JPH0426119B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a low bit rate waveform encoding system for audio signals, and in particular to an encoding/decoding apparatus that reduces the amount of transmitted information to 10 kbit/sec or less. An effective method for encoding an audio signal with a transmission information amount of about 10k bits/second or less is to minimize the error between the input signal and the signal reproduced using the driving sound source signal sequence of the audio signal. As a condition,
A method of searching at short intervals is well known.
These methods are called tree coding (TREE CODING) and vector quantization (VECTORQUANTIZATION) depending on their search method. In addition to these methods, a plurality of pulse sequences representing the driving sound source signal sequence are transmitted at short intervals.
On the encoder side, A-b-S( A NALYSIS- B Y-
Recently, a method has been proposed that attempts to obtain the information sequentially using the SYNTHESIS method. The present invention relates to this method. For details on this method, please refer to B.S.A.T.A.R.
Eye by S.ATAL) et al. C. A. S. S. Proceedings of ICASSP, 1982
“A. New Model of L.P.C. Excitement for Producing Natural Sounding Speech at Low Bit Rates” (“ A NEW MODEL OF LPC
EXCITATION FOR PRODUCING
NATURAL−SOUNDING SPEECH ATLOW
BIT RATES") (Reference 1), so we will briefly explain it here. Figure 1 shows the processing on the encoder side in the conventional method described in Reference 1. 1 is a block diagram. In the figure, 100 indicates an encoder input terminal, into which an A/D converted audio signal sequence x(n) is input. 110 is a buffer memory circuit;
1 frame of audio signal sequence (e.g. 10msec, 8K
In the case of Hz sampling, 80 samples) are accumulated. The output value of 110 is output to a subtracter 120 and a K parameter calculation circuit 180. However, according to Reference 1, the reflection.. is used instead of the K parameter. COE FICCIENTS (REFLECTION
COEFFICIENTS), which is the same parameter as the K parameter. Using the output value of 110, the K parameter calculation circuit 180 divides the K parameter Ki representing the audio signal spectrum for each frame into 16th order (1
≦i≦16) and outputs these to the synthesis filter 130. 140 is a sound source pulse generation circuit, which generates a predetermined number of pulse sequences in one frame. Here, this pulse sequence is denoted as d(n). An example of a sound source pulse sequence generated by 140 is shown in FIG. In FIG. 2, the horizontal axis represents discrete time, and the vertical axis represents amplitude. Here, a case is shown in which eight pulses are generated within one frame. The pulse sequence d(n) generated by 140 drives the synthesis filter 130. Synthesis filter 130
inputs d(n), obtains the reproduced signal x〓(n) corresponding to the audio signal x(n), and sends this to the subtracter 120.
Output to. Here, the synthesis filter 130 is K
Input the parameters Ki, convert them to prediction parameters ai (1≦i=16), and use ai to calculate x〓(n)
Calculate. x〓(n) can be expressed as shown below using d(n) and ai. x〓(n)=d(n)+ _P〓i ⁼¹ ai·x〓(n−i)−(1) In the above equation, P indicates the order of the synthesis filter, and here P=16. The subtracter 120 converts the original signal x
(n) and the reproduced signal x(n) is calculated and output to the weighting circuit 190. 190 inputs e(n), uses the weighting function w(n), and calculates the weighting error ew(n) according to the following equation. ew(n)=w(n) ^* e(n) −(2) In the above equation, the symbol “*” represents a convolution integral. Furthermore, the weighting function w(n) performs weighting on the frequency axis, and its Z-transformed value is
(Z), the prediction parameter ai of the synthesis filter is
It is expressed by the following formula using . W(Z)=(1- _P 〓 ⁱ⁼¹ aiZ _-i )/(1- _P 〓 ⁱ⁼¹ ai・r _i・Z _-i ) −(3) In the above equation, r is 0≦r≦1. It is a constant and determines the frequency characteristics of W(Z). That is, when r=1, W(Z)=1, and the frequency characteristics are flat. On the other hand, when r=0, W(Z) has a frequency characteristic opposite to that of the synthesis filter. Therefore, the characteristics of W(Z) can be changed depending on the value of r.
