JPH0378638B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a low bit rate waveform encoding method for audio signals, and particularly to an encoding method that reduces the amount of transmitted information to 10 kbit/sec or less.

As an effective method for encoding an audio signal with a transmission information amount of less than about 10k bits/second, the driving sound source signal sequence of the audio signal is set on the condition that the error between the reproduced signal and the input signal is minimized. A method of searching every short period of time is known. BSATAL of Bell Telephone Laboratories, USA
Analysis by Synthesis, in which a driving sound source signal sequence is represented by multiple pulses, and the amplitude and phase are analyzed at short intervals on the encoder side.
The method of obtaining by-Synthesis (A-b-S) method is effective. An explanation of this is published in the 1982 ICASSP proceedings, pages 614-617 (Reference 1), so a detailed explanation will be omitted here.
Since the conventional method disclosed in Reference 1 uses the A-b-S method as a means for determining the pulse sequence, it has the disadvantage that the amount of calculation is extremely large. On the other hand, Patent Application No. Sho 57-231603 (Reference 2) proposes a method for greatly reducing the amount of calculations required to obtain the above-mentioned pulse sequence. It has been reported that these methods can provide good playback quality at transmission rates of 10k bits/second or less.

The conventional method disclosed in Document 2 (Patent Application No. 1983-231603) will be briefly explained. A driving sound source sequence consisting of K pulse sequences within one frame is expressed as follows.

d(n)= _K 〓 ^k=1 g _k δ(n-l _k ), n=0, 1,..., N-1 −(1) Here, δ(・) is KRONECKER's δ . N is the frame length,
g _k represents the amplitude of the pulse standing at position l _k . d(n)
The reproduced signal x obtained by inputting to the synthesis filter
(n) is the prediction coefficient of the synthesis filter α _i (i=1,
..., M, M is the order of the synthesis filter), then it can be written as follows.

x〓(n)=d(n)+ _M〓 ⁱ⁼¹ α _i x〓(n-i) −(2) 1 of input audio signal x(n) and reproduced signal x〓(n)
The weighted squared error within the frame is J= _N-1 〓 ⁿ⁼¹ ((x(n)−x〓(n))*w(n)) ² −(3). Here, * is a symbol for convolution integral, and w(n) represents a weighted function. x(n),
The Z transformations of x〓(n) and w(n) are respectively X(z) and X〓
(z) and W(z), it is expressed as follows.

J=|X(z)W(z)−X〓(z)W(z)| ²⁻ (4) Here, |·| represents an absolute value. Also (2)
From the relationship of the formula, x〓(z) becomes as follows.

x〓(z)=H(z)D(z) −(5) where H(z) is the Z transformation of the synthesis filter, and D(z) is the Z transformation of the driving sound source. Substituting (5) into (4), J=|X(z)W(z)−H(z)W(z)D(z)| ²⁻ (6).

Therefore, the inverse Z-transformed signals of X(z)W(z) and H ₍ z)W(z) are respectively expressed as
(n) and hw(n)=h(n)*w(n), (6)
becomes as follows.

J= _N-1 〓 ^k=1 (x _w (n) − _K 〓 ^k=1 g _k h _w (n-l _k )) ² −(7) Sound source pulse sequence that minimizes equation (7) amplitude of
To find g _k and position l _k , the equation (7) is partially differentiated with respect to g _k and set to 0, that is, Take advantage of the relationship between

Here, _xh (・) represents the cross-correlation function sequence calculated from x _w (n) and h _w (n), and _hh (・) represents the autocorrelation function sequence of h _w (n), as follows. is expressed in Note that _hh (・) is also called the covariance function.

_xh (l _k ) _N-1 〓 〓 ^k=1 x _w (n) h _w (n−l _k )= _hx (−l _k )0≦l _k ≦N−
1(9) _hh (l _i , l _j )= _N-(li-lj)+1 〓 ⁿ⁼⁰ h _w (n-l _i ) h _w (n-l _j )0≦l _i , l _j ≦ N-1 (10) The conventional method determines the amplitude and position of the k-th pulse by regarding g _k in (8) as a function of only l _k . In other words, the l _k that maximizes |g _k | in (8) is k
Let g k be the position of the kth pulse, and g _k at that time be the amplitude of the kth pulse. This method is g _k
If g is exactly a function of only l _k , the sound source pulse sequence that minimizes equation (7) can be calculated, but this is not the case for actual audio signals, and in general g _k is expressed as l ₁ , l ₂ ,
..., l _k , etc.

