JPS60186898A - Voice encoding system and apparatus - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a low bit rate waveform encoding method for audio signals, and particularly to an encoding method and apparatus for reducing the amount of transmitted information to 10 bits/second or less.

(Prior art and its problems) An effective method for encoding an audio signal with a transmission information amount of less than about 10 bits per second is to reproduce the driving sound source signal sequence of the audio signal using the A well-known method is to search for short periods of time, subject to the minimum error between the signal and the input signal.

These methods use tree encoding (TRI) depending on their search method.
C0DING), vector quantization (VECTOQ)
It is called UA, NT I ZAT I ON).
There is. In addition to these methods, analysis pie synthesis (ANALYSIS) is also used to generate a plurality of pulse sequences representing the drive excitation signal sequence at short intervals on the encoder side.
-BY-8YNTHE8IS;A-B-8) A method has recently been proposed in which the method is attempted to be sequentially performed. The present invention relates to this method.

For more information on this method, see Pi Nis Attar (B.
-8-ATAL) et al.'s I.C.N.N.I.N.S.P. (Engineering/C-A-8-8-P) proceedings,
[A New Model of LPC Excitation, published in 1982, pp. 614-617.
"For Producing Natural Sounding Speech at Low Bits" (A
NEW MODEL 0FLPCBXCITATION
FORPRODUCING NATURAL -8O
UNDING 8PFiECHAT LOW BIT
Since it is explained in a paper entitled "RATES" (Reference 1), a brief explanation will be given here.

FIG. 1 is a block diagram showing the processing on the encoder side in the conventional method described in Document 1.

In the figure, 100 indicates an encoder input terminal, into which an A/D converted audio signal sequence x(,) is input.

Reference numeral 110 is a backup memory circuit, which stores the audio signal series as 1.
A frame (for example, in the case of 8 KHz sampling and a frame length of 10 m5ec, 80 samples) is accumulated. The output value of 110 is output to a subtracter 120 and a parameter calculation circuit 180. However, according to Document 1, the parameters of the reflection coefficient (REFIJCTIONCOEFFICIEN)
TS ), which is the same parameter as . The K parameter is also called the PARCOR coefficient. The K-parameter calculation circuit 180 uses the output value of 110 to obtain 16-order parameters Ki (1≦i≦16) representing the audio signal spectrum for each frame according to the covariance method, and outputs these to the synthesis filter 130. do. 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 a (n). Sound source pulse generation circuit 140
FIG. 2 shows an example of a sound source pulse sequence generated by. 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 the sound source pulse generation circuit 140 is
The synthesis filter 130 is driven. The synthesis filter 30 inputs d(n), obtains the reproduced signal ˜(n) corresponding to the audio signal (n), and outputs this to the subtracter 120 .

Here, the synthesis filter 30 inputs the K parameters Ki, converts them into prediction parameters ai (k≦1≦16), and calculates the reproduced signal X(n) using ai.

1(n) can be expressed as shown below using d(n) and ai.

In the above formula, P indicates the order of the synthesis filter, here P=1
It is set at 6. The subtracter 120 calculates the difference, (,) between the original signal Ma(n) and the reproduced signal x(n), and outputs the weighting to the circuit 190. 190 inputs e(n), uses the function w(n) for the weighting, and calculates the weighting error ew(n) according to the following equation.

ew(n) = W (,) e(n) (2) In the above equation, the symbol -1 represents a convolution integral.

In addition, the weighting function w(n) is a weighting function on the frequency axis, and if its two-converted value is W(z), it can be expressed by the following equation using the prediction parameter ai of the synthesis filter. It will be done.

In the above equation, r is a constant of O≦r≦1, and determines the frequency characteristics of W(Z). If J, y=1, W(Z)
= 1, the Norr wavenumber characteristic becomes flat.

On the other hand, when r=0, W(2)] has a frequency characteristic opposite to that of the composite nolter. Therefore, depending on the value of r, W
The characteristics of (Z) can be changed. Furthermore, as shown in equation (3), the reason W is determined depending on the frequency characteristics of the synthesis filter is because an auditory masking effect is utilized. In other words, in places where the spectral power of the input audio signal is large (for example, near formants),
Even if the error with the spectrum of Reproduction No. 46 is a little large, the error is difficult to hear due to the audible property. FIG. 3 shows the spectrum of the input audio signal in a certain frame and an example of the frequency characteristics of W (Z). Here, it was set as r-08. In the figure, the horizontal axis represents frequency (maximum 4 KHz), and the vertical axis represents logarithmic amplitude (maximum 60 dB). The upper curve represents the spectrum of the audio signal, and the lower curve represents the frequency characteristics 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 calculate the weighted squared error ε according to the following equation.