Furthermore, as shown in equation (3), W(Z) is determined depending on the frequency characteristics of the synthesis filter because an auditory masking effect is utilized. In other words, this is due to the perceptual property that even if the error with the spectrum of the reproduced signal is a little large, the error is hard to notice at a location where the input audio signal has a large spectral power (for example, near a formant). FIG. 3 shows an example of the spectrum of the input audio signal and the frequency characteristics of W(Z) in a certain frame. Here, r=0.8. In the figure, the horizontal axis shows frequency (maximum 4KHz) and the vertical axis shows logarithmic amplitude (maximum 60dB). Further, the upper curve represents the spectrum of the audio signal, and the lower curve represents the frequency characteristic of the weighting function. Returning to FIG. 1, the weighted error ew(n) is fed back to the error minimization circuit 150. The error minimization circuit 150 stores the values of ew(n) for one frame, and uses them to perform weighting 2 according to the following formula.
Calculate the multiplicative error ε. ε= _N 〓 ⁿ⁼¹ ew(n) ² −(4) Here, N indicates the number of samples for calculating the squared error. In the method of Reference 1, this time length is set to 5 msec.
This corresponds to N=40 in the case of 8KHz sampling. Next, the error minimization circuit 15
0 provides pulse position and amplitude information to the sound source pulse generation circuit 140 so as to reduce the squared error ε calculated by the above equation (4). 140 generates a sound source pulse sequence based on this information. The synthesis filter 130 uses this sound source pulse sequence as a driving source to calculate a reproduced signal x〓(n). Next, subtractor 12
0, the currently determined reproduced signal x〓(n) is subtracted from the previously calculated error e(n) between the original signal and the reproduced signal,
Let this be the new error e(n). Weighting circuit 1
90 inputs e(n), calculates a weighted error ew(n), and feeds it back to the error minimization circuit 150. 150 again calculates the squared error and
The amplitude and position of the sound source pulse sequence are adjusted to reduce this. In this way, a series of processes from generation of the sound source pulse sequence to adjustment of the sound source pulse sequence by error minimization are repeated until the number of pulses in the sound source pulse sequence reaches a predetermined number.
A sound source pulse sequence is determined. This concludes the explanation of the conventional method. In the case of this method, the information to be transmitted is the K parameter Ki (1≦i≦16) of the synthesis filter and the pulse position and amplitude of the sound source pulse sequence. Any transmission rate can be achieved. Furthermore, the transmission rate
It is considered to be one of the effective methods since good playback quality can be obtained for the range of 10Kbps or less. However, this conventional method has the disadvantage that the amount of calculation is extremely large. When calculating the amplitude of a pulse position in a sound source pulse sequence, this calculates the error and squared error between the reproduced signal and the original signal based on the pulse, and feeds them back to adjust the pulse position and amplitude. This is due to what you are doing. Furthermore, this is caused by repeating this process until the number of pulses reaches a predetermined value. Furthermore, according to this conventional method, the length of the analysis frame is fixed, and when frames are switched at a portion where the power of the input audio signal sequence is large, waveform irregularities occur near the frame boundaries in the reproduced signal sequence. There is a drawback that deterioration due to continuity occurs, which greatly impairs the quality of reproduced audio. The purpose of the present invention is to use a relatively small amount of calculations, almost no quality deterioration near frame boundaries,
The object of the present invention is to provide a high-quality audio encoding/decoding device that can be applied to transmission rates of 10 Kbps or less. The audio encoding/decoding device of the present invention includes, on the transmission side, a subtraction circuit that inputs a discrete audio signal sequence and subtracts a response signal sequence derived from a drive sound source signal sequence obtained in the past from the audio signal sequence. , a parameter calculation circuit that extracts and encodes a parameter representing the short-time spectral envelope of the audio signal sequence or the subtraction result; and an impulse response sequence calculation circuit that calculates an impulse response sequence based on the parameter representing the spectral envelope. a correlation sequence calculation circuit that inputs the output sequence of the impulse response sequence calculation circuit and calculates a correlation sequence, and creates a target signal sequence based on the subtraction result, and generates a target signal sequence and the impulse response sequence. a cross-correlation sequence calculation circuit that calculates a cross-correlation sequence with