FIG. 1 is a block diagram showing an embodiment of the speech encoding method described in this specification. Second
FIG. 1 is a flowchart showing a processing procedure for determining the amplitude g _k and position l _k of a sound source pulse sequence, which is performed in the sound source pulse sequence calculating circuit 140 of FIG. 1 according to the conventional method of Document 2.

Hereinafter, the components of the embodiment of the speech encoding apparatus shown in FIG. 1 and the sound source pulse sequence search algorithm according to the conventional method shown in Reference 2 shown in FIG. 2 will be explained in detail.

In FIG. 1, each component performs processing for each frame. 100 indicates the encoder input terminal, A/
A D-converted audio signal sequence x(n) is input.
110 is a buffer memory circuit that stores one frame worth of audio signal series. K parameter calculation circuit 1
80 inputs the audio signal x(n) stored in the buffer memory circuit 110 and calculates a predetermined number of K parameters K _i (1≦i≦M). This value is output to the K parameter encoding circuit 190. K parameter encoding circuit 190 encodes K _i based on, for example, a predetermined number of quantization bits, and outputs the code I _ki to multiplexer 160 . Further, the K parameter encoding circuit 190 decodes I _ki and decodes the decoded value K′ _i (1≦i≦
M) is output to the impulse response calculation circuit 120 and the weighting circuit 200. The weighting circuit 200 is
Inputting the input audio signal x(n) and the K-parameter decomposed value K′ _i , the above-mentioned x _w (n) is calculated using the weighting function w(n) that depends on the frequency characteristics of the synthesis filter. x _w (n) is output to the cross-correlation calculation circuit 135. Note that the weighting function w(n) used here is such that, for example, the Z transformation W( _z ) is expressed as W(z)=( 1− _P 〓 ⁱ⁼¹ α _i Z ^-i )/
(1− _P 〓 ⁱ⁼¹ α _i r ⁱ Z ⁻ⁱ ) is adopted.
The impulse reaction calculation circuit 120 inputs K′ _i ,
The above-mentioned n _w (n) (convolution integral of the impulse response and the above-mentioned weighted function) is calculated for a predetermined number of samples, and the obtained h _w (n) is calculated by the covariance function calculation circuit 130 and the cross-correlation function. It is output to the calculation circuit 135. The covariance function calculation circuit 130 inputs a predetermined number of samples h _w (n),
According to equation (10) above, _hh (l _i , l _j ) (0≦l _i , l _j ≦N
−
1) and sends it to the sound source pulse sequence calculation circuit 1.
Output to 40. Cross correlation calculation circuit 135
calculates the cross-correlation number between the input x _w (n) and h _w (n), and executes the sound source pulse sequence calculation circuit 140.
Output to. Next, the sound source pulse sequence calculation circuit will be explained. The sound source pulse sequence calculation circuit 140 _{receives xh} (l _k ) (0≦l _k ≦
N-1) from the covariance function calculation circuit 130 to _hh
Input (l _i , l _j ) (0≦l _i , l _j ≦N−1), respectively.
The amplitude g _k and position l _k of the sound source pulse sequence are calculated using the above-mentioned pulse calculation algorithm equation (8).

FIG. 2 is a flowchart showing the processing procedure performed by the sound source pulse sequence calculation circuit 140 in the conventional method of Document 2. For the first pulse, in equation (8), set K = 1, the amplitude g ₁ is a function of the position l ₁ , g ₁ = _xh
Expressed as (l ₁ )/ _hh (l ₁ , l ₁ ). Next, select l ₁ that maximizes |g ₁ |, and let l ₁ and g ₁ at that time be the first pulse position and amplitude. The second pulse is
In equation (8), set k = 2 and maximize |g ₂ |
Select l ₂ and let l ₂ and g ₂ be the position and amplitude of the second pulse. The third and subsequent pulses are calculated in the same manner until the predetermined number of pulses is reached. In FIG. 2, 1 initializes to 1 a calculation counter that calculates the number of pulses.
2 is a comparison, in which it is determined whether the number of pulses is larger or smaller than a predetermined number, and if it is larger than the predetermined number, the pulse sequence calculation process is finished. 3 calculates equation (8), where l ₁ ,..., l _k-1 and g ₁ ,..., g _k-1 are known, and |g _k | is maximized. Then, g _{k and} l _k at that time are output as the amplitude and position of the k-th pulse. 4 is addition, which increments the contents of the calculation counter that calculates the number of pulses by one. This concludes the description of the sound source pulse calculation circuit 140.

