JPH09326772A - Voice coding device and voice decoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the transmission characteristic by providing an audible sense weighting filter and a reflection coefficient discrimination means so as to use the filter when the reflection coefficient is at the outside of a specified range and not using the filter when the reflection coefficient is within the specified range. SOLUTION: A reflection coefficient discrimination section 17 receiving a reflection coefficient calculated by an audible sense weighting filter adaptive device 7 provides an output of '1' when the reflection coefficient is within a specified range and provides an output of '0' when the reflection coefficient is at the outside of the range so as to allow usual filter operation. A switch control section A18 monitors a signal on a signal line 101 and sends a control signal with a content of denoting that a switch 19 selects a signal not passing through an audible sense weighting filter 8 to a signal line 102 when the signal is '1' and sends the control signal with a content of denoting that the switch 19 selects a signal passing through the audible sense weighting filter 8 to the signal line 102 when the signal is '0'. The switch 19 selects any signal under this control. Thus, even in the case of reception of a DTMF signal not a voice signal, production of distortion is prevented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a telephone band voice encoding apparatus and decoding apparatus.

[0002]

2. Description of the Related Art In many speech coding systems, a perceptual weighting filter and a quantization noise shaping filter (hereinafter referred to as "hearing correction filter") are used in an encoder in order to reduce quantization noise and improve auditory sound quality. However, a post filter is used in the decoder. This is, for example, 4
Monthly IEEE Communications Section Newsletter (IEEE TRANSACTION ONCOMMU
NICATIONS), VOL.COM-30.No.4, "Predictive Coding of Speech at low bit rate"
Low Bit Rates): A noise shaping filter by BISHNS.S.ATAL, and "adaptive postfiltering of 16 kbit / s ADPCM voice" (ADAPTIVE POSTFILTE) published in 1986.
RING OF 16kbit / s-ADPCM SPEECH): NSJAYANT, V.RAM
A post filter is described by AMOORTHY.

As a first conventional technique of a voice encoding system,
The TTC standard JT-G.3, which uses a perceptual weighting filter for the encoder and a post filter for the decoder, is used. I will list 728. FIG. 35 shows "16 kbit / s speech coding method using JT-G.728 low delay code excitation linear prediction (LD-CELP)" (TTC standard Vol. V third volume high layer protocol [coding method] Heisei It is a block diagram of the LD-CELP encoder of the 5th telephone and telephony technical committee edit. In the figure, 1 is an excitation VQ codebook, 2 is a gain adjustment unit,
3 is a synthesis filter, 4 is a backward gain adaptor, 5 is a backward synthesis filter adaptor, 6 is a signal adder, 7
Is a perceptual weighting filter adaptor, 8 is a perceptual weighting filter, and 9 is a least mean square error calculator.

Next, the operation of the encoder will be described. The input signal is treated as five consecutive input samples (hereinafter referred to as vectors), and the auditory weighting filter adaptive unit 7
Is input to the signal addition unit 6. The perceptual weighting filter adaptor 7 analyzes the input signal by the Levinson-Durbin method, and outputs the filter coefficient indicating the characteristic of the input signal to the perceptual weighting filter 8. On the other hand, for each input vector, the encoder passes the 1024 codebook vector candidates through the gain adjusting unit 2 and the synthesis filter 3, and selects an input signal from the 1024 quantized signal vector candidates as a result. On the other hand, the frequency-weighted root-mean-square error for improving the perceptual quality is determined to be the minimum. The encoder corresponds to the optimum code vector for generating the optimum quantized signal vector determined by the above procedure.
Send the codebook index of bits to the decoder. The optimal code vector is then passed through the gain adjuster 2 and the synthesis filter 3 in order to update the filter memory in preparation for coding the next signal vector. The coefficients and gain of the synthesis filter are updated by backward adaptation based on the adjusted excitation signal.

FIG. 36 shows "16 kbit / s speech coding method using JT-G.728 low delay code excitation linear prediction (LD-CELP)" (TTC standard Vol. Method] A block diagram of an LD-CELP decoder (edited by the Telegraph and Telephone Technical Committee in 1993). In the figure, 1 to 5 are the same as the encoder, and 10 is a post filter.