a driving excitation signal sequence calculation circuit that inputs the correlation sequence and the cross-correlation sequence and calculates and encodes the driving excitation signal sequence of the audio signal sequence; A response signal sequence calculation circuit that calculates the response signal sequence, and a multiplexer circuit that combines and outputs the output code sequence of the parameter calculation circuit representing the spectral envelope and the code sequence of the drive excitation signal sequence, and on the receiving side, the code sequence a demultiplexer circuit that inputs a code sequence of the driving excitation signal sequence and separates a code sequence of the parameter representing the spectrum envelope, and a demultiplexer circuit that decodes the driving excitation signal sequence from the separated code sequence to generate an excitation pulse sequence. A sound source pulse sequence generation circuit to generate a sound source pulse sequence, a decoding circuit to decode a code sequence of parameters representing the separated and obtained spectral envelope, and a reproduction and output of an audio signal sequence using the parameters representing the decoded spectral envelope. The invention is characterized in that it has a synthesis filter circuit. The speech encoding device of the present invention includes a subtraction circuit that inputs a discrete speech signal series and subtracts a response signal series from the speech signal series, and a short-time spectral envelope of the speech signal series or the output system of the subtraction circuit. a parameter calculation circuit that extracts and encodes parameters representing the spectral envelope; an impulse response sequence calculation circuit that calculates an impulse response sequence based on the parameters representing the spectral envelope; a correlation sequence calculation circuit for calculating a number sequence; and a cross-correlation for calculating a cross-correlation sequence between the output sequence of the subtraction circuit or a signal obtained by applying a predetermined correction to the output sequence of the subtraction circuit and the impulse response sequence. a driving excitation signal sequence calculation circuit that inputs the correlated sequence and the cross-correlated sequence and calculates and encodes a driving excitation signal sequence of the audio signal sequence; and a driving excitation signal sequence calculation circuit that inputs the driving excitation signal sequence. a response signal sequence calculation circuit that calculates the response signal sequence derived from the driving excitation signal sequence; and a multiplexer circuit that combines and outputs the output code sequence of the parameter calculation circuit and the code sequence of the driving excitation signal sequence. It is characterized by having the following. The audio decoding device of the present invention subtracts a response signal sequence derived from a drive sound source signal sequence obtained in the past from a discrete audio signal sequence, and extracts a parameter representing the short-time spectral envelope of the audio signal sequence or the subtraction result. and encode it as
searching for and encoding a driving excitation signal sequence using a cross-correlation sequence calculated using the impulse response sequence obtained from the parameters and the subtraction result, and a correlation sequence calculated using the impulse response sequence; a demultiplexer circuit inputting a code sequence output by combining the code sequence of the parameter representing the spectral envelope and separating the code sequence representing the drive excitation signal sequence and the code sequence of the parameter representing the spectral envelope; an excitation pulse sequence generation circuit that generates an excitation pulse sequence by decoding a code sequence representing the driving excitation signal sequence obtained by the above-described method; and a decoding circuit that decodes a code sequence of parameters representing the spectral envelope obtained by separation. and a synthesis filter circuit which inputs the output sequence of the sound source pulse sequence generation circuit and reproduces and outputs the audio signal sequence using the output parameters of the decoding circuit. One of the features of the speech encoding device according to the present invention is the algorithm for calculating the sound source pulse sequence.
Therefore, in the following, this algorithm will first be explained in detail. First, the sound source pulse sequence d(n) at an arbitrary time n within one frame is expressed by the following equation. d(n)= _k 〓 ^k=1 g _k・δ _o , mk −(5) Here, δn, mk represent Kronetzker's delta, which is 1 when n=m _k and when n≠m _k is 0
It is. Furthermore, g _k represents the amplitude of the pulse at position m _k . The reproduced signal x〓(n) obtained by inputting d(n) to the synthesis filter is the predicted parameter of the synthesis filter a _i (1≦i≦N _p ; Here, N _p indicates the order of the synthesis filter) Then, it can be written as the following formula. x〓(n)=d(n)+ _Np 〓 ⁱ⁼¹ aix〓(n-i) −(6) Next, within one frame of the input audio signal x(n) and the playback signal x〓(n) The weighted squared error J can be written as follows. J= _N 〓 ⁿ⁼¹ ((x(n)−x〓(n))＊w(n)) ² −(7) Here, w(n) is the impulse response of the weighting circuit, and for example, in the conventional example They may have the same characteristics.