ε−Σe w (n)2(4) −1 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 m5ec, which is N = 40 in the case of 8KH2 sampling.
corresponds to Next, the error minimization circuit 150 performs the above (4)
) The pulse position and amplitude information is given to the sound source pulse generation circuit 140 so as to reduce the squared error ε calculated by the equation. 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; (n). Next, the subtractor 120
Now, the error e(n) between the original signal and the reproduced signal calculated earlier is
The playback signal currently obtained from ? (n) is subtracted and this is set as a new error e(n). The weighting circuit 190 is
(n) is input, the weighting calculates the error eW(n), and feeds this back to the error minimization circuit 150. The error minimization circuit 150 calculates the squared error again, and adjusts the amplitude and position of the sound source pulse sequence to reduce the squared error.

In this way, a series of processes from generation of a 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, and the 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 parameter Ki (1≦1≦16) of the synthesis filter, the pulse position and amplitude of the sound source pulse sequence, and is arbitrary depending on the number of pulses generated within one frame. transmission rate can be achieved. Furthermore, it is considered to be one of the effective methods since good reproduction sound quality can be obtained for a transmission rate range of 16 Kbps to 10 Kbps.

However, this conventional method has the disadvantage that the amount of calculation is extremely large. When calculating the position and amplitude of a pulse in a sound source pulse sequence, this calculates the error and square error between the reproduced signal and the original signal based on the pulse, and feeds them back to reduce the square error. This is due to the fact that the pulse position and amplitude are adjusted accordingly. Furthermore, this is caused by repeating this process until the number of pulses reaches a predetermined value.

Furthermore, this conventional method has the disadvantage that, at a bit rate of about 10 Kbps or less, the reproduction quality deteriorates when an input signal with a high pitch frequency is input, for example when a female voice is input. This is because, in order to reproduce this pitch waveform satisfactorily, a larger number of sound source pulses is required compared to the case of a speaker with a low pitch frequency. Therefore, for this reason, it has been difficult to significantly reduce the transmission bit rate, that is, to significantly reduce the number of pulses within one frame, without deteriorating quality.

(Objective of the Invention) The object of the present invention is to achieve a speed of 10 Kbps with a relatively small amount of calculation.
An object of the present invention is to provide an audio encoding method and an apparatus thereof that can reproduce high-quality audio even at the following bit rates.

(Structure of the Invention) According to the present invention, on the transmitting side, a discrete audio signal sequence is input, a pitch parameter representing a pitch fine structure and a spectral parameter representing a short-time spectral envelope are extracted and encoded, and the spectral parameters are also calculate the autocorrelations of the impulse response sequence according to the short-time spectral envelope, calculate the crosscorrelations according to the audio signal sequence and the impulse response sequence, and calculate the autocorrelations and the A first pulse sequence that can satisfactorily represent the audio signal sequence is encoded using the cross-correlation coefficients and the pitch parameter, and a code representing the first pulse sequence, the pitch parameter, and the spectrum are encoded. A receiving side inputs the combined code and separates the code representing the first pulse sequence, the code representing the pitch parameter, and the code representing the spectrum parameter. decode, determine a second pulse sequence based on the decoded first pulse sequence, taking into account the decoded pitch parameter, and use the second pulse sequence and the decoded spectral parameter. There is obtained an audio encoding system characterized in that the audio signal sequence is reproduced by using the audio signal sequence.

Further, according to the present invention, a parameter calculation circuit inputs a discrete audio signal sequence and extracts and encodes a pitch parameter representing a pitch fine structure and a spectral parameter representing a short-time spectral envelope from the audio signal sequence; an autocorrelation calculation circuit that inputs the output sequence of the calculation circuit and calculates the autocorrelation of the impulse response sequence according to the short-time spectrum of the audio signal sequence; and an output sequence of the audio signal sequence and the parameter calculation circuit. a cross-correlation calculation circuit that calculates a cross-correlation number represented by the audio signal sequence and an impulse response sequence corresponding to the short-time spectrum; The output series of the cross-correlation number calculation circuit and the output series of the parameter calculation circuit are input, and the audio signal series is well represented using the autocorrelation number and the cross-correlation number and taking the pitch parameter into consideration. a first pulse sequence calculation circuit that encodes the obtained first pulse sequence; and an output code of the parameter calculation circuit and the first pulse sequence.
and a multiplexer circuit that combines and outputs the output sequence of the pulse sequence calculation circuit.

Further, according to the present invention, a demultiplexer circuit inputs the combined code sequence and separates the code representing the sound source pulse sequence, the code representing the pitch parameter, and the code representing the spectral parameter;
a first pulse sequence decoding circuit inputting and decoding a code representing the pulse sequence obtained separately; a pitch parameter decoding circuit inputting and decoding a code representing the separately obtained pitch parameter; a spectral parameter decoding circuit that inputs and decodes a code representing the spectral parameter; and a spectral parameter decoding circuit that inputs and decodes a code representing the spectral parameter; a second pulse sequence generation circuit that generates a second pulse sequence based on the pulse sequence by taking into account the decoded pitch parameter; and an output sequence of the second pulse sequence generation circuit and the spectral parameter decoding circuit. There is obtained an audio decoding device characterized in that it has a synthesis filter circuit which inputs an output sequence of , and reproduces and outputs an audio signal sequence.