Next, the operation of the decoder will be described. The decoding operation is also performed for each vector. Based on the received 10-bit index, the decoder extracts the corresponding code vector from the excitation VQ codebook 1. The extracted code vector is passed through the gain adjustment unit 2 and the synthesis filter 3 to generate the decoded signal vector at that time, and the post filter 1 is used to improve the perceptual quality.
It is passed through 0 and output. The synthesis filter coefficient and gain are
Updated in a manner similar to the encoder, the post filter coefficients are updated periodically with the information available at the decoder.

Next, as a second conventional technique, an encoder has an auditory weighting filter, and a decoder has a post filter.
Furthermore, pitch information is required for encoding and decoding,
A forward type CELP speech coding method will be given as an example of the speech coding method characterized in that the information is captured in a part of the coded data and transmitted. Figure 37 shows the IEEE Global Telecommunications issued in 1989.
Conference & Exhibition, Conference Record Vol.2
of 3, "16 kbit / s Low Delay CELP Speech Coding System"
(A ROBUST LOW-DELAY CELP SPEECH CODER AT 16 KBIT /
S): Conventional forward adaptive CE by Juin-Hwey Chen
It is a block diagram of an LP encoder. In the figure, 1 is an excitation VQ codebook, 2 is a gain adjustment unit, 11 is an LPC analysis unit, 12 is a data multiplexing unit, 14 is a long period prediction synthesis filter, 13 is a short period prediction synthesis filter, 8 is a perceptual weighting filter, Reference numeral 7 is an auditory weighting filter adaptation unit, and 9 is a least mean square error calculation unit.

Next, the operation of the encoder will be described. The input signal is treated as a plurality of continuous input samples (hereinafter referred to as vectors), and is input to the auditory weighting filter adaptive unit 7, the LPC analysis unit 11 and the signal addition unit 15.
The perceptual weighting filter adaptor 7 analyzes the input signal by the Levinson-Durbin method, and passes the perceptual weighting filter coefficient indicating the characteristics of the input signal to the perceptual weighting filter 8. On the other hand, in the LPC analysis unit 11, LPC analysis is performed on each input vector, and the long-period prediction synthesis filter coefficient obtained by the LPC analysis is included in the long-period prediction synthesis filter 14 and the short-period prediction synthesis filter coefficient. To the short cycle prediction synthesis filter 13 to update the long cycle prediction synthesis filter coefficient and the short cycle prediction synthesis filter coefficient. The encoder adds all the codebook vector candidates to the gain adjusting unit 2 and the signal that has passed the signal one vector before through the long cycle prediction synthesis filter 14 and further predicts that signal and the signal one vector before the short cycle prediction. From all the quantized signal vector candidates to which the signals passed through the synthesizing filter 13 are added, the one having the smallest frequency-weighted root-mean-square error for improving the perceptual quality of the input signal is determined. To do. The encoder uses the codebook index corresponding to the optimum code vector for generating the optimum quantized signal vector determined in the above procedure, the gain information, and the long cycle prediction synthesis filter coefficient and the short cycle obtained by the LPC analysis unit 11. The predictive synthesis filter coefficient is multiplexed by the data multiplexing unit 12 and transmitted to the decoder.

FIG. 38 shows the conventional forward adaptive C described above.
It is a block diagram of the decoder corresponding to an ELP encoder. In the figure, 1, 2, 13 and 14 are the same as the encoder, 1
Reference numeral 0 is a post filter, and 16 is a data separation unit. Next, the operation of the decoder will be described. The decoding operation is also performed for each vector. The coded index, the gain information, the long period prediction synthesis filter coefficient, and the short period prediction synthesis filter coefficient are extracted from the received encoded data in the data separation unit 16, and the excitation VQ codebook 1, the gain adjustment unit 2, and the long period prediction are extracted, respectively. It is sent to the synthesis filter 14 and the short cycle prediction synthesis filter 13. The decoder uses the transmitted codebook index to extract a corresponding code vector from the excitation VQ codebook 1, and the extracted code vector is used to generate a decoded signal vector at that time. 2, long cycle prediction synthesis filter 1
4 and the short cycle prediction synthesis filter 13, and is passed through the post filter 10 and output in order to improve the quality of hearing. The synthesis filter coefficients and gains are updated in a manner similar to the encoder, and the post filter coefficients are updated periodically with the information available at the decoder.