Returning to FIG. 1, the encoding circuit 150 inputs the amplitude g _k and position l _k of the pulse sequence output from the excitation pulse calculation circuit 140 and encodes them. Conventionally well-known methods can be used to encode the amplitude g _k and the position l _k . Regarding the amplitude g _k , for example, a method can be considered in which the maximum value of the amplitude of the pulse sequence within one frame is used as a normalization coefficient, the amplitude of each pulse is normalized with this value, and then quantized and encoded. For the position l _k , it is conceivable to use run-length encoding, which is well known in the field of facsimile signal encoding, for example. This represents the length of the code "0" using a predetermined code sequence. The multiplexer 160 includes a K-parameter encoding circuit 190
and the output code of the encoding circuit 150 are input, and by combining these,
Output from 0 to the communication path.

The method of searching for a driving sound source pulse sequence in the conventional method of Document 2 has been described above.

In the conventional method of Reference 2, in an algorithm for determining the amplitude and position of a sound source pulse sequence, it is assumed that the amplitude of a pulse is a function only of the position at which the pulse stands. However, the above assumption does not necessarily hold true for actual audio signals;
In the conventional method, g _k is generally
It becomes a function of l ₁ , l ₂ ..., l _k , etc. Therefore, the sound source pulse sequence determined by the conventional method of Document 2 does not truly reduce J in the above equation (7), and a more suitable sound source pulse sequence exists. In a method in which the driving sound source signal sequence is represented by multiple pulses, in order to obtain even better audio quality in the region where the transmission rate is 10k bits/second or less, it is necessary to find a more suitable amplitude and position of the sound source pulse sequence. Become.

An object of the present invention is to provide a speech encoding method that improves the sound source search algorithm described above and provides high quality speech with a relatively small amount of calculation.

According to the present invention, a discrete audio signal sequence is input, the audio signal sequence is divided into short time periods to obtain a short time audio signal sequence, and a parameter representing a spectral envelope is extracted from the short time audio signal sequence.
Calculating the autocorrelation sequence of the impulse response sequence corresponding to the spectral envelope, calculating the cross-correlation sequence of the impulse response sequence corresponding to the spectral envelope and the short-time audio signal sequence, When sequentially determining a sound source pulse sequence suitable as a drive sound source signal sequence, the autocorrelation sequence and the cross-correlation sequence are used to adjust the amplitude of the previously determined sound source pulse sequence and the amplitude of the newly determined sound source pulse. While doing so, find the position of the newly determined sound source pulse and use the adjusted amplitude as the amplitude of the sound source pulse sequence to find the drive sound source pulse,
A speech encoding method is obtained, characterized in that the parameter representing the spectral envelope and the driving excitation pulse are encoded, and the code of the parameter representing the spectral envelope and the code representing the driving excitation signal sequence are combined and output. .

Further, according to the present invention, a discrete audio signal sequence is input, the audio signal sequence is divided into short-term audio signal sequences to obtain a short-term audio signal sequence, and a parameter representing a spectral envelope is extracted from the short-term audio signal sequence. , adding a predetermined correction to the spectrum envelope, calculating an autocorrelation sequence of an impulse response sequence having the corrected spectrum, and applying a predetermined correction based on the short-time audio signal sequence. When calculating a cross-correlation sequence using the added target signal sequence and sequentially obtaining a sound source pulse sequence suitable as a driving sound source signal sequence for the short-time audio signal sequence, the autocorrelation sequence and the cross-correlation sequence are Drive by finding the position of the newly determined sound source pulse while adjusting the amplitude of the previously determined sound source pulse sequence and the amplitude of the newly determined sound source pulse using a number sequence and using the adjusted amplitude as the amplitude of the sound source pulse sequence. A sound source pulse is obtained, a parameter representing the spectral envelope and the driving sound source pulse are encoded, and the code of the parameter representing the spectral envelope and the code representing the driving sound source signal sequence are combined and output. An encoding method is obtained.

