JP3006790B2 - Voice encoding / decoding method and apparatus - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a speech encoding / decoding method and apparatus for encoding and decoding an audio signal efficiently at a low bit rate.

(Prior Art) As a method of transmitting an audio signal at a low bit rate, for example, about 16 Kb / s or less, a multi-pulse encoding method and the like are known. In these, the sound source signal is represented by a combination of a plurality of pulses (multi-pulse), the characteristics of the vocal tract are represented by a digital filter, and the information of the sound source pulse and the coefficient of the filter are represented by:
It is obtained and transmitted for each fixed time section (frame). For details of this method, see, for example, "Multipulse Excited Speechcoder Bastion on Maximum Cross-Correlation Search Algorithm" by Araceki, Ozawa, Ono, and Ochiai (I.
EE Global Telecommunication '83 Lecture number 23.3, Reference 1) (Araseki, Ozawa, Ono)
and Cchial, `` Multi-pulse Excited Speech Coder Bas
ed on Maximum Cross-correlation Search Algorith
m ", (GLOBECOM 83, IEEE Global Telecommunication, 23.
3,1983)). In this method, a good voice signal is output after decoding by separating and expressing the vocal tract information and the sound source signal, respectively, and using a combination (multi-pulse) of a plurality of pulse trains as means for expressing the sound source signal. . Reference 1 by performing pitch prediction using a pitch parameter representing a fine structure of pitch
Since the pitch prediction multi-pulse method for improving the sound quality of the above method is described in Japanese Patent Application No. 58-139022 (Reference 2), it will be briefly described here with reference to FIG.

An audio signal of a section divided for each frame is input from an input terminal 800 in the figure. The spectrum parameter calculation unit 810 determines the spectrum parameter of the audio signal by a known method. The pitch parameter calculation unit 820 obtains a pitch parameter representing a fine structure of the pitch by a known method such as an autocorrelation method. These outputs are input to a pitch prediction excitation pulse calculation unit in addition to being transmitted as they are. The pitch prediction excitation pulse calculation unit 830 predicts the pitch and obtains and outputs a multi-pulse train representing the excitation signal. However, since this method is described in the above-mentioned document 2, its description is omitted here.

(Problems to be Solved by the Invention) However, in the conventional method of Document 1, the sound quality is good when the bit rate is sufficiently high and the number of sound source pulses is sufficient, but as the bit rate is lowered, the sound quality is improved. Had declined. In particular, the conventional method has a drawback that, in the case of an input signal having a high pitch frequency, for example, when a female voice is input, the reproduced sound is deteriorated. This means that when the pitch frequency is high, many pitch waveforms are included in the frame of the pulse calculation, and in order to reproduce this pitch waveform well, compared to a speaker with a low pitch frequency, , Because more multi-pulses are required. Therefore, for this reason, the transmission bit rate is significantly reduced without lowering the sound quality.
That is, it has been difficult to significantly reduce the number of pulses in one frame.

On the other hand, in the conventional method of Document 2, although pitch prediction is performed using a pitch parameter based on a correlation for each pitch, a large-amplitude excitation signal or a small-amplitude excitation signal is used.
The sound source signal is represented using multi-pulses and pitch prediction. However, a large-amplitude excitation signal is considered to have a high correlation for each pitch, but a small-amplitude excitation signal is considered to have a relatively low correlation. Therefore, in order to further improve the sound quality, the role of the small-amplitude multi-pulse train or the small-amplitude sound source signal among the multi-pulse trains representing the sound source signal is important. This is especially important for consonant audio signals. In the conventional method, as a multi-pulse train expressing a sound source signal, only a set number is obtained and transmitted in descending order of amplitude. Therefore, in the conventional method, a sufficient number of small-amplitude pulses cannot be obtained due to the preset upper limit of the information amount, and the approximation degree of the sound source signal is not sufficient.
There was a limit in the quality of reproduced sound. This was particularly remarkable when the bit rate was low.

An object of the present invention is to provide an audio encoding / decoding system and an apparatus therefor that are capable of reproducing sound better than before even when the bit rate is high or at a low bit rate with a relatively small amount of computation. Is to do.