(Example) In general, a sound source sequence has very strong periodicity in voiced parts. In the present invention, by searching for sound source pulses using this periodicity, good sound quality is provided with a small number of sound source pulses.

The configuration of the audio encoding system according to the present invention will be explained in detail below using the drawings. FIG. 4(a) is a block diagram showing an embodiment of the encoder side of the speech encoding system according to the present invention, and FIG. 4(b) is a block diagram showing an embodiment of the post-encoder side. It is. In FIG. 4(a), the audio signal sequence, (
n) is input from the input terminal 195, separated by a predetermined number of samples, and sent to the buffer memory circuit 3.
It is accumulated to 40. Next, the parameter calculation circuit 280 inputs a predetermined number of samples of the audio signal stored in the buffer memory circuit 340, and uses this to calculate LPG parameters of a predetermined order P. Calculate according to the method (e.g. linear predictive analysis method). Various LPC parameters can be considered, but below, the parameter Ki (1≦i≦P
) will be used in the explanation. The K parameter is the same parameter as the Percoll coefficient. The K parameter Ki is output to the parameter encoding circuit 200. K
The parameter encoding circuit 200 encodes , based on a predetermined number of quantization bits, and generates codes 1, 1
is output to multiplexer 450. Further, the K parameter encoding circuit 200 uses 'i as the decoded parameter value obtained by decoding lki, converts it into a prediction coefficient value 'i (i≦i≦p) according to a well-known method, and generates an impulse The response calculation circuit 210 and weighting are output to a circuit 410 and a synthesis filter circuit 400.

Next, the pitch analysis circuit 37.0 uses the buffer memory circuit 3
The pitch period pd is calculated using the audio signal for one frame, which is the output of 40. As a method for calculating Pd, for example, R. V. Cox (R, V.

IEE Transactions on NissP (rEEE
TRANSACTIONS ON A-8-8-P
) magazine, February 1983 issue, pages 258-272.
Speech for Late Modification and Coding
LEMENTATION OF TIME DO1'1
AIN HARMONIC5 CALING OF 5P
EECHFORRATE MODIFICATIONA
There is a known method of calculation using the autocorrelation coefficients of the audio signal, which is described in a paper titled "ND C0DING" (Reference 2).In addition, calculations using other well-known methods are also possible. It can also be calculated from the prediction residual signal after predicting the audio signal.Pitch encoding circuit 3
80 inputs the pitch period Pd, performs quantization encoding using a predetermined number of quantization bits, and outputs the code 1d to the gate circuit 460. Further, the pitch encoding circuit 380 outputs the PA obtained by decoding 1d to the pulse calculation circuit 390 and the pulse generation circuit 420.

Next, the impulse response calculation circuit 210 calculates the prediction coefficient value a'
H (1≦i≦P), and the impulse response h representing the transfer function of the weighted synthesis filter expressed by the following equation.
w(n) is calculated for a predetermined number of samples.

I-1w(Z)=W(Z)/(1-Σa'1Z-')
(5) Here, HW(Z) represents the transfer function of the weighted synthesis filter on Z transformation. Further, W(Z) is expressed in the above-mentioned equation (3), and the weighting is a two-transform expression of the function. The impulse response calculation circuit 210 has an impulse response hw
(n) is output to the autocorrelation number calculation circuit 360 and the cross correlation number calculation circuit 350.

Next, the autocorrelation number calculation circuit 360 inputs the impulse response hW (n) from the impulse response calculation circuit 210, and calculates the autocorrelation number Rhh(·) according to the following formula, and the autocorrelation number Rhh(τ) It is output to the pulse calculation circuit 390.

Next, the subtracter 285 inputs the audio signal x(n) accumulated in the buffer memory circuit 340, subtracts the output series of the synthesis filter circuit 400 by one frame sample from X(n), and subtracts the result e(n). ) is weighted and output to the circuit 410.

Next, the weighting circuit 410 inputs the subtraction result e(n) from the subtracter 285 and also inputs the prediction coefficient value a'1 from the parameter calculation circuit 200, and weights e(n). Output eW(n). Here, ew (n) can be written as a two-transform expression as follows.

EV(Z)=E (z) −w(z) ('t) Here, EW(Z) and mZ) represent the Z-transformed values of e and (n), and the Z-transformed value of e(n), respectively. Further, W (Z) represents the two-transformed value of the weighted function shown in equation (3) above. The weighting circuit 4.10 outputs e-trial, > to the cross-correlation calculation circuit 350.

Next, the cross-correlation calculation circuit 350 is weighted by the circuit 410.
eW(n) is input from the impulse response calculation circuit 210, and the impulse response hw(n) is input from the impulse response calculation circuit 210, and the cross-correlation number ψhx (n) is calculated for a predetermined number of samples according to the following equation.