[0010]

The conventional speech coding apparatus is configured as described above. In the former first configuration, the auditory weighting filter is provided on the encoder side and the post filter is provided on the decoder side. It is always inserted and frequency weighting is applied to improve the perceptual quality of the audio signal. In addition to this, DTMF (Dual Tone Multi Frequency) which is a non-voice signal
) Even when a signal is input, since the perceptual weighting filter and the post filter perform frequency weighting, the signal is distorted, and the DTMF
There is a problem of deteriorating the signal transmission characteristics. Also,
In the latter second configuration, in addition to the above problem, pitch prediction is performed on a signal with a sharp increase in level,
There is a problem in that the pitch prediction is not successful and, conversely, distortion occurs at the rising portion of the signal and distortion occurs at the rising portion of the DTMF signal that is a burst signal, and the transmission characteristics of the signal deteriorate.

The present invention has been made to solve the above problems, and encodes / decodes a DTMF signal in a speech encoding / decoding device having a hearing correction filter, a post filter, and a long-period predictive synthesis filter. In this case, it is an object of the present invention to obtain a speech encoding / decoding device having good transmission characteristics by performing encoding / decoding processing without passing through each of the above filters.

[0012]

A speech coding apparatus according to the present invention obtains a filter coefficient which is characteristic of an input signal in coding a signal in an input speech band to improve a perceptual quality and a frequency characteristic. A perceptual weighting filter that changes,
A reflection coefficient determining means for comparing a reflection coefficient obtained by analyzing an input signal with a set value, and an auditory weighting filter is used when the reflection coefficient is out of a specified range, and an auditory weighting filter is used when the reflection coefficient is within the specified range. The use switching means which does not use is provided.

Furthermore, the reflection coefficient is obtained by an auditory weighting filter adaptation section which obtains an auditory weighting filter coefficient indicating the characteristics of the input audio signal.

Furthermore, the reflection coefficient is obtained by a synthesis filter adaptation unit which obtains a coefficient when a vector candidate from the codebook is passed through the synthesis filter.

Furthermore, a signal level monitoring means for monitoring the change of the level of the input signal is added, and the use switching means switches the use of the auditory weighting filter by the combination of the reflection coefficient and the change of the level of the input signal. .

Furthermore, data reflection means for multiplexing the reflection coefficient obtained by analyzing the input signal and the vector code selected from the codebook is added and transmitted.

Furthermore, the use switching means sets the filter coefficient of the auditory weighting filter to the current value instead of switching between use and non-use of the auditory weighting filter, or
The initial value is switched.

Alternatively, in the decoding of the received code, a post filter for enhancing the perceptual quality of the output signal from the synthesis filter unit for decoding the code vector from the code book corresponding to the reception code, and its coefficient in the synthesis filter unit. And a reflection coefficient determination means for comparing the reflection coefficient obtained by this synthesis filter adaptation means with a set value, and if this reflection coefficient is out of the specified range, a post filter is used, The use switching means that does not use the post filter is provided inside.

Furthermore, signal level monitoring means for monitoring the change in the level of the output signal from the synthesis filter section is added, and the use switching means switches the use of the post filter by the combination of the reflection coefficient and the change in the level of the output signal. I decided to do it.

Further, in place of the reflection coefficient determining means, a data separating means for demultiplexing the encoded data and the reflection coefficient from the received encoded data is added, and the reflection coefficient obtained by separating by this data separating means. The use of the post filter is switched depending on the value of.

Alternatively, in encoding a signal in the voice band, the above signal is converted into an LPC (Linear Prediction).
Coding) Reflection coefficient determination means for comparing the reflection coefficient obtained by the LPC analysis section with a set value, and auditory weighting for weighting vector candidates based on the codebook with auditory weighting filter coefficients obtained by analyzing the input signal. A filter, a long-period prediction synthesis filter that synthesizes a signal using pitch information, and a use switching unit that switches between use and non-use of the auditory weighting filter and the long-period prediction synthesis filter according to the result determined by the reflection coefficient determination unit. It was

Furthermore, a signal level monitoring means for monitoring the level of the input signal is added, and the result of the reflection coefficient determination means and the value of the sound level obtained by the signal level monitoring means are used for the auditory weighting filter. Switched between use and non-use.

Furthermore, instead of obtaining the reflection coefficient from the LPC analysis section, the reflection coefficient is obtained from the auditory weighting filter adaptation section that obtains the auditory weighting filter coefficient.