(Means for Solving the Problems) In a speech encoding / decoding method according to the present invention, a discrete speech signal is input, and a pitch parameter representing a pitch structure of the speech signal and a spectrum parameter representing a spectrum envelope of the speech signal. Is determined, the audio signal is classified into a predetermined type, and a sound source signal of the audio signal is multi-pulse train obtained by pitch prediction according to the type, or a code book is used in addition to the multi-pulse train. It is determined whether or not the performance is improved, and if the performance is improved, the pitch parameter, the spectrum parameter, the multipulse, and the code are represented by using both a codebook and the multipulse train to represent the excitation signal. Transmitting the index of the book and the type, the codebook and the multi-pulse train according to the type And outputting a synthesized speech signal representing the speech signal using the pitch parameter and the pitch parameter.

A speech coding apparatus according to the present invention includes a pitch parameter calculation circuit that obtains and encodes a pitch parameter representing a pitch structure from an input discrete speech signal sequence, and a spectrum parameter that represents a short-time spectrum characteristic of the speech signal. A spectrum parameter calculation circuit for encoding the audio signal, a discrimination circuit for classifying the audio signal into a predetermined type, and a sound source signal of the audio signal using the pitch parameter and the spectrum parameter according to an output of the discrimination circuit. Use a multi-pulse train obtained by pitch prediction, or determine whether the performance is improved by using a codebook in addition to the multi-pulse train, and if the performance is improved, use both the codebook and the multi-pulse train. An excitation signal calculation circuit for expressing and encoding the pitch parameter; And a multiplexer circuit for combining an output for outputting the index and the discrimination circuit of code and the codebook represents the sign and the multi-pulse train representing the spectrum parameter.

An audio decoding device according to the present invention is an audio decoding device that receives an output signal of the audio encoding device via a transmission path and restores the audio signal, and includes a code representing a sound source multi-pulse sequence representing a sound source signal. A code representing a pitch parameter representing a fine structure of a pitch of a speech signal, a code representing a spectrum parameter representing a short-time spectrum characteristic of the speech signal, a code representing a sound source signal, a discrimination code, and an output code of a sound source discrimination circuit are inputted. A demultiplexer circuit for separating and decoding, and a sound source signal obtained by restoring a multi-pulse train according to an output code of a sound source discriminating circuit in the decoded sound source device and reproducing a pitch by using the decoded pitch parameter. Depending on the desired pitch reproduction circuit, the discrimination code and the output code of the sound source discrimination circuit, one of a plurality of codebooks is selected or A sound source signal restoring circuit for restoring a driving sound source signal using an output of the pitch reproducing circuit by using a pulse; and a sound using the restored driving sound source signal or the reproduced sound source signal and the decoded spectrum parameter. And a spectrum envelope filter circuit for synthesizing signals.

(Function) In the present invention, in the pitch prediction multi-pulse coding method of the document 2, in order to express a sound source signal more efficiently than a conventional method with a small amount of transmission information, an audio signal for each frame is predetermined. The sound source signal for representing the audio signal is classified into a vowel part, a consonant part or a voiced part, and an unvoiced part. In the case of using the pulse method and the codebook, and further using the codebook, the effect of the characteristics due to the use of the codebook is obtained by once reproducing the signal by the codebook for each subframe shorter than the frame and comparing the signal with the input signal. And if the degree of improvement in the characteristics is high,
The codebook is used in the subframe. On the other hand, when the degree of improvement is low, the use of a codebook in the subframe is stopped and the number of pitch prediction multipulses is increased using only pitch prediction multipulses.

Here, vowels and consonants are discriminated using methods known in the art, for example, using parameters such as frame power, changes in the spectrum between the current frame and the previous frame under the power between the current frame and the previous frame, and pitch characteristics. Can be.
On the other hand, parameters such as pitch characteristics can be used for discriminating voiced and unvoiced portions.

In addition, for determining the effect of the characteristic by the selected codebook, for example, the ratio of the power or RMS of the reproduced signal reproduced by driving the synthesis filter using the codebook to the power or RMS of the input signal, or Depending on the method, a method of determining the magnitude of the codebook gain can be used.