Further, N indicates the number of samples in one frame. Equation (7) can be further transformed as follows. J= _N 〓 ⁿ⁼¹ (x(n)＊w(n) −x〓(n)＊w(n)) ² −(8) Here, the term x〓(n)＊w(n) is as follows It is transformed according to the formula. Let x〓(w)(n)=x〓(n)＊w(n) −(9). When both sides of equation (9) are Z-transformed, it is multiplied by X〓(Z)=X〓(Z)・W(Z) −(10). X〓(Z) is further multiplied as follows. X〓(Z)=H(Z)・D(Z) −(11) Here, D(Z) is the Z transformation of the sound source pulse sequence equation (5), and H(Z) is the impulse response of the synthesis filter. Indicates the Z-transform value. Substituting equation (11) into equation (10), we get X _w (Z)=D(Z). H(Z). W(Z) −(12), and H _w (Z)=H(Z). Letting W(Z) be the inverse Z-transform of equation (12) and the inverse Z-transformed value of Hw(Z) to be hw(n), the following equation is obtained. x _w (n)=d(n)*h _w (n) −(13) Here, h _w (n) represents the impulse response of the cascade-connected filter of the synthesis filter and the weighting circuit. (1
Substitute equation (5) into equation 3) to obtain the following equation. X〓 _w (n)= _K 〓 ⁱ⁼¹ g _i h _w (n−mi ₎ −(14) Here, K indicates the number of pulses generated in one frame. Substituting equations (14) and (9) into equation (8), J= _N 〓 ⁿ⁼¹ (x _w (n)− _K 〓 ⁱ⁼¹ g _i h _w (n−m _i )) ² Multiply by −(15). Therefore, equation (7) can be expressed as equation (15). The amplitude of the sound source pulse sequence that minimizes equation (15)
The calculation formulas for g _k and position m _k are derived as follows. By partially differentiating equation (15) with respect to g _k and setting it to 0, the following equation is derived. Here, ψxh (·) represents the cross-correlation sequence calculated from Xw(n) and h _w (n). Also, ψhh(・) is
represents the correlation sequence of the impulse response hω(n). A covariance function sequence or an autocorrelation function sequence is known as a correlation sequence of an impulse response. The cross-correlation sequence and covariance function sequence can be expressed as follows. Note that ψ _hh (·) is often called a covariance function in the field of audio signal processing. ψ _xh (−m _k )= _N 〓 ⁿ⁼¹ x _w (n)h _w (n−m _k ) = ψ _hx (m _k ), (1≦m _k ≦N) −(17) ψ (m _i , m _k )= _N-(ni-nk) 〓 ^n-1 h _w (n-m _i ) h _w (n-m _k ), (1≦m _k ≦N) −(18) In equation (16) According to the above, the amplitude g _k corresponding to the position m _k can be calculated using the pulse position m _k as a parameter.
As for the pulse position m _k, it is sufficient to select the m _k at which |g _k | is maximum for each pulse. This can be proven by solving equation (16) for g _i , but the proof is omitted here. This concludes the explanation regarding the derivation of this algorithm. Another feature of the speech encoding device according to the present invention is that there is almost no quality deterioration near frame boundaries, and this will be explained next using an example. FIG. 4 is a block diagram showing an example of the configuration of an encoder using the excitation pulse calculation algorithm according to equation (16). In the figure, the components given the same numbers as in FIG. 1 have the same functions as in FIG. 1, and therefore their explanations will be omitted here. In FIG. 4, each component performs the following processing for each frame. Also, let N be the number of samples in one frame. The K parameter calculation circuit 280 inputs the audio signal sequence x(n) stored in the buffer memory circuit 110 and calculates K parameters K _i (1≦i≦
N _p ). K _i is output to the K parameter encoding circuit 200. K parameter encoding circuit 2
00 encodes K _i based on, for example, a predetermined number of quantization bits, and outputs the code l _ki to the multiplexer 260 . Further, the K parameter encoding circuit 200 decodes l _ki and obtains a decoded value k _i ′(1
≦i≦N _p ) with the impulse response calculation circuit 210,
Weighting circuit 290 and synthesis filter circuit 320
Output to. The impulse response calculation circuit 210 is
Input k _i ′ and calculate h _w (n) (impulse response of a filter consisting of a cascade of a synthesis filter and a weighting circuit) in equation (13) for a predetermined number of samples. ta h _w (n)
is output to the covariance function calculation circuit 220 and the cross-correlation calculation circuit 235. The covariance function calculation circuit 220 inputs a predetermined number of samples h _w (n), and calculates the covariance ψ _hh (m _i , m _k ) of h _w (n) according to the above-mentioned equation (18). (1≦i,
K≦N) and sends it to the pulse sequence calculation circuit 2.