The speech encoding method according to the present invention is characterized by an algorithm for obtaining the above-mentioned sound source pulse sequence. Hereafter, when the above equation (7) is given, the amplitude g _k of the sound source pulse sequence that minimizes J in equation (7), k=1, ..., K
The algorithm of the present invention for determining the positions l _k , k=1, . . . , K will be explained.

The formula that expresses the squared error when K pulses are added is J= _N-1 〓 ⁿ⁼⁰ (x _w (n) − _K 〓 ^k=1 g _k h _w (n−l _k )) ² −(11 ) with respect to g _k , k=1, ..., K and set it to 0, we get
We get the following relation. _xh (l _k ) _K 〓 ⁱ⁼¹ g _{i hh} (l _i , l _k ), k=1,..., K −(12) g _i , i=1,..., K that satisfies equation (12) is as follows It is found as the solution of the simultaneous equations.

The squared error at this time is calculated as follows.

J= _{N-1 〓} ⁿ⁼⁰ x ² _w (n ₎ − _K 〓 ^k= 1 g _{k 1} ,..., g _k-1 and l ₁ ,..., l _k-1 are determined, and by viewing equation (12) as a function of only l _k , we can sequentially calculate g so that |g _k | is maximum. This is to find _k and l _k . The present invention also uses a sequential method, but when calculating the sound source pulse sequence, l ₁ ,...
Assuming that l _k-1 has been determined, gi, i=1, . . . , k and l _k that minimize equation (14) are found by varying l _k .

Next, as an embodiment of the present invention, an algorithm that can efficiently calculate a sound source pulse sequence by decomposing the matrix of equation (13) will be described.

In order to solve the simultaneous linear equations in equation (13), the K×K positive symmetric matrix on the left side is
(CHOLESKY) When decomposed, it is expressed as follows.

〓〓〓 ^t 〓=〓 −(15) In equation (15), 〓 is a K×K lower triangular matrix with mutually orthogonal column vectors, 〓 is a K×K diagonal matrix, 〓 is g _i , i =1, . . . , a column vector with K as an element, 〓 is _xh (l _i ), i=1, .

Now, if we represent the elements of 〓 as v _ij , 1jiK, and the elements of 〓 as d _i , 1iK, then Therefore, the following gradual relationship exists between the elements of 〓 and 〓.

v _ii = 1 − (17) v _ij = { _hh (l _i , l _j ) − _j-1 〓 ^k=1 v _ik d _k v _jk } /d _j , 1ji-1 − (18) d ₁ = _hh (l ₁ , l ₁ ) −(19) d _i = _hh (l _i , l _i ) − _i-1 〓 ^k=1 v ² _ik d _k , 2iK − (20) With these 〓 and 〓, the column vector 〓 and the squared error J are calculated.

〓〓〓 ^t 〓=〓 −(21) If we set 〓〓=〓 −(22), 〓=〓〓 ^-1 〓 However, 〓 ^t =〓 ^-1 −(23) The squared error J is (14) From formula, (22), and (23), J= _N-1 〓 ⁿ⁼⁰ x ² _w (n)−〓 ^t 〓= _N-1 〓 ⁿ⁼⁰ x ² _w (n)−〓 ^t 〓 ^{- 1} 〓 ^t 〓= _N-1 〓 ⁿ⁼⁰ x ² _w (n)−〓 ^t 〓 ^-1 〓= _N-1 〓 ⁿ⁼⁰ x ² _w (n)− _N-1 〓 ⁿ⁼⁰ y ² _k /d _k −(24) Here, y _i , i=1, .

y ₁ = _xh (l ₁ ) − (25) y _i = _xh (l _i ) − _i=1 〓 ^j=0 v _ij y _j , 2iK − (26) Next, the recurrence formula derived above (17) Equation, (18),
An algorithm for determining the acoustic pulse position l _k , k=1, . . . , K will be explained using equations (19), (20), (25), and (26).