Alternatively, in the decoding of the received code, the data separation means for demultiplexing the received coded data and the reflection coefficient from the received signal, and the synthesis filter section for decoding the code vector from the codebook corresponding to the received coded data are used. A post filter for improving the perceptual quality of the output signal and a use switching means for switching between use and non-use of the post filter by the reflection coefficient obtained by the data separating means are provided.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. 1 is a configuration block diagram of a speech coding apparatus according to Embodiment 1 of the present invention. In the figure, 1 to 9
Are the same elements as the encoder of FIG. 35 shown in the first conventional art, and the description thereof will be omitted. A new element 17
Is a signal which has received the first-order and second-order reflection coefficients calculated by the auditory weighting filter adaptive unit 7 which enhances the auditory quality and changes the frequency characteristic, determines the numerical values thereof, and which has passed through the auditory weighting filter 8. It is a reflection coefficient determination unit that determines selection of a signal that does not pass. 18 is a reflection coefficient determination unit 17
It is a switch control unit that controls a switch (use switching unit) 19 that selects a signal that has passed through the auditory weighting filter 8 or a signal that has not passed according to the received signal. FIG.
The operation of the encoder is basically the same as that of the encoder shown in the first prior art, and a reflection coefficient determination unit 17, a switch control unit A18 and a switch 19 are added.

According to the present invention, when a DTMF signal which is a non-speech signal is input, when the reflection coefficient is within the range of a predetermined value prepared in advance, a signal which has not passed through the auditory weighting filter 8 is selected, When the value is out of the specified range,
A speech coding code characterized by performing coding processing without passing through the perceptual weighting filter 8 by using a use switching unit that selects a signal that has passed through the perceptual weighting filter 8, that is, returns to normal processing. It is a vessel.

Next, the method for calculating the specified value of the reflection coefficient will be described in detail. FIG. 2 is a flowchart of the Levinson-Durbin method described in the document: "Basics of Speech Information Processing", published by Ohmsha. In the figure, each variable, v _0: 0-order autocorrelation coefficients v _1: 1 order autocorrelation coefficient alpha _i: forward linear prediction coefficient omega _n: correlation u _n between forward prediction error and a backward prediction error : Reflection coefficient p: order of the prediction filter

The sampling frequency is 8 kHz and the delay unit for autocorrelation is 0.125 msec. Here, 0.125 msec is replaced with the symbol τ.
Here, the reflection coefficient of a sine wave with each frequency ω ₀ and amplitude A will be calculated. This sine wave is expressed by the following equation (1). It is expressed as f (t) = Asin ω ₀ t (1). The 0th-order autocorrelation coefficient value is represented by Expression (2). ^{_{v 0 = A 2/2 (}} 2) n following self correlation coefficient value is represented by the formula (3). ^{_{v n = A 2/2 ×}} cosω 0 nτ (3) where, in accordance with Levinson-Durbin technique of Figure 2, when the first-order reflection coefficient of k1,2 order reflection coefficient is k2, the following equation (4) , Equation (5) is obtained. k1 = cos ω ₀ nτ (4) k2 = −1.0 (5) As described above, regarding the sine signal, k1, peculiar to the signal,
Since the value of k2 is taken, when a signaling signal of a sine wave signal is input, if the values of k1 and k2 are investigated, if the range of values taken by k1 and k2 is specified in advance, When inputting the signaling signal of the sine wave signal,
Code without passing through the auditory sense correction filter or post filter
A decryption process can be performed.

Next, compared with a sine wave with each frequency ω ₀ and amplitude A, and each frequency ω ₀ , each frequency is different by Δω,
The reflection coefficient of the addition signal with the sine wave having the amplitude A will be calculated. This signal is expressed by equation (6). The zero-order autocorrelation coefficient value is given by equation (7). v ₀ = A ² (7) The autocorrelation coefficient value of the nth order is as shown in Expression (8). Here, when the reflection coefficients k1 and k2 are obtained according to the Durbin method of FIG. 2, Equations (9) and (10) are obtained.

[0030]

[Equation 1]

In this case, the more the difference between the frequencies of the two sine waves is eliminated, the closer to k2 = -1.0. For example, the frequency used for the DTMF signal is 697 Hz to 1477 Hz, which is smaller than the sampling frequency,
And since the difference between the two frequencies is also small, equation (1
K2 of 0) has a value close to −1.0. Also, k1
Takes a value peculiar to the signal from equation (9). Therefore,
Code without passing through the auditory sense correction filter or post filter
If you input the DTMF signal you want to decode, k
If the values of 1 and k2 are investigated and the ranges of k1 and k2 are defined in advance, the encoding / decoding process can be performed without passing through the auditory sense correction filter or the post filter. K even if the input signal is a voice signal
The values of 1 and k2 are within the range of the specified value, and the encoding / decoding process may be performed without passing through the auditory sense correction filter or the post filter. However, the frequency at which the audio signal falls within the range of the specified value. Is so small that the effects of the hearing correction filter and the post filter are not reduced.