ψhxGrl'-ΣeV(n)-hW (n-m)
FC1≦m≦N) (8) r+=1 The cross-correlation number ψhx(·) is output to the pulse calculation circuit 390.

Next, the pulse calculation circuit 390 calculates the cross-correlation number ψhx(・)
and the autocorrelation number hhh(・), the pitch period P'd is
The first pulse sequence, which is the basis of the driving sound source pulse sequence, is calculated by taking into account the following. Specifically, a pulse sequence calculation using the pitch period d and a pulse sequence calculation not using P'd are performed.

First, a pulse sequence calculation algorithm when P'd is not used will be described. As for the weighting of the input audio signal and the synthesized audio signal, a pulse sequence that minimizes the error power is sequentially calculated pulse by pulse according to the following equation.

−1 Here, yi indicates the amplitude of the i-th pulse within the frame. k indicates the total number of pulses generated within the frame, and town indicates the position within the frame of the i-th pulse.

In equation (9), the pulse position mi is determined from the position in the frame where the absolute value of c is the maximum value.

Next, a method of calculating a pulse sequence using P'd will be explained. The voiced portion of the audio signal has very strong periodicity, and the driving sound source pulse sequence is arranged periodically. Therefore, if you use the pitch period every time you calculate one sound source pulse and extrapolate the pulse to a position separated by the pitch period, you can increase the number of sound source pulses isometrically and greatly improve the characteristics. can do. FIG. 5 is a diagram showing an example of the process of creating a pulse sequence when the pitch period P'd is used.

The calculation of the pulse sequence follows the above equation (9). Figure 5(a)
indicates the cross-correlation number for one frame calculated by the cross-correlation number calculation circuit 350. Here the frame length is 160
It is used as a sample. FIG. 5(b) is a diagram showing the first pulse calculated according to equation (9).

FIG. 5(c) shows that the first pulse is removed using the pitch period P'd.

Figure 5(d) shows three pulses (g
,.

, L,',gu). FIG. 5(e) is a diagram showing the pulse I2. FIG. 5(f) is a diagram in which the pulse g2 is removed using the pinch period P'd. In this example, only two pulses, J71 and J12, are required to be transmitted to the decoder side for the pulse sequence created as described above. On the decoder side, by removing the pulses using the transmitted pitch period P'd,
This is because it can generate

Therefore, very good characteristics can be obtained even with a small number of pulses. This concludes the explanation of the pulse sequence calculation method.

Returning to FIG. 4 C8), the pulse calculation circuit 390 calculates whether or not the pitch period is used for the pulse sequence calculated using the pitch period and the pulse sequence calculated without using the pitch period. The error power between the input signal and the reproduced signal for each case is calculated according to the following equation.

J = PLee(o)−Σgiψhx(mi) QO
)-1 Here, Fi represents the pulse amplitude of equation (9), and ψhx(·) represents the number of cross-correlation. The weighted "ee" (o) indicates the power for N samples of the output value ew(n) of the circuit 410. The error power when using the pinch period is JN, and the error when using the pitch period is JN. Let JP be the electric power. JN and Jp are output to the comparison circuit 430. Also, the pulse sequence obtained using the pitch period and the pulse sequence obtained without using the pitch period are output to the switching circuit 440. Ru.

Next, the comparison circuit 430 compares the error power JN and Jp. If JP is smaller than JN, it is determined that the characteristics are better using the pitch period, and this information is output to switching circuit 440, comparison circuit 430, and gate circuit 460. Further, if Jp is shorter than JN, information indicating that the pitch period is used more often is output to switching circuit 440, comparison circuit 430, and gate circuit 460.

Next, the switching circuit 440 inputs the comparison information from the comparison circuit 430 and outputs one of the two types of pulse sequences to the encoding circuit 470 according to this information.

Next, the gate circuit 460 inputs the comparison information from the comparison circuit 430, and if it is better to use the pitch period,
The code ld is output as is to the multiplexer 450.

If it is still better not to use the pitch period, a code 1d representing pitch period 0 is output to the multiplexer 450.

Next, the encoding circuit 470 inputs the pulse sequence from the switching circuit 440 and encodes the nine amplitude positions of each pulse with a predetermined number of bits. Also, the amplitude 7 of each pulse
The decoded position value 1' i Hn'l'i is output to the pulse generation circuit 420 . Here, various methods of encoding the pulse sequence can be considered. One is to encode nine amplitude positions of the pulse train separately, and the other is to encode two amplitude positions together. An example of the former method will be explained.