The operation of the present invention will be described with reference to FIG. An audio signal x (n) of a section divided for each frame is input from an input terminal 300 in the figure. Spectrum parameter calculator 31
0 calculates the spectral parameters from the audio signal of the current frame. The pitch parameter calculator 320 calculates a pitch parameter representing a fine structure of the pitch. Voiced / unvoiced discrimination section 305 determines whether the current frame is voiced or unvoiced using the pitch gain, and outputs a voiced / unvoiced discrimination code. The pitch prediction excitation pulse calculation section 330 obtains a predetermined number of excitation pulses using pitch prediction according to the discrimination signal. For a method of obtaining a multi-pulse using pitch prediction, reference can be made to Reference 2. Switch 340
When the current frame is a voiced section (or vowel section), both are connected to the lower side to output a multi-pulse. When the current frame is a voiceless section (or consonant section), the switches 340 are connected to the upper side first, and the obtained multipulse is input to the signal reproducing section 350. The signal reproducing unit 350 reproduces the synthesized speech signal x (n) by driving the pitch reproduction filter and the synthesis filter using the input sound source pulse train, pitch parameter, and spectrum parameter. The subtractor 360 subtracts the synthesized speech waveform x (n) from the input signal x (n). And this residual signal e
(N) is input to the codebook selection unit 370. The codebook selecting unit 370 calculates a small-amplitude sound source signal representing the characteristic of the small-amplitude component of the sound source signal for each of the small sections (for example, about 5 msec.) Obtained by dividing the frame section into several sections. Here, this small-amplitude sound source signal has almost random phase characteristics and is considered to be almost close to a noise signal.
In order to encode such a signal very efficiently,
A vector quantization technique of preparing and encoding a codebook (codebook) of a plurality of small amplitude excitation signals created in advance can be used. Regarding vector quantization, for example, “Vector Quantization for Speech Coding and Recognition” by Earl M. Gray (Journal of the Acoustical Society of America 80, Q1,198)
6, Reference 3) (RMGray, “Vector quantization for spe
ech coding and recognition "(J.Acoust.Soc.America,
vol.80, Suppl.1, Q1, 1986)).

Hereinafter, the operation of the codebook selection unit 370 will be described.
The reproduced signal x (n) is converted to the original audio waveform x (n) by the subtractor 360.
, The residual signal e (n) resulting from the subtraction is uniformly divided in time into small sections shorter than the frame, and one of the codebooks (codebooks) is input, and the gain is passed through a gain circuit. After the combination, the combined residual signal e (n) is combined.

Then, from the input residual signal e (n), the combined residual signal e
(N) is subtracted, and the result is weighted to obtain a weighted error power. Then, an optimal code is selected from a code book so as to minimize the weighted error power, and the index and gain are output.

Next, an actual method for expressing a small-amplitude sound source signal using a codebook and selecting the codebook will be described below using equations. As a codebook selection method, calculation is performed so as to minimize the error power E defined by the following equation.

Here, e (n) is the codebook selection unit 370 in FIG.
Is the input residual signal, g is the gain, and e (n) is the residual signal reproduced by the selected one type of code and the synthesis filter. w (n) indicates the impulse response of the weighting filter considering the sense of hearing. When the expression (1) is minimized with respect to g, the expression (2) is obtained.

Here, e _w (n) = e (n) * w (n) = n (n) * h (n) *
w (n) (3a) e _w (n) = e (n) * w (n) (3b) The symbol * represents convolution. (2) the denominator autocorrelation e _w (n) (strictly covariance), molecules and e _w (n) e
_w (n) is the cross-correlation. Also, n (n) in the equation (3a)
Is a signal represented by one selected code in the record book. H (n) indicates the impulse response of the synthesis filter.

At this time, the error power E can be written as the following equation. The codebook that minimizes E may be selected so as to maximize the second term of equation (4), that is, maximize | g |.

As a method for further greatly reducing the amount of calculation for selecting a codebook, the following configuration can be considered. The multipulse train representing the sound source signal is searched for using the cross-correlation. Since the method of obtaining is equal to the above-mentioned documents 1, 2 and the like, the description is omitted here.However, in the pitch prediction excitation pulse calculation unit 330, by using the modified cross-correlation function Φ _xh 'after _obtaining the pitch prediction multi-pulse train, In addition, the codebook can be selected after significantly reducing the calculation amount compared to the above-described method. In the method described below, the signal e _w (n) does not need to be reproduced at the time of codebook selection, so that the amount of calculation can be greatly reduced while keeping the characteristics almost the same as in the above-described method. The derivation method will be described below. First,
Φ _xh ′, e _w (n) can be written as:

Φ _xh '= Σe _w (n) h _w (n) (5) e _w (n) = n (n) * h _w (n) (6) Substituting equation (6) into equation (2), Using equation 5),
The following modifications are possible.