The position l _k is sequentially determined so as to minimize the squared error expressed by equation (24), that is, to maximize ² _y /d _k . For l ₁ ,
d ₁ in equation (19), y ₁ to l ₁ in equation (25) = {l｜max(〓〓(l)/ _hh (l, l), 0l
N-1} - (27). Now, suppose that the positions l ₁ , l ₂ , ..., l _i-1 have been determined, and consider finding l _i . Since up to l ₁ , ..., l _i-1 have been determined, from the assumption, from equation (18), the elements of 〓 have been calculated up to row (i-1), and therefore, equation (20) Therefore, the values of d ₁ , ..., d _i-1 , and from equation (26), the values of y ₁ , ..., y _i-1 are also determined. In this case, v _iv is a function of only l _i , and l _i is It is required as. v _ij in equation (28) is v _iv = _hh (l ₁ , l ₁ ) −(29) v _ij = { _hh (l, l _j ) − _j-1 〓 ^k=1 v _ik d _k v _jk } /d _j −(30), which is calculated while including the variable l.

If l _i , i=1, ..., K are determined by equation (28), the elements of the matrices 〓, 〓, and 〓 are determined, and 〓
can be found recursively from the following relationship from equation (23).

g _i =y _i /d _i − _K 〓 ^j=i+1 v _ji g _j・1ik−1 −(31) The initial condition is g _k =y _k /d _k −(32). This concludes the explanation regarding the algorithm of the present invention.

As described above, the present invention is characterized by an algorithm for determining a sound source pulse sequence. Here, one embodiment implemented by the sound source pulse sequence calculation circuit 140 according to the present invention will be described in more detail using a flowchart. In Figure 3, 5
is to calculate l ₁ by calculating the above equation (27).
6 is for setting initial values, and d ₁ and y ₁ are calculated from l ₁ determined in 5 using the above equation (19) and the above (25). 7 is a comparison, the calculated number of pulses is
It is determined whether it is larger or smaller than a predetermined number, and when the predetermined number is reached, the process of calculating the position of the pulse is stopped and the process moves to step 12 to calculate the amplitude of the pulse. 8 calculates vi _ij over 1ji-1 using equations (29) and (30). Reference numeral 9 determines the pulse position l _i and detects the maximum value of the equation (28). 10 is
This is to calculate d _i , and the l _i found in step 9 is calculated using
Determined by substituting into the formula. 11 is for calculating y _i , which is determined by substituting l _i determined in 9 into the above equation (26). 12 is for finding the amplitude of the pulse, and using the above equation (31) and the above equation (32), g _i , i=1, . . . , K are calculated. Above,
This concludes the explanation of the sound source pulse calculation circuit according to the present invention.

According to the configuration of the present invention, in calculating the sequence of sound source pulses, the optimum amplitude when l ₁ ,...l _K are given by equation (13) is determined, and the optimum position is determined sequentially by equation (28). Therefore, there is no assumption that the amplitude of a pulse is a function only of the position of the pulse, as seen in the conventional method of Reference 2, and it is possible to obtain a more suitable sound source pulse sequence in the sense of reducing the square error. can. Therefore, there is an effect that better sound quality can be obtained than in the conventional method. In addition, since the sound source pulse sequence search algorithm of the present invention is all given by first-order recurrence formulas as described above, it has the effect of significantly reducing the amount of calculation compared to the conventional method of Reference 1. be.

In addition, in the calculation algorithm of the sound source pulse sequence according to the present invention described above, all pulses were calculated at once based on the above formula (13), but _hh (・)
Because hh decays exponentially, it can be said that the influence of _hh (・) on equation (13) is small when its order is large. Therefore, the recurrence formula (20) mentioned above
When the absolute value of l _i −l _j is larger than a predetermined value in v _ij that appears in the calculations of equations (26), (28), (30), and (31), v _ij Even if the value of is ignored, an approximately optimal sound source pulse sequence can be calculated. Such a configuration lowers the order of the matrix on the left side of equation (13), and can greatly reduce the amount of calculation required.

Furthermore, although the calculation of the sound source pulse sequence according to the present invention is performed in frame units as described above, the frame may be divided into several subframes and the pulse sequence may be calculated for each subframe. According to this configuration, when the number of frame divisions is m, the amount of calculation can be increased by approximately 1/m compared to the configuration shown in FIG.

Further, in the configuration example described above, the frame length is kept constant, but it may be made variable. The characteristics will improve if it is made variable. Further, although the K parameter is used as the parameter representing the spectral envelope of the short-time audio signal sequence, other well-known parameters (for example, LSP parameters, etc.) may also be used. Furthermore, the aforementioned weight function w
(n) may be omitted.