Output to 40. Next, the subtracter 285 extracts the audio signal sequence x(n) stored in the buffer memory circuit 110.
, the output series of the synthesis filter circuit 320 is subtracted by one frame, and the subtraction result is sent to the weighting circuit 290.
Output to. Here, as will be described later, in the synthesis filter circuit 320, a response signal sequence is obtained by using the sound source pulse sequence one frame past the current frame as a drive signal, and then the signal sequence is extended to the current frame with the drive signal set to 0 for one frame. minutes have been accumulated. In other words, if the number of meaningful samples of the impulse response of the synthesis filter is at most two frames, the audio signal sequence of the current frame is the synthesis filter output signal driven by the sound source pulse of one frame past. This is based on the idea that the signal sequence can then be expressed as the sum of the signal sequence extended to the current frame by setting the drive signal to 0, and the synthesis filter output signal sequence driven by the sound source pulse sequence of the current frame. Weighting circuit 290
inputs K _i ′ from the K-parameter encoding circuit 200 and calculates the weighting function w(n) using, for example, the conventional method.
Calculate according to formula (3). This may be calculated using other frequency weighting methods. Further, the weighting circuit 290 inputs the subtraction result of the subtracter 285, performs convolution integral calculation with this and w(n), and transmits the obtained x _w (n) to the cross-correlation coefficient calculation circuit 285.
Output to 5. The cross-correlation calculation circuit 235 is
Input x _w (n) and h _w (n), and according to equation (17) above,
The cross-correlation number ψ _xh (-m _k ) (1≦m _k ≦N) is calculated and output to the pulse sequence calculation circuit 240 . Next, the pulse sequence calculation circuit 240 receives ψ _xh (−m _k ) from the cross-correlation calculation circuit 235 and ψ _hh (mi, m _k ) (1≦
m _i , m _k ≦N), and the pulse amplitude g _k is calculated using the above-mentioned sound source pulse calculation formula (16). For example, in (16) for the first pulse, the amplitude g ₁ is determined as a function of the position m ₁ with k=1. Next, select m ₁ that maximizes |g ₁ |, and let m ₁ and g ₁ at that time be the position and amplitude of the first pulse. Next, the second pulse is found by setting k=2 in equation (16). According to equation (16), it means that the second pulse is found by subtracting the influence of the first pulse. The third and subsequent pulses can be calculated in the same way, and either the predetermined number of pulses is reached, or the error value obtained by substituting g _k and m _k of the determined pulses into equation (15) is Pulses continue to be calculated until they fall below a predetermined threshold. g _k and m _k representing the amplitude and position of the pulse sequence are
50. The encoding circuit 250 includes the excitation pulse calculation circuit 24
0, the amplitude g _k and position m _k of the sound source pulse sequence are input, these are encoded using normalization coefficients to be described later, and codes representing g _k , m _k and the normalization coefficient are output to the multiplexer 260 . It also decodes this and outputs the decoded values g _{k ′} and m k ′ of g _k and m _k to the pulse sequence generation circuit 300 . Here, although various encoding methods can be considered, a conventionally well-known method can be used to encode the amplitude g _k . For example, a method can be considered in which the amplitude probability distribution is assumed to be a normal type and an optimal quantizer for the normal type is used. Regarding this, please refer to I. by Mr. J.MAX. R. E.
Transactions. on. Information. Theory (IRETRANSACTIONS ON
INFORMATION THEORY) March 1960 issue,
“Quantizing Four.” published on pages 7-12.
minimum. "Destruction"
(“QUANTIZING FOR MINIMUM
This is explained in detail in the paper entitled "DISTORTION" (Reference 2), so the explanation will be omitted here.