First, as a method for encoding the amplitude of a pulse sequence, the maximum value of the amplitude of the pulse sequence within a frame is used as a normalization coefficient, and after normalizing the amplitude of each pulse, quantization and encoding are performed. There are ways to do this. Another possible method is to assume that the amplitude probability distribution is a normal type and use an optimal quantizer for the normal type. Regarding this, Mr. J-MAX (J-MAX)
IR, E TRANSACTIO
NS ON INFORMATION THgoay)
1 Quantizing for Minimum Distortion” (QU) published in March 1960 issue, pages 7-12.
ANTIZINGFou MINIMTJMJ)IST
Since it is explained in detail in the paper entitled "O 几Tl0N" (Reference 3), etc., the explanation is omitted here.Furthermore, after converting the amplitude of each pulse to other parameters in an orthogonal relationship, quantization, It may be encoded.Also, the bit assignment may be changed for each pulse amplitude.Next, various methods can be considered for encoding the pulse position.For example, in the field of facsimile signal encoding, A well-known run-length code etc. may be used. This is the code "0" or "
■The length of “” is expressed using a predetermined code sequence. Also, to encode the normalization coefficient,
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 to FIG. 4(a), the pulse generation circuit 420 generates the pulse sequence decoded value +j? 'i Hrn' rl using i
A pulse train having an amplitude y'1 is generated at the position l'i. At this time, when using the pitch period based on the information input from the comparison circuit 430, the pitch encoding circuit 38
Using the pitch period decoded value p/d inputted from 0, the pulse is removed to a position separated by the pitch period p/d from the pulse sequence decoded value inputted from the encoding circuit 470. The driving sound source pulse sequence thus obtained is output to the synthesis filter circuit 400. The synthesis filter circuit 400 is
A driving sound source pulse sequence is input from the pulse generation circuit 420. Furthermore, a prediction coefficient value a'1 is input from the K-parameter encoding circuit 200 to configure a synthesis filter. The synthesis filter circuit 400 performs a filter operation using the input driving sound source pulse, calculates a response signal for one frame, and outputs it to the subtracter 285. Here of the response signal x(n) ◇(
, ) are calculated for two frames (1≦n≦2N).

d(n) represents a driving sound source signal, and when 1≦n≦N, the driving sound source pulse output from the pulse generation circuit 420 is used. Still, when N+1≦n≦2N, all O sequences are used.

A(n) in the second frame of A(n) determined by formula 09
) (N+1≦n≦2N) is output to the subtracter 285.

Next, the multiplexer 450 inputs the output code of the parameter encoding circuit 200 and the output code of the gate circuit 460 to the output code of the encoding circuit 470, combines them, and outputs them from the transmission side output terminal 480 to the communication path. . This concludes the section (1) and (1) on the encoder side of the audio encoding system according to the present invention.

Next, the decoder side of the audio encoding system according to the present invention will be explained with reference to FIG. 4(b). Demultiplexer 50
0 inputs the code from the decoder side input terminal 490. Of the input codes, the demultiplexer 500 separates a code sequence representing the K parameter and a code sequence representing pitch information,
A parameter decoding circuit 520 separates the code sequence representing the first pulse sequence and converts the code sequence representing the K parameter into a parameter decoding circuit 520.
The pulse sequence decoding circuit 530 outputs a code sequence representing the first pulse sequence to the pulse sequence decoding circuit 530. K-parameter decoding circuit 520 and pitch decoding circuit 510 decode the input code sequence and output it to synthesis filter circuit 550.

The pulse sequence decoding circuit 530 inputs the code sequence representing the first pulse sequence, decodes it, and outputs it to the pulse generation circuit 540 as amplitude two-position information of the pulse sequence. The pulse generation gate 54o receives the amplitude and two position information of the first pulse sequence and generates a driving sound source pulse sequence corresponding to the second pulse sequence.

At this time, the pitch periodic value P is sent from the pitch decoding circuit 510.
Id is input, and if this value is not 0, the pulse is removed to a position separated by P'd with respect to the input pulse sequence. The drive sound source pulse sequence thus obtained is output to the synthesis filter circuit 550. Synthesis filter circuit 550
inputs 'i as a parameter decoded value from the K parameter decoding circuit 520, generates a composite signal M(n) using the output pulse train of the pulse generation circuit 540 as a driving source, and outputs it from the receiving side output terminal 560. This completes the explanation of the decoder side according to the present invention.

According to the configuration of this embodiment, the error power is calculated on the encoder side between when a pulse sequence is determined using the pitch period and when the pulse sequence is determined without using the pitch period, and the error power is reduced. Since the configuration is such that a pulse sequence that can reproduce the input audio more faithfully is transmitted and used for reproduction on the decoder side, there is a possibility that errors in extraction of transient parts of the input audio signal or pitch parameters may occur. This has the effect of preventing deterioration.

Incidentally, as a more convenient method for determining whether to use the pitch period or not, it is also possible to use the pitch gain. Here, the pitch gain is determined from the value of the correlation coefficient with a delay equal to the pitch period. The pitch gain obtained in this way is compared with a predetermined threshold, and if the pitch gain is less than or equal to the threshold, the pitch period is not used when calculating the sound source pulse. Further, the pitch period transmitted to the decoder side is set to 0. By adopting such a configuration, the error power calculation and comparison circuit 430 of Equation 101 becomes unnecessary, and the amount of calculation can be reduced.