Here, Φ _xh ′ is a cross-correlation function after _{obtaining a} multi-pulse train by pitch prediction, and R _hh (0) is power of an impulse response of a filter composed of a cascade connection of a spectrum envelope filter and a weighting circuit. R _nn (0) is the power of the signal n (n) represented by the code when one type of code is selected from the code book. (7)
The numerator of the equation is the cross-correlation function between Φ _xh 'and the signal n (n) represented by the selected code. As in the case of the above equation (2), the codebook may be selected so as to maximize g in the equation (7).

Note that the code book may be created by training in advance using residual signals of sound sources after obtaining a predetermined number of large-amplitude pitch prediction multi-pulse trains. A plurality of noise signals having characteristic characteristics may be created by changing the phase characteristics variously and stored in the codebook. The latter method is described by M. A. Schroeder and BS Atar in "Code Excited Linear Prediction (CELP): High-Quantity Speech.
At Very Low Bit Rates ”(I.C.S.S.S.P. Lectures, Volume 1, Lecture No. 25.1.1, 1985 Reference 4) (MRShroeder and BSAta
l: “Code-Excited linear prediction (CELP): high-
quality speech at very low bitrates, ", Proc, ICA
SSPvol.1, paper No. 25.1.1, March, 1985).

Next, the switch 385 is connected to the upper side when selecting a codebook, and outputs the codebook and the gain to the small amplitude sound source reproducer 372. The small-amplitude sound source reproducer 372 reproduces a small-amplitude sound source using the code and the gain selected by the codebook selection unit 370.

 Hereinafter, the determination unit 380 will be described.

The signal e (n) reproduced by using the code book and the output signal e (n) of the subtractor 360 are temporally divided by the time division circuit into the same length as the small section reproduced signal, and these are divided in divided units. By comparing the two signals, it is determined whether or not there is an effect of improving the characteristics by representing the excitation signal of the current subframe using a codebook. As described above, as described above, for example, the power ratio between e (n) and e (n), the RMS ratio, and the like can be used.

As a result of the determination, if it is determined that the codebook is necessary for the current subframe, the code is output, and the codebook selection unit 370, the pitch prediction sound current pulse calculation unit 330, and the switch 340 are output.
Output a signal to With this signal, the switches 340 are both connected to the lower side, and the pitch prediction excitation pulse calculation unit 330 outputs the obtained excitation pulse train, and the codebook selection unit
At 370, the index and gain are output. When it is determined that the codebook is unnecessary for the current subframe, the code is output to the codebook selection unit 370, the pitch prediction excitation pulse calculation unit 330, and the switch 340 without using the codebook. Output a signal to modify the signal. With this signal, the switches 340 are both connected to the lower side, and the pitch prediction sound source pulse calculation unit 330 increases the sound source pulse train by the determined number and outputs it in addition to the multi-pulse train obtained so far. Selection section 370 does not output the index and gain of the codebook for the current subframe.

The transmission information on the transmitting side is a voiced / unvoiced discrimination code, a pitch parameter, a spectrum parameter, an amplitude and a position of a multi-pulse whose pitch is predicted, a discrimination code output from the discrimination unit, and an index and a gain of a codebook to be used.

(Examples) Next, the present invention will be described in more detail by way of examples.

FIG. 1 is a block diagram showing an example of a speech coding apparatus and a speech decoding apparatus according to the present invention. Referring to FIG. Will be described.

1, a discrete audio signal x (n) is input from an input device 500. The time division circuit 510 converts the input audio signal into temporally uniform frames (for example, 20 ms).
ec.). The pitch parameter calculation circuit 515 calculates a pitch parameter representing the pitch structure. As a calculation method, a method as shown in the aforementioned reference 2 is used. The quantizer 516 quantizes the obtained pitch parameter. The inverse quantizer 518 performs inverse quantization using the result of quantization and outputs the result. The spectrum parameter calculation circuit 520 obtains a spectrum parameter representing the spectrum of the audio signal in the divided section by a known LPC analysis method.