Another possible method is to use the maximum value of the amplitude of a pulse sequence within one frame as a normalization coefficient, normalize each pulse amplitude with this value, and then quantize and encode it. In the case of the former method,
rms within one frame (ROOT MEAN
SQUARE) value may be used as the normalization coefficient. Next, various methods can be considered for encoding the pulse position. For example, a run-length code well known in the field of facsimile signal encoding may be used. This represents the length of the code "0" using a predetermined code sequence. Further, for encoding the normalization coefficients, conventionally well-known logarithmic compression encoding or the like can be used. It should be noted that the coding of the pulse sequence is not limited to the coding method described here, and it goes without saying that the best known method can be used. Returning again to FIG. 4, the pulse sequence generation circuit 30
0 uses the input g _k ′ and m _k ′ to calculate one frame of a sound source pulse sequence having an amplitude g _k ′ at the position m _k ′, and uses this as a drive signal to send to the synthesis filter circuit 3.
Output to 20. The synthesis filter circuit 320 inputs the K parameter quantized value K _i ' (1≦i≦N _p ) from the K parameter encoding circuit 200 and converts it into the prediction parameter a _i (1≦i≦N _p ) using a well-known method. Convert it using . Next, the synthesis filter circuit 320 inputs one frame of the drive sound source signal from the pulse generation circuit 300, adds one frame of zero to this one frame of signal, and generates a response signal sequence for the two frames of signal. Find x〓(n)′. Furthermore, when calculating a response signal sequence using the zero signal sequence of the second frame, the synthesis filter circuit 320 calculates a new K _i ′ (1≦
i≦N _p ) and use this. This is shown in the following equation. Here, the drive sound source signal d(n) represents an output pulse series from the pulse generation circuit 300 when 1≦n≦N, and represents a series of all 0s when N+1≦n≦2N. Also, in (19), a ^j _i is K _i ′ at current frame time j
The prediction parameters calculated from (1≦i≦N _p ) are
a ^j-1 _i is one frame time past frame time j-1
The prediction parameters calculated from K _i ′ of are shown respectively. Of the x〓′(n) obtained according to formula (19), the second frame x〓′(n) (N+1≦n≦2N) is the subtracter 28
5. Next, the multiplexer 260 outputs the output code of the K-parameter encoding circuit 200 and the output code of the encoding circuit 25.
Input the output sign of 0 and combine them,
It is output from the transmission side output terminal 270 to the communication path. This concludes the description of the speech encoding device according to the present invention. Next, a speech decoding device according to the present invention will be explained. FIG. 5 shows an example of the configuration of a speech decoding device according to the present invention. In the figure, decoder input terminal 35
A code sequence is input for each frame from 0, and the demultiplexer 360 separates this code sequence into a K parameter code sequence, a code sequence representing the amplitude and position of the excitation pulse sequence, and a code representing a normalization coefficient, The K-parameter code sequence is output to the K-parameter decoding circuit 380, and the remaining code sequence is output to the decoding circuit 370. The decoding circuit 370 first decodes the code representing the normalization coefficient, uses this to decode the code sequence of the excitation pulse sequence, and outputs the amplitude g _k ′ and position m _k ′ of the pulse to the pulse sequence generation circuit 420
Output to. Here, the pulse sequence generation circuit 420
performs the same operation as the pulse sequence generation circuit 300 on the encoder side in FIG. 4, generates a pulse sequence within one frame, and outputs it to the synthesis filter circuit 440. The synthesis filter circuit 440 inputs N _p K parameter decoded values K _i ′ (1≦i≦N _p ) from the K parameter decoding circuit 380 and converts them into prediction parameters a _i (1≦i≦N _p ). Convert to next,
1 drive sound source signal from the pulse sequence generation circuit 420
A frame worth of data is input, and this is used to reproduce one frame worth of the audio signal sequence. Inside this synthesis filter 440, the response signal sequence determined from the sound source pulse sequence of one frame past is added to the reproduction signal sequence determined from the sound source pulse sequence of the current frame, and the audio signal sequence is reproduced. The reproduced audio signal sequence x〓(n) is output to the buffer memory circuit 470. After the buffer memory circuit 470 accumulates x〓(n) for one frame, it outputs it through the decoder side output terminal 410.
This concludes the description of the audio decoding device according to the present invention. According to the configuration of the present invention, since the sound source pulse sequence is calculated according to equation (16), the synthesis filter is driven by the pulses seen in the conventional method of Document 1, the reproduced signal is obtained, and the reproduction signal is compared with the original signal. There is no path to adjust the pulse by feeding back errors and squared errors, and there is no need to repeat the process.
