to reduce the amount of calculation, but the sound quality is degraded.

Note that, as is well known in the field of digital signal processing, the autocorrelation function corresponds to the power spectrum on the frequency axis, and the cross-correlation function corresponds to the cross-power spectrum, so it is also possible to calculate from these. These calculation methods are described in “Digit” by Oppenheim et al.
al Signal Processing” (Pr.
Reference may be made to the publication entitled Entice-Hall, 1975).

(Effects of the Invention) As described above, according to the present invention, the sound source signal (representative interval) in the l pitch interval is processed using a small number of pulses giving two amplitude phases and a codebook representing the characteristics of the sound source signal. Therefore, at a bit rate of about 4.8 kb/s, there is a great effect that the approximation of the sound source signal is higher than in the conventional method, and good synthesized speech can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a speech encoding device and a speech decoding device for explaining an embodiment of the speech encoding and decoding method according to the first invention, and FIG. 2 is a block diagram of a speech encoding and decoding method according to the second invention. FIGS. 3 and 4 are block diagrams of a speech encoding device and a speech decoding device for explaining one embodiment of the encoding method, and FIGS. 3 and 4 are diagrams for explaining the present invention in detail. 110 ... 130 ... 140 ... 150 ... 160 ... 170 ... 175, 350°180 ... 205 ... 210 ... 220 ... 225 ... Parameter calculation circuit pitch encoding circuit in buffer memory pitch calculation circuit Parameter encoding circuit Impulse response calculation circuit...Code book Autocorrelation function calculation circuit Pitch division circuit Cross correlation function calculation circuit Sound source signal calculation circuit Amplitude/phase correction calculation circuit 230 260 260 28L 360,730 283, 285. 340°300 ・ ・ ・ ・ ・ Encoder multiplexer...Synthesis filter 342...Drive signal restoration circuit demultiplexer

Claims (2)

[Claims] (1) On the transmitting side, from the input discrete audio signal, a spectral parameter representing the spectral envelope and a pitch parameter representing the pitch are obtained for each frame of a predetermined time length, and the audio signal of the frame is The audio signal is divided into pitch sections corresponding to the pitch period determined from the parameters, and the sound source signal of one of the pitch sections is represented by a small number of pulses and a codebook representing the characteristics of the sound source signal. The amplitude and phase of the pulse are determined so as to reduce the error between the pulse and the composite signal obtained by the codebook, one codeword is selected from the codebook, and the pitch parameter, the spectral parameter, and the outputting the amplitude and phase of the pulse and an index representing the code word;
On the receiving side, a sound source signal of the pitch section is generated using the amplitude and phase of the pulse and the index, and a sound source signal of the other pitch section is restored by interpolation, and a synthesizer configured using the spectral parameters is generated. A speech encoding/decoding method that drives a filter to obtain and output synthesized speech. (2) On the transmitting side, from the input discrete audio signal, a spectral parameter representing the spectral envelope and a pitch parameter representing the pitch are obtained for each frame of a predetermined time length, and the audio signal of the frame is The sound source signal of one of the pitch periods is represented by a small number of pulses and a codebook representing the characteristics of the sound source signal, and In other pitch sections other than the section, a correction coefficient for correcting at least one of the amplitude and phase of the pulse is determined, and the pitch parameter, the spectrum parameter, the amplitude and phase of the pulse, and the selected codeword of the codebook are calculated. On the receiving side, the amplitude and phase of the pulse and the index of the code word are used to generate a sound source signal for the pitch section, and for other pitch sections, the signal of the pitch section is output. A speech encoding/decoding method that restores a sound source signal using a sound source signal and the correction coefficient, and uses the restored sound source signal to drive a synthesis filter configured using the spectral parameters to obtain and output synthesized speech.