The obtained spectral parameters are quantized by a spectral parameter quantizer 525. As a quantization method, slaker quantization as shown in Japanese Patent Application No. 59-272435 (Reference 5) or vector quantization as described in Reference 4 may be performed. Inverse quantizer 53
“0” is inversely quantized using the result of quantization and output.
The weighting circuit 540 weights the divided audio signal using the dequantized spectral parameters. The weighting method is described in the weighting circuit 200 of
Can be referred to. Impulse response calculation circuit 550
Calculates the impulse response using the inversely quantized pitch parameter and the inversely quantized spectral parameter. The specific method can be referred to the above-mentioned document 2. The autocorrelation function calculation circuit 560 calculates the autocorrelation function of the impulse response and outputs the calculated autocorrelation function to the sound source pulse calculation circuit 580. The calculation method of the autocorrelation function is based on the autocorrelation function calculation circuit 180 of the aforementioned reference 2.
Can be referred to. The cross-correlation function calculation circuit 570 calculates a cross-correlation function between the weighted signal and the impulse response, and outputs the result to the sound source pulse calculation circuit 580. The specific method can be referred to the above-mentioned document 2.

The determination circuit 575 determines whether the current frame is, for example, a vowel section or a consonant section, and outputs a determination code indicating the result to the excitation pulse calculation circuit 580. For the discrimination, as shown in the section of the operation, for example, well-known parameters such as spectrum change, power, and power change can be used.

When the output of the discrimination circuit 575 is a code indicating a vowel, the sound source pulse calculation circuit 580 obtains a predetermined number (L1) of multi-pulses by pitch prediction. For the calculation method of the multi-pulse train, reference can be made to the sound source pulse calculation circuit 210 of Reference 2.

If the output of the discrimination circuit 575 is a code indicating a vowel, the small amplitude excitation signal is not calculated, and the calculation for the excitation signal ends here. Therefore, in this case, the quantizer 585, the pulse generator 600, the pitch reproduction filter 605, the synthesis filter 610, the subtractor 615, and the small amplitude sound source calculation circuit 620 do not operate.

The quantizer 585 quantizes the source multipulse train and outputs a code. This output is inversely quantized by an inverse quantizer 590, and a multi-pulse is generated by a pulse generator 600. The pitch reproduction filter 605 receives the reproduced multi-pulse and the pitch parameter dequantized by the dequantizer 518 as inputs and outputs a sound source signal whose pitch has been reproduced. By passing the source signal and the spectral parameters output from the inverse quantizer 530 through a synthesis filter 610, a synthesized speech signal x by the source pulse
(N) is obtained.

When the output of the discrimination circuit 575 indicates a consonant, the multi-pulse train of pitch prediction of L2 pieces (L2 <L1) is obtained by the above-described configuration to obtain the composite signal x (n).

Further, the subtractor 615 subtracts the synthesized audio signal x (n) from the audio signal x (n) to generate a residual signal e (n).
A small amplitude excitation signal is calculated for (n).

As described in the operation section, the small-amplitude sound source calculation circuit 620 converts the small-amplitude sound source signal of a small section (for example, 5 msec.) Divided into sections shorter than a frame using a plurality of codebooks. And a composite signal e ′ of a small section
When it is determined that the codebook is unnecessary by reproducing (n) and comparing with the residual signal e (n), the output of the small-amplitude sound source calculation circuit is modified so that the codebook is not used. Then, the number of multi-pulses is increased by a predetermined number using the sound source pulse calculation circuit 580.

A code indicating whether the current frame is a vowel section or a consonant section, a determination code representing a small amplitude excitation signal, a codebook index and a gain, a code obtained by quantizing a multi-pulse train output from the quantizer 585, a quantum A code obtained by quantizing the pitch parameter output from the quantizer 516, and further a quantizer
The codes obtained by quantizing the spectral parameters output from the 525 are input to the multiplexer 630, respectively. However, when the current frame is a vowel section, the index and gain of the codebook representing the small-amplitude sound source signal or the amplitude and position of the multipulse in the small section are not input. The multiplexer 630 combines and outputs the above codes.