This has the great effect of significantly reducing the amount of calculations and providing good playback quality. Furthermore(1
6) In the calculation of equations, ψ _xh (−m _k ) and ψ _hh (m _i , m _k )
By calculating the values of (1≦m _i , m _k ≦N) for each frame in advance, the calculation of equation (16) is a very simplified operation of multiplication and subtraction. This has the effect of further reducing the amount of calculation. Furthermore, compared to other conventional methods for searching for sound source pulse sequences, the method according to the present invention has the advantage that better quality can be obtained for the same amount of transmitted information. Furthermore, according to the configuration of the present invention, not only when the analysis frame length is not constant, but even when the analysis frame length is constant, there is almost no deterioration of the reproduced signal near the frame boundaries due to waveform discontinuity. There is a big effect that there is no. This effect occurs on the encoder side when calculating the sound source pulse sequence of the current frame by extending the response signal sequence obtained by driving the synthesis filter using the sound source pulse sequence of the frame one frame past to the current frame. This is because the sound source pulse sequence of the current frame is calculated using the result obtained by subtracting this from the input audio signal sequence as the target signal sequence. This effect also occurs on the decoder side, which reproduces an audio signal sequence using a signal sequence that is reproduced using the received and decoded sound source pulse as a driving source, and a response signal sequence derived from the sound source pulse sequence one frame past. This is due to the configuration. In the above-described embodiment of the present invention, the sound source pulse sequence within one frame is encoded by the component 250 in FIG. 4 after all pulse sequences have been determined. It may also be configured such that encoding is included in the pulse sequence calculation, and each time one pulse is calculated, encoding is performed and the next pulse is calculated. By adopting such a configuration, a pulse sequence that minimizes errors including encoding distortion can be found, so that quality can be further improved. Furthermore, in the above-mentioned embodiment, the calculation of the pulse sequence was performed on a frame-by-frame basis, but the frame may be divided into several subframes, and the pulse sequence may be calculated for each subframe. . According to this configuration, if the frame length is N, the amount of calculation can be approximately 1/d times as large as that of the configuration shown in FIG. Here, d indicates the number of frame divisions. For example, if d=2, the amount of calculation can be reduced to about 1/2. Of course, equivalent characteristics can be obtained. Furthermore, according to the configuration of the present invention, in the embodiment on the encoder side shown in FIG. was input, and the response signal sequence was extended to the current frame. In this case, when the synthesis filter is driven by the sound source pulse sequence of one frame past, the K parameter value input one frame past is used as is, but when the sound source pulse sequence of all zeros is input for one frame, In this case, the K parameter value input at the current frame time is used. Here, even when a sound source pulse sequence of all zeros in one frame is input, the K parameter value input one frame in the past may be used as it is as the K parameter value of the synthesis filter circuit 320. In such a configuration, the configuration on the decoder side shown in FIG. 5 requires the same changes as the encoder side. In addition, in the configuration example described above, the K parameter was used as the parameter representing the spectral envelope of the short-time audio signal sequence, but it is also possible to use other well-known parameters (such as LSP parameters). good. Furthermore, the weighting function w(n) in the above equation (7) may be omitted. Note that, as is well known in the field of digital signal processing, the autocorrelation function may be calculated from the power spectrum. Alternatively, the cross-correlation function may be calculated from the cross-power spectrum. In addition, in the sound source pulse calculation formula (16) according to the present invention, the covariance function was calculated according to formula (18) with ψ _hh (゜), but this is based on calculating the autocorrelation sequence as shown in the following formula. It is also possible to configure such a configuration. Ψ _hh (|m _i −m _k |)=N−|mi−mk| 〓 n=1h _w (n) h _w (n−|m _i −m _k |, (1≦|m _i −m _k | ≦N)
(20) By adopting such a configuration, it is possible to significantly reduce the amount of calculation required to calculate ψ _hh (°), and the overall amount of calculation can also be reduced. Further, in one embodiment of the encoder having the configuration of the present invention shown in FIG. 4, the buffer memory circuit 1
Although the subtraction circuit 285 is placed after the buffer memory circuit 110, the subtraction circuit 285 is placed after the buffer memory circuit 110.