In addition, in the sound source pulse calculation formula (13) according to the present invention, the covariance function was calculated for _hh (・) according to formula (10), but this is calculated by calculating the autocorrelation sequence as shown in the following formula. It may be configured like this.

_hh (l _i , l _j )= _N-(li-lj)+1 〓 〓 ⁿ⁼⁰ h _w (n) h _w (n-|l _i , l _j |), 0|l _i −l _j |
N-1-(33) By adopting such a configuration, it is possible to significantly reduce the amount of calculation required to calculate _hh (.), and there is an effect that the amount of calculation as a whole can also be reduced.

Furthermore, in the present invention, when calculating the autocorrelation sequence of the synthesis filter, the impulse response of the synthesis filter is first determined and then calculated according to equation (10), but the autocorrelation sequence is calculated by inverse Fourier transforming the power spectrum of the synthesis filter. It can be found by In addition, in the present invention, the cross-correlation sequence between the impulse response of the synthesis filter and the input audio signal is calculated according to equation (9), but the product of the power spectrum of the synthesis filter and the power spectrum of the input audio signal is inverse Fourier transformed. It can be found by

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the audio encoding method of the present invention, FIG. 2 is a flowchart showing the processing procedure performed by a conventional sound source pulse sequence calculation circuit, and FIG. 3 is a sound source according to the present invention. 3 is a flowchart showing a processing procedure of an embodiment implemented by a pulse sequence calculation circuit. In the figure, 110...buffer memory circuit,
120... Impulse response calculation circuit, 130...
Covariance function calculation circuit, 135...Cross correlation sequence calculation circuit, 140... Sound source pulse sequence calculation circuit,
150... Encoding circuit, 160... Multiplexer, 180... K parameter calculation circuit, 190...
...K parameter encoding circuit, 200...Weighting circuit, 1...Initialization, 2...Comparison, 3...Pulse calculation formula, 4...Addition, 5, 6...Initialization, 7...
Comparison, 8, 9, 10, 11...Pulse position calculation,
12... each shows pulse amplitude calculation.

Claims (1)

[Scope of Claims] 1. A discrete audio signal sequence is input, the audio signal sequence is divided into short time periods to obtain a short time audio signal sequence, and a parameter representing a spectral envelope is extracted from the short time audio signal sequence. , calculate the autocorrelation sequence of the impulse response sequence corresponding to the spectral envelope, calculate the cross-correlation sequence of the impulse response sequence corresponding to the spectral envelope and the short-time audio signal sequence, and calculate the cross-correlation sequence of the impulse response sequence corresponding to the spectral envelope and the short-time audio signal sequence When successively determining a sound source pulse sequence suitable as a driving sound source signal sequence of a series, the amplitude of the previously determined sound source pulse sequence and the newly determined sound source pulse are calculated using the autocorrelation sequence and the cross-correlation sequence. determining the position of a newly determined sound source pulse while adjusting the amplitude and using the adjusted amplitude as the amplitude of the sound source pulse sequence to obtain a driving sound source pulse, encoding the parameter representing the spectral envelope and the driving sound source pulse, A speech encoding method characterized in that a code of the parameter representing the spectral envelope and a code representing the drive excitation signal sequence are combined and output. 2. Input a discrete audio signal sequence, divide the audio signal sequence into short-time intervals to obtain a short-time audio signal sequence, extract a parameter representing a spectral envelope from the short-term audio signal sequence, and add a parameter to the spectral envelope in advance. A predetermined correction is applied, an autocorrelation sequence of an impulse response sequence having the corrected spectrum is calculated, and a signal sequence to which a predetermined correction is added to the short-time audio signal sequence is added to the signal sequence and the above correction is added. When calculating the cross-correlation sequence with the impulse response sequence and sequentially obtaining the sound source pulse sequence suitable as the driving sound source signal sequence of the short-time audio signal sequence, the autocorrelation sequence and the cross-correlation sequence are calculated. The position of the newly determined sound source pulse is determined while adjusting the amplitude of the previously determined sound source pulse sequence and the amplitude of the newly determined sound source pulse using , the parameter representing the spectral envelope and the driving excitation pulse are encoded, and the code of the parameter representing the spectral envelope and the code representing the driving excitation signal sequence are combined and output. Method.