On the other hand, on the receiving side, the demultiplexer 710 outputs a code of a multi-pulse train, a code of a pitch parameter, a code of a spectrum parameter, a determination code indicating whether the current frame is a vowel section or a consonant section, In some cases, the judgment code and the index and gain code representing the small amplitude excitation signal are separated and output. The excitation pulse decoder 720 decodes the amplitude and position of the multi-pulse. The spectrum parameter decoder 750 includes an inverse quantizer 530 on the transmission side.
Works the same as. The small-amplitude excitation decoder 730 has the same codebook as the small-amplitude excitation calculation circuit 620 on the transmission side, and receives a code indicating that the current frame is a consonant section, and responds to the determination code. In the part using the codebook, a code representing the small amplitude excitation signal is selected and output using the received index. Gain circuit 735
Determines the amplitude of the small-amplitude excitation signal using the code of the received gain according to the determination code when receiving a code indicating that the current frame is a consonant section. The pitch parameter decoder 745 performs the same function as the inverse quantizer 518 on the transmission side. The pulse generator 725 generates a sound source signal based on the multi-pulse train according to the determination code and the determination code. A pitch reproduction filter 755 reproduces a synthesized excitation signal whose pitch has been reproduced by using the obtained excitation signal and the decoded pitch parameter as inputs. The adder 740 adds the sound source signal obtained by regenerating the pitch and the output signal of the gain circuit 735 when receiving a code indicating that the current frame is a consonant section, obtains a driving sound source signal, and obtains a spectrum envelope filter. Drives the circuit 760. The synthesis filter circuit 760 obtains and outputs a synthesized speech waveform using the driving excitation signal and the decoded spectrum parameter.

The configuration described above is only one configuration of the present invention, and various modifications are possible.

If it is determined that a small-amplitude excitation signal is unnecessary in a certain subframe section, the excitation pulse calculation circuit 580 increases the number of pulses by a predetermined number, and then obtains the pulse. It is also possible to re-calculate the small-amplitude excitation signal in an appropriate subframe section.

In order to further reduce the amount of calculation for obtaining the small-amplitude sound source signal, as described in Equations (5) to (7) in the operation section, after calculating the large-amplitude multipulse by pitch prediction, A code book can be selected using the cross-correlation function Φ _xh ′. As described above, the signal e _w (n) does not need to be reproduced at the time of codebook selection, as described in the section of the operation, so that the amount of calculation is significantly reduced as compared with the configuration shown in FIG. it can.

For the consonant part, different codebooks may be created in advance in accordance with the properties of the consonant (for example, burstiness, friction, etc.), and these may be switched and used.

Further, as a method of calculating a multi-pulse,
Various known methods can be used in addition to the method shown in FIG.

Further, other well-known parameters (a line spectrum pair, a cepstrum, a mel cepstrum, a logarithmic cross-sectional area ratio, etc.) can also be used as the spectrum parameter. Further, as a method of quantizing the spectrum parameter, vector quantization or the like can be used in addition to scalar quantization. Reference 3 can be referred to for the vector quantization.

Although the frame length is fixed, it may be variable according to the characteristics of the audio signal.

(Effects of the Invention) According to the present invention, the sound source code is classified into predetermined types (for example, vowel parts, consonant parts or voiced parts, unvoiced parts) as compared with the conventional example, and vowels or In the voiced part, the sound source is transmitted by a relatively small number of pitch-predicted multi-pulse trains. It can be represented by the amount of information. Furthermore, when the effect of improving the characteristics by the codebook is low, there is an effect that the sound quality can be further improved by increasing the number of multi-pulses without using the codebook. Therefore, even if the bit rate is the same as that of the conventional method, there is a great effect that a better reproduced audio signal can be obtained not only in the vowel part but also in the consonant section as compared with the conventional method. Further, this effect becomes more remarkable when the bit rate is reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a speech encoding apparatus and speech decoding apparatus according to the present invention, FIG. 2 is a block diagram showing the operation of the present invention, and FIG. It is a block diagram showing the conventional example. 500, 300, 800 ... input terminal, 510 ... time division circuit, 515
…… Pitch parameter calculation circuit, 320,820… Pitch parameter calculation unit, 520 …… Spectrum parameter calculation circuit, 310,810 …… Spectrum parameter calculation unit, 305…
Voiced / unvoiced discriminator, 516, 525, 585… Quantizer, 518, 53
0,590 …… Dequantizer, 540 …… Weighting circuit, 550 ……
Impulse response calculation circuit, 560 ... autocorrelation function calculation circuit, 570 ... cross correlation function calculation circuit, 575 ... discriminator circuit,
580 …… Sound source pulse calculation circuit, 600,725 …… Pulse generator, 755,605 …… Pitch reproduction filter, 610,760 …… Spectral envelope filter, 620 …… Small amplitude sound source calculation circuit, 630
…… Multiplexer, 710 …… Demultiplexer, 720…
… Excitation pulse decoder, 730 …… Small amplitude excitation decoder, 740,3
75 …… Adder, 745 …… Pitch parameter decoder, 750…
... Spectral parameter decoder, 770 ... Output terminal, 810
... Sound source calculator, 615,360 ... Subtractor, 370 ... Codebook selector, 735 ... Gain circuit, 340,385 ... Switch, 380 ... Comparison / determination unit, 330,830 ... Pitch prediction sound source pulse calculator, 350 ... ... Signal playback unit.