It may also be configured to be placed before. Furthermore, in FIG. 4, the K parameter calculation circuit 280 is connected before the subtraction circuit 285, and is configured to analyze the output series of the buffer memory circuit 110, but the K parameter calculation circuit 280 is connected before the subtraction circuit 285.
It may be configured such that it is connected after 5 and analyzes the 285 output series.

[Brief explanation of drawings]

Figure 1 is a block diagram showing the configuration of the conventional system, Figure 2 is a diagram showing an example of a sound source pulse sequence, and Figure 3 shows the frequency characteristics of the input audio signal sequence and the frequency characteristics of the weighting circuit shown in Figure 1. Diagram showing an example, No. 4
FIG. 5 is a block diagram showing an embodiment of the speech encoding apparatus according to the present invention, and FIG. 5 is a block diagram showing an embodiment of the speech decoding apparatus according to the invention. In the figure, 110,470...Buffer memory circuit, 120,285...Subtraction circuit, 130,
320, 440...Synthesis filter circuit, 140,
300, 420...Sound source pulse generation circuit, 150
... Error minimization circuit, 180, 280 ... K parameter calculation circuit, 190, 290 ... Weighting circuit, 200 ... K parameter encoding circuit, 240
... Sound source pulse calculation circuit, 210 ... Impulse response calculation circuit, 220 ... Covariance function calculation circuit,
235 shows a cross-correlation calculation circuit, and 250 shows an encoding circuit.

Claims (1)

[Scope of Claims] 1. On the transmission side, a subtraction circuit inputs a discrete audio signal sequence and subtracts a response signal sequence derived from a drive sound source signal sequence obtained in the past from the audio signal sequence; Alternatively, a parameter calculation circuit extracts and encodes a parameter representing the short-time spectral envelope of the subtraction result, an impulse response sequence calculation circuit calculates an impulse response sequence based on the parameter representing the spectral envelope, and the impulse response a correlation sequence calculation circuit that inputs the output sequence of the sequence calculation circuit and calculates a correlation sequence; and a correlation sequence calculation circuit that generates a target signal sequence based on the subtraction result and calculates the cross-correlation between the target signal sequence and the impulse response sequence. a cross-correlation sequence calculation circuit that calculates a number sequence; a driving excitation signal sequence calculation circuit that receives the correlation sequence and the cross-correlation sequence and calculates and encodes a driving excitation signal sequence of the audio signal sequence; a response signal sequence calculation circuit that calculates the response signal sequence derived from the driving sound source signal sequence; and a multiplexer circuit that combines and outputs the output code sequence of the parameter calculation circuit representing the spectral envelope and the code sequence of the driving sound source signal sequence. on the receiving side, a demultiplexer circuit receives the code sequence and separates the code sequence of the driving excitation signal sequence from the code sequence of the parameter representing the spectral envelope; a sound source pulse sequence generation circuit that decodes the driving sound source signal sequence to generate a sound source pulse sequence; a decoding circuit that decodes a code sequence of parameters representing the separated spectral envelope; and a decoding circuit that decodes a code sequence of parameters representing the separated spectral envelope; 1. An audio encoding/decoding device comprising: a synthesis filter circuit that reproduces and outputs an audio signal sequence using parameters representing the audio signal. 2. A subtraction circuit that inputs a discrete audio signal sequence and subtracts a response signal sequence from the audio signal sequence, and a parameter that extracts and encodes a parameter representing the short-time spectral envelope of the audio signal sequence or the output sequence of the subtraction circuit. a calculation circuit; an impulse response sequence calculation circuit that calculates an impulse response sequence based on parameters representing the spectral envelope; and a correlation sequence calculation circuit that receives an output sequence of the impulse response sequence calculation circuit and calculates a correlation sequence. a cross-correlation sequence calculation circuit for calculating a cross-correlation sequence between the output series of the subtraction circuit or a signal obtained by subjecting the output sequence of the subtraction circuit to a predetermined correction and the impulse response sequence; a driving sound source signal sequence calculation circuit that inputs a number sequence and the cross-correlation sequence and calculates and encodes a driving sound source signal sequence of the audio signal sequence; and a multiplexer circuit that combines and outputs the output code sequence of the parameter calculation circuit and the code sequence of the drive excitation signal sequence. conversion device.