In the pulse calculation method shown in formula 00), pulses were calculated one by one in sequence. In this method, when calculating the next pulse, the amplitudes of a plurality of pulses accumulated in the past may be readjusted. This improves the characteristics when the pulses are not independent of each other.

Further, as a method for calculating the sound source pulses, other good pulse sequence calculation methods, such as a method for calculating an optimal pulse sequence, can be used.

In addition, in the configuration of this embodiment, if it is determined that the characteristics are better when the pitch period is used, removal processing is performed on all requested pulse sequences. This removing process does not necessarily need to be applied to all pulse sequences; it is also possible to select a specific pulse that has a large pulse removing effect and apply the removing process only to this selected pulse. . Various methods can be considered for selecting a specific pulse. For example, since it is considered that pulses with large amplitudes have stronger periodicity, processing may be performed to remove only M pulses with large amplitudes out of the pulses o that are included in a frame. Here M
The value may be a predetermined value or may be changed for each frame. Further, δ information (1 bit) may be added to each pulse in order to determine whether or not to perform removal processing. With such a configuration, although the amount of transmitted information increases slightly, the characteristics are significantly improved.

In the configuration of this embodiment, when calculating the autocorrelation number of an impulse response sequence representing a short-time spectral structure, the impulse response calculation circuit 210 calculates the impulse response using the decoded parameter values, and then calculates the impulse response using the decoded parameter values. An autocorrelation number calculation circuit 360 calculates an autocorrelation number using the response. As is well known in the field of digital signal processing, the autocorrelation coefficients of an impulse response have a corresponding relationship with the power spectrum. Therefore, a configuration may be adopted in which the power spectrum is first determined using the decoded parameter values, and then the autocorrelation number is calculated using this correspondence. On the other hand, when calculating the cross-correlation number between the audio signal and the impulse response representing the short-time spectral envelope, in the configuration of this embodiment, the weight is determined by the output value eW(n) of the circuit 410 and the parameter decoded value. The impulse response hw(n) calculated by the impulse response calculation circuit 210 using 1 is used to calculate the cross-correlation number ψhx(·).

As is well known, the cross-correlation number corresponds to the cross-power spectrum. Therefore, the configuration may be such that the cross power spectrum is first calculated using ew(n) and '1, and then the cross-correlation coefficients are calculated. Regarding the correspondence between the power spectrum and the autocorrelation number, and the correspondence between the cross power spectrum and the cross-correlation number, see N. V. Oppenheim (A-V-O
“Digital Signal Processing” by Mr. PPENI (ETM) et al.
Since it is explained in detail in Chapter 8 of the book entitled "NG") (Reference 4), the explanation will be omitted here.

Further, according to the present invention, there is a great effect that there is almost no deterioration of the reproduced signal near the frame boundary due to waveform discontinuity at the frame boundary.

The effect of this is that on the encoder side, when calculating the pulse sequence of the current frame, the response signal sequence obtained by driving the synthesis filter using the driving excitation pulse sequence l frames past is extended to the current frame. This is due to the fact that the frame pulse sequence is calculated by looking at the result of subtracting this from the input audio signal sequence.

Further, in this embodiment, a case where the frame length is constant has been described, but a variable length frame may be used in which the frame length is changed over time. Further, the number of sound source pulses generated within one frame does not need to be constant.

For example, the number of pulse sequences in each frame may be changed so as to keep the S/N constant.

In this embodiment, as a criterion for selecting a pulse sequence that can reproduce the input signal more faithfully between a pulse sequence obtained using the pitch period and a pulse sequence obtained without using the pitch period, ( The error power shown in equation 10) was used. This can be done using other best methods. For example, a configuration may be adopted in which a predicted gain when using the pitch is calculated from the pitch gain, and this value is compared with a predetermined threshold value.

Furthermore, in the embodiment of the present invention described above, the encoding of the pulse sequence within one frame is performed by the encoding circuit 470 of FIG. 4(a) after all the pulse sequences have been completed. 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 created, so that the quality can be further improved.

In the example described above, the parameter was used as the parameter representing the spectral envelope of the short-time audio signal sequence, but this can be done using other well-known parameters (such as the LSF parameter). Good too.

Furthermore, the above-mentioned equation (5), (in the method, the weighting is the function W
(Z) may be omitted.

In addition, in this embodiment, in order to prevent quality deterioration due to discontinuity of the reproduced waveform at frame boundaries, a response signal sequence derived from the drive sound source pulse one frame past the current frame is calculated, and The pulse sequence was calculated after subtracting this response signal from the voice, but as shown in Figure 6, the data used for calculating the pulse sequence was
Pulse NT indicates the frame in which the pulse is transmitted, and N indicates the frame in which the sound source pulse is calculated. With such a configuration, there is an effect that there is no need to calculate a response signal sequence derived from a driving sound source pulse one frame past.