Claims (3)

(57) [Claims] 1. A transmitting side receives a discrete voice signal, divides the signal into frames of a predetermined time length, and generates a pitch parameter representing a pitch structure of the voice signal and a spectrum representing a spectrum envelope of the voice signal. Determining a parameter; classifying the audio signal into a predetermined type and outputting a first discriminating code indicating a classification result; if the classification result indicates a specific type, calculating by pitch prediction It is determined whether or not performance is improved by representing the sound source signal by using the multi-pulse train and the codebook together, as compared with representing the sound source signal of the audio signal of the frame using the multi-pulse train. Generating a second discriminant code indicating the result of the judgment, and when the performance is improved, both the codebook index and the multi-pulse train are encoded to represent the excitation signal When the performance is not improved, the multi-pulse sequence is encoded and output as a code representing the excitation signal. When the classification result does not indicate a specific type, the multi-pulse sequence is encoded and Outputting as a code representing an excitation signal; the pitch parameter, the spectrum parameter, a code representing the excitation signal, the first discrimination code, and the second
On the receiving side, a code representing the excitation signal, a code representing the pitch parameter, a code representing the spectrum parameter, and the first
And a second identification code; a code representing the excitation signal, a code representing the pitch parameter, and a code representing the spectrum parameter are decoded; a multi-pulse train of the decoded excitation signal And restore
A sound source signal whose pitch has been reproduced is obtained using the restored multi-pulse train and the decoded pitch parameter. It is determined whether a codebook is used based on the first discrimination code and the second discrimination code. In the case where the codebook is used, the driving excitation signal is restored using the excitation signal reproduced from the pitch and the index of the codebook included in the decoded excitation signal, and the codebook is not used. Wherein the sound source signal obtained by reproducing the pitch is restored as a drive sound source signal; and the sound signal is output using the restored drive sound signal and the decoded spectrum parameter. Encoding / decoding method. 2. A pitch parameter calculation circuit for obtaining and encoding a pitch parameter representing a pitch structure from an input discrete speech signal; and a spectrum for acquiring and encoding a spectrum parameter representing a short-time spectrum characteristic of the speech signal. A parameter calculation circuit; a classification circuit that classifies the audio signal into a predetermined type and outputs a first determination code indicating the result; and a case where the first determination code indicates a specific type. Compared with representing the sound source signal of the audio signal using a multi-pulse sequence obtained by pitch prediction using the pitch parameter and the spectral parameter, the sound source signal using a codebook in addition to the multi-pulse sequence And whether or not the performance is improved by expressing is output, and a second determination code indicating the determination result is output. If the performance is improved, both the codebook index and the multi-pulse train are coded and output as a code representing the excitation signal. If the performance is not improved, the multi-pulse train is coded to represent the excitation signal. A sound source signal calculation circuit that outputs the code as a code, and when the first discrimination code does not indicate a specific type, encodes the multi-pulse train and outputs the code as a code representing a sound source signal of the audio signal; And a multiplexer circuit for combining and outputting a code representing the spectrum parameter, a code representing the excitation signal, the first discrimination code, and the second discrimination code. Encoding device. 3. A speech decoding device for receiving an output code of the speech encoding device according to claim 2 via a transmission line and restoring the speech signal, wherein the code represents the excitation signal and the pitch parameter. A code, a code representing the spectral parameter, and the first
And a demultiplexer circuit that receives and discriminates the discrimination code and the second discrimination code from each other; a code indicating the excitation signal, a code indicating the pitch parameter, and a code indicating the spectrum parameter separated by the demultiplexer circuit. And a decoding circuit for decoding the sound source signal;
A pitch reproduction circuit for obtaining an excitation signal whose pitch has been reproduced using the restored multi-pulse train and the decoded pitch parameter; and determining whether a codebook is used based on the first discrimination code and the second discrimination code. Judgment, when the codebook is used, the drive excitation signal is restored using the codebook index and the output of the pitch reproduction circuit, and when the codebook is not used, the excitation signal reproduced from the pitch is used. A driving excitation signal restoring circuit for restoring as a driving excitation signal; and a synthesis filter circuit for synthesizing the audio signal using the restored driving excitation signal and the decoded spectrum parameter. Audio decoding device.