Next, the procedure of performing the encoding process without passing through the auditory weighting filter 8 will be described in detail with reference to FIGS. 3 and 4 are configuration diagrams of the reflection coefficient determination unit 17 and the switch control unit A18 shown in FIG. 1, respectively. In FIG. 3, 100 and 101 are signal lines, 17a is a reflection coefficient distribution unit, 17b is a k1 determination unit, 17c is a k2 determination unit, 17d is a comprehensive determination unit, and in FIG.
Reference numeral 02 is a signal line, 18a is a signal monitoring unit, and 18b is a switch control signal transmitting unit.

Next, the operation will be described. In the perceptual weighting filter adaptive unit 7 of FIG. 1, the reflection coefficient determination unit 17 that has received the reflection coefficient calculated by the Levinson-Durbin method, as shown in FIG. Received more k
The reflection coefficients of 1 and k2 are calculated as k1 determination units 17b and k2, respectively.
It is passed to the determination unit 17c and it is determined whether it is within the range of the specified value. When both coefficients are determined to be within the specified value range by the comprehensive determination unit 17d, the signal "1" is output to the signal line 101, and when the coefficients are out of the range, the normal filter operation is performed to "0". Is output.

As shown in FIG. 4, the switch control section A18 monitors the signal on the signal line 101 by the signal monitoring section 18a,
If the signal is "1", the switch control signal transmission unit 18b outputs a control signal for selecting a signal that the switch 19 does not pass through the auditory weighting filter 8 to the signal line 102.
To send to. If the signal is “0”, the switch 19 sends a control signal to the signal line 102 to select the signal passed through the auditory weighting filter 8. The switch 19 selects one of the signals according to the above control. By the above, the encoding process that does not pass the auditory weighting filter 8 can be performed.

Further, instead of selecting whether the signal after processing the input signal is passed or not passed through the auditory weighting filter 8, it is selected whether the filter coefficient of the auditory weighting filter 8 is switched to the initial value or the current value. You can
That is, the initial value of the perceptual weighting filter coefficient is a value that is not frequency-weighted, and if the perceptual weighting filter 8 is used to pass the value, no weighting is performed as a result. FIG. 5 shows the auditory weighting filter 8 according to the present embodiment.
It is a figure which shows the structure which makes the initial value of the filter coefficient of above into said value. The present invention is not limited to the embodiment having the auditory weighting filter, but can be applied to the case having a quantization noise shaping filter.

In the above embodiment, the case where the auditory weighting filter 8 is switched between use and nonuse by using the reflection coefficient calculated by the auditory weighting filter adaptive unit 7 has been described. A speech coding apparatus having another configuration will be described below. A local decoding unit, that is, an excitation VQ codebook 1, a gain adjusting unit 2, a synthesizing filter 3, a gain adaptor 4, is provided inside the encoder.
In the case where the synthesis filter adaptive unit 5 is provided, use or non-use of the auditory weighting filter 8 may be switched using the reflection coefficient calculated by the synthesis filter adaptive unit 5. FIG. 6 is a configuration block diagram of the encoding device in this case. 1 to 9 and 17 to 19 are shown in FIG.
This is the same element as, and its description is omitted.

The operation of the coding apparatus having the configuration shown in FIG. 6 is basically the same as that of the apparatus having the configuration shown in FIG. 1, and the place where the reflection coefficient is extracted is changed from the auditory weighting filter adaptive unit 7 to the synthesis filter adaptive unit 5. It was done. Similar to the configuration of FIG. 1, the auditory weighting filter 8 may be used or not used instead of selecting the initial value or the current value as the filter coefficient of the auditory weighting filter 8.