(Effects of the Invention) As explained in detail above, according to the present invention, when calculating the pulse sequence, the pinch period is used to remove pulses and increase the number of pulses, so the transmission bit rate can be reduced. It has the effect of being able to obtain good reproduced audio even when the number of sound source pulses is low (the number of sound source pulses is small).In particular, it is possible to obtain a high-pitched female voice with a high pitch frequency, where the male voice quality deteriorated in the conventional method. Good reproduced audio can be obtained with the amount of information transmitted.Also, the sound source pulse (
10), so it is determined according to the formula 1. Unlike the conventional method, there is no path to drive a synthesis filter with a sound source pulse to obtain a reproduced signal, and then adjust the pulse by feeding back the squared error with the original signal, and there is no need to repeat the process nine times. Therefore, there is an effect that the amount of calculation can be significantly reduced. Furthermore, the method of removing pulses using the pitch period has the advantage that it can be realized with a small amount of additional calculation.

[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, Figure 3 is a diagram showing the frequency characteristics of the input audio signal sequence, and the weighting shown in Figure 1 is the frequency of the circuit. A diagram showing an example of the characteristics, a diagram showing an example of the search process for the sound source pulse of the fourth combination, and FIG. 6 are diagrams for explaining the positional relationship between the pulse transmission frame and the sound source pulse calculation frame. In the figure, 110, 340...buffer memory circuit, 120, 285...subtraction circuit, 130, 40
0, 550...Synthesis filter circuit, 14.0, 4
20, 540... Pulse generation circuit, 150... Error minimization circuit, 180, 280... K parameter calculation circuit, 190, 410... Weighting circuit, 200
. . . K parameter encoding circuit, 210 . . . Impulse response division circuit, 350 . . . Cross-correlation calculation circuit, 36
0... Autocorrelation calculation circuit, 370... Pitch analysis circuit, 380... Pitch encoding circuit, 390... Pulse calculation circuit, 430... Comparison circuit, 440... Switching circuit, 470 ... Encoding circuit, 450... Multiplexer, 460... Gate circuit, 500...
Demultiplexer, 510... Pitch decoding circuit, 52
0: K parameter decoding circuit; 530: sound source pulse decoding circuit. , λ;:8. ．． l 'F Takushi Naiba tf, Xi Figure 1 Figure 2 Figure 5 Figure 16 Procedural Amendment (Voluntary) 60.5.29 Showa Year Month Day 1, Case Indication 1982 Patent Application No. 042305
No. 2, Title of the invention Audio encoding system and its device 3, Relationship with the amended person's case Applicant 5-33-1 Shiba, Minato-ku, Tokyo (423) NEC Corporation Representative Tadahiro Sekimoto 4; Agent 7″″-＼・te11. ml, :,,,su(-1 5. Subject of amendment (1) Claims column of the specification (2) Detailed explanation of the invention column 6 of the specification Contents of the amendment (1) Claims Amend the range as shown in the attached sheet. (2) Amend "less" on page 16, line 16 of the specification to "less". (3) Amend page 23, line 8 of the specification. The phrase "subtracted" should be corrected to "subtracted from the cross-correlation numbers in Figure 5 (a)." (4) "Transmitted" on page 23, line 14 of the specification. (5) In the 7th line of page 24 of the specification, the statement "gl is" is corrected to "gi is when the pitch period is not used." (6) In the 7th line of page 24 of the specification, ``ψhx' is corrected to ``If pitch period is used, all pulse amplitudes including such pulses, ψhx.'' (7) Specification No. The phrase "by doing so" on page 31, line 19 is corrected to "by doing so, the number of pulses in the frame is increased." (8) On page 37, line 17 of the specification, "(11・−＼、Patent Attorney Uchihara !−ﾛAttachment Claims (1) On the transmitting side, the discrete audio signal sequence is A pitch parameter representing a fine structure and a spectral parameter representing a short-time spectral envelope are extracted and encoded, and based on the spectral parameters, an autocorrelation number of an impulse response sequence corresponding to the short-time spectral envelope is calculated; A first pulse sequence that can satisfactorily represent the audio signal sequence is encoded using the pitch parameter and the autocorrelation coefficient, and the code representing the pitch parameter and the spectral parameter is combined and output.
On the receiving side, the one combined code is decoded by separating the code representing the first pulse sequence, the code representing the pitch parameter, and the code representing the spectral parameter, and the decoded first A second pulse sequence is determined based on the pulse sequence using the decoded pitch parameter, and the audio signal sequence is reproduced using the second pulse sequence and the decoded spectral parameter. Characteristic voice encoding method. (2) a parameter calculation circuit that inputs a discrete audio signal sequence and extracts and encodes a pitch parameter representing a pitch fine structure and a spectral parameter representing a short-time spectral envelope from the audio signal sequence; an autocorrelation coefficient calculation circuit that receives an output sequence and calculates a number of autocorrelations of an impulse response sequence according to a short-time spectrum of the audio signal sequence; and inputs the audio signal sequence and the output sequence of the parameter calculation circuit. a mutual correlation calculation circuit for calculating a cross-correlation number according to the speech signal sequence and an impulse response sequence corresponding to the short-time spectrum; and an output series of the autocorrelation calculation circuit and the cross-correlation calculation circuit. a first pulse calculation circuit that receives the output sequence of the numerical calculation circuit and the output sequence of the parameter calculation circuit and encodes a first pulse sequence that can satisfactorily represent the audio signal; A speech encoding device comprising: a multiplexer circuit that combines and outputs the output code and the output sequence of the first pulse sequence calculation circuit. (3) The first one that can satisfactorily represent the input audio signal sequence.
input a code sequence in which a code representing a pulse sequence of , a code representing a pitch parameter representing a pitch fine structure of the audio signal sequence, and a code representing a spectral parameter representing a short-time spectral envelope of the audio signal sequence are combined. a demultiplexer circuit that separates a code representing the first pulse sequence, a code representing the pitch parameter, and a code representing the spectrum parameter;
a first pulse sequence decoding circuit that receives and decodes a code representing the first pulse sequence obtained by separation; and a pitch parameter that inputs and decodes a code representing the pitch parameter obtained by separation. a decoding circuit, a spectral parameter decoding circuit inputting and decoding a code representing the separated spectral parameter, an output series of the first pulse sequence decoding circuit and an output series of the pitch parameter decoding circuit; a second pulse sequence generation circuit that generates a second pulse sequence using the decoded pitch parameter based on the input and decoded first pulse sequence; An audio decoding device comprising: a synthesis filter circuit that inputs an output sequence and an output sequence of the spectral parameter decoding circuit, reproduces and outputs an audio signal sequence. 17N5 Agent Patent Attorney Uchihara -(゛＼Nino