Embodiment 2 In this embodiment, a signal level monitoring unit A20 is added to the first embodiment shown in FIG. 1 and a switch control unit B23 is provided instead of the switch control unit A18, and its configuration block diagram is shown in FIG. . In the figure, reference numerals 1 to 9, 17 and 19 are the same as those of the encoding apparatus of FIG. In addition, FIG.
FIG. 3 is a configuration diagram of a signal level monitoring unit A20 and a switch control unit B23, respectively. In FIG. 8, 103, 104
Is a signal line, 20a is a buffer A, 20b is a buffer B,
20c is a power calculation unit A, 20d is a power calculation unit B, 2
0e is a signal power comparison unit, and in FIG.
2, 104 are signal lines, 23a is a reflection coefficient monitoring unit, and 23b
Is a logical sum calculation unit, 23c is a reset signal transmission unit, and 23d
Is a switch control signal transmitter.

Next, the operation will be described. In FIG. 7, the input signal of the signal line 103 is fetched by the buffer A20a in the signal level monitoring unit A20 shown in FIG. The signal taken into the buffer A 20a is supplied to the power calculation unit A
After transmitting the signal to 20c, it is moved to the buffer B20b, and when the next signal is input to the buffer A20a, it is moved to the power calculating unit B20d. That is, the buffers A and B referred to here correspond to the present and previous signals, respectively. The size of the buffers A and B is 1 msec. Power calculator A20c and power calculator B20
In d, the powers of the signals received from the buffers A and B are calculated and sent to the signal power comparison unit 20e. The signal power comparison unit 20e receiving the result compares the two signal powers, and it is confirmed that the signal power of the buffer A 20a is higher than the signal power of the buffer B 20d by 25 dB or more, that is, a sharp level increase of 25 dB or more. Only in this case, "1" having a signal width of 1 msec is transmitted to the signal line 104. In other cases, the signal "0" is transmitted.

In the switch control unit B23 shown in FIG. 9, the signal on the signal line 101 output from the reflection coefficient determination unit 17 enters the reflection coefficient monitoring unit 23a, and the values of k1 and k2 are out of the predetermined range. Is monitored and if it is out of the range, the reset signal sending unit 23c is instructed to send a reset signal. The reset signal transmission unit 23c that has received the reset instruction operates the switch control signal transmission unit 23c.
A reset signal is issued so that the signal held by d is always "0". On the other hand, when the reflection coefficient is within the range of the predetermined value in the reflection coefficient determination section 17, the logical sum calculation section 23b makes a sudden change in the signal line 101 and the signal level monitoring section A20 to which the signal of "1" is transmitted. When it is confirmed that the level has risen, the logical sum of the signals on the signal line 104 to which the signal of "1" of 1 msec is transmitted is calculated, and the information is passed to the switch control signal transmission section 23d. When the received signal becomes "1", the switch control signal sending section 23d sends a signal of "1" to the signal line 102 and continues to hold it until reset by the reset signal sending section 23c.

The switch 19 receives the control signal from the signal line 102 and, when a control signal for selecting a signal not passing through the auditory weighting filter 8 is received, selects a signal not passing through the auditory weighting filter 8. If the above method is used, the encoding process can be performed without passing through the auditory weighting filter 8. In this method, the rising edge of the DTMF signal which is a burst signal is detected,
Since the signal that does not pass the auditory weighting filter 8 is selected, the signal that does not pass the auditory weighting filter 8 is prevented from being selected when the reflection coefficient at the time of a voice signal falls within the range of the specified value. . Further, as in the first embodiment, the auditory weighting filter 8 may be used or not used, and an initial value or a current value may be selected as a filter coefficient of the auditory weighting filter 8. . FIG. 10 is a configuration diagram in that case.
Further, FIG. 11 is a detailed configuration diagram of the filter control unit A26 of FIG.

In the filter control unit A26 shown in FIG. 11, the signal line 101 output from the reflection coefficient determination unit 17 is used.
Signal enters the reflection coefficient monitoring unit 26a, and it is monitored whether the values of k1 and k2 are out of the predetermined range. If they are out of the range, the reset signal is sent to the reset signal sending unit 26c. Give instructions. The reset signal transmission unit 26c that has received the reset instruction operates the filter control signal transmission unit 26c.
A reset signal is issued so that the signal held by d is always "0". On the other hand, when the reflection coefficient is within the range of the predetermined value in the reflection coefficient determination section 17, the logical sum calculation section 26b outputs a signal of "1" in the signal line 101 and the level monitoring section A20. When it is confirmed that the level has risen, a logical sum of the signals on the signal line 104 to which a signal of "1" of 1 msec is transmitted is calculated, and the information is passed to the filter control signal transmission unit 26d. Filter control signal transmitter 2
When the received signal becomes "1", 6d sends a control signal from the signal line 106 for the filter to select an initial value, and continues to hold the reset signal sending unit 26c until it is reset.