Claims (3)

[Claims] (1) On the transmitting side, a pitch parameter representing a pitch fine structure and a spectral parameter representing a short-time spectral envelope are extracted and encoded from a discrete audio signal sequence, and the short-time spectral envelope is encoded based on the spectral parameters. calculate the autocorrelation number of the impulse response sequence according to the speech signal sequence and the impulse response sequence, calculate the cross-correlation number according to the speech signal sequence and the impulse response sequence, and calculate the first pitch parameter that can satisfactorily represent the negative signal sequence. The pulse sequence is
If it is less than a predetermined threshold, the first pulse code sequence is determined using only the autocorrelation function and the cross-correlation function, the first pulse sequence is encoded, and the first pulse sequence is A code representing the first pulse sequence, a code representing the pitch parameter, and a code representing the spectral parameter are combined and output, and on the receiving side, the combined code is combined with a code representing the first pulse sequence, a code representing the pitch parameter, and a code representing the spectral parameter. A second pulse sequence is generated using the decoded pitch parameter based on the decoded first pulse sequence. An audio encoding method characterized in that the audio signal sequence is reproduced using the second pulse sequence and the decoded spectrum parameter. (2) a parameter calculation circuit that receives a complete input of a discrete audio signal sequence and extracts and encodes a pitch parameter representing a pitch fine structure and a spectral variation representing a short-time spectrum envelope from the audio signal sequence; and the parameter calculation circuit. a cross-correlation calculation circuit that receives all the output series of the input pulse response sequence and calculates the auto-correlation number of the impulse response sequence corresponding to the short-time spectrum of the audio signal sequence; , when the output series of the autocorrelation coefficient calculation circuit, the output series of the cross correlation coefficient calculation circuit, and the output series of the parameter calculation circuit are input, and the pitch parameter exceeds a predetermined threshold value, A first pulse sequence that can represent the audio signal well is determined and encoded from the pitch parameter, the autocorrelation function, and the cross-correlation function, and when the pitch parameter is less than or equal to the aperture, the autocorrelation function and the cross-correlation function are encoded. a first pulse calculation circuit that encodes the first pulse sequence from the cross-correlation function; and outputs a combination of the output code of the parameter calculation circuit and the output sequence of the first pulse sequence calculation circuit. 1′4″1′”° characterized by having a multiplexer circuit that (3) The first step is to obtain a good representation of the input audio signal sequence.
a code representing a pulse sequence of the audio signal sequence, a code representing a pitch parameter representing the pitch fine structure of the audio signal sequence, and a code representing a spectral parameter representing the short-time spectral envelope of the audio signal sequence. a demultiplexer circuit that separates a code representing the first pulse sequence, a code representing the pitch parameter, and a code representing the spectral parameter;
a first pulse sequence decoding circuit that inputs and decodes a code representing the first pulse sequence obtained by separation; a pulse sequence decoder that inputs and decodes a code representing all of the pitch parameters obtained by separation; a spectral parameter decoding circuit that inputs and decodes codes representing all of the separated spectral parameters;
The output sequence of the pulse sequence decoding circuit and the output sequence of the pitch parameter decoding circuit are input, and a second pulse sequence is calculated using the decoded pitch parameter based on the decoded first pulse sequence. a second pulse sequence generation circuit, and an output sequence of the second pulse sequence generation circuit and the spectral parameter recovery circuit;