In the above embodiment, the signal that has passed through the auditory weighting filter 8 is selected again, that is, returned to normal processing when the monitored reflection coefficient falls outside the specified range. However, the signal level monitoring unit B21 may monitor the level of the input signal and return the level when the level of the input signal becomes very small. A configuration block diagram in this case is shown in FIG. FIG.
, 1 to 9, 17 and 19 are the same as those of the encoding device of FIG.

13 and 14 are block diagrams of the signal level monitoring unit B21 and the switch control unit C24, respectively. In FIG. 13, 103, 104 and 105 are signal lines, 21a is a buffer A, 21b is a buffer B and 21c.
Is a power calculation unit A, 21d is a power calculation unit B, 21e is a level monitoring unit, 21f is a signal power comparison unit, and 21g is a reset signal transmission unit. In FIG. 14, 101,1
Reference numerals 04 and 105 are signal lines, 24a is a logical sum calculation unit, and 24b.
Is a switch control signal transmitter.

Next, the operation will be described. In FIG. 12, the input signal of the signal line 103 is captured by the buffer A21a in the signal level monitoring unit B21 shown in FIG. The signal taken into the buffer A21a is sent to the power calculation unit A21c, and then the signal is sent to the buffer B21b.
When the next signal is input to the buffer A21a, it is transferred to the power calculation unit B21d. That is, the buffers A and B referred to here correspond to the present and previous signals, respectively. The size of buffers A and B is 1 msec.
Is the length of Power calculator A21c and power calculator B
In 21d, the powers of the signals received from the buffers A and B are calculated and sent to the signal power comparison section 21f.
When the signal power comparison unit 21f receives the result, 2
The two signal powers are compared with each other, and only when it is confirmed that the signal power of the buffer A 21a is higher than the signal power of the buffer B 21d by 25 dB or more, that is, a rapid level increase of 25 dB or more is confirmed, the signal line 104 having a signal width of 1 msec 1 ”is transmitted. In other cases, the signal "0" is transmitted.

On the other hand, in the level monitor 21e, the signal on the signal line 103 is constantly monitored, and the signal level is -40d.
Monitor whether the signal is smaller than Bm0, -40dB
If the signal is smaller than m0, the control signal is sent to the reset signal sending unit 21g, and the reset signal sending unit 21g receiving the control signal sends the reset signal to the signal line 105. In the switch control unit C24 shown in FIG. 14, when the signal on the signal line 105 output from the signal level monitoring unit B21 is a reset signal, the signal held by the switch control signal sending unit 24b is always "0". Reset to "". On the other hand, when the reflection coefficient is within the range of the predetermined value in the reflection coefficient determination section 17, the logical sum calculation section 23b outputs the signal "1" to the signal line 101.
When a rapid level rise is confirmed in the signal level monitoring unit A20, the logical sum of the signals of the signal line 104 to which the signal of "1" of 1 msec is transmitted is calculated, and the information is passed to the switch control signal transmission unit 24b. . When the received signal is "1", the switch control signal sending unit 24b
A signal of "1" is transmitted to the signal line 102, and is kept held until a reset signal is received from the signal line 105.

The switch 19 receives the control signal from the signal line 102, and when a control signal for selecting a signal not passing through the auditory weighting filter 8 is received, selects a signal not passing through the auditory weighting filter 8. If the above method is used, the encoding process can be performed without passing through the auditory weighting filter 8. In this method, the rising edge of the DTMF signal which is a burst signal is detected,
Since the signal that does not pass the auditory weighting filter 8 is selected, the signal that does not pass the auditory weighting filter 8 is prevented from being selected when the reflection coefficient at the time of a voice signal falls within the range of the specified value. .

As in the other embodiments, the auditory weighting filter 8 can be used or not used instead of the initial value or the current value as the filter coefficient of the auditory weighting filter 8. FIG. 15 is a configuration diagram thereof, and FIG. 16 is a detailed configuration diagram of the filter control unit B27 of FIG.

In the filter control section B27 shown in FIG. 16, the signal line 1 output from the signal level monitoring section B21.
When the signal 05 is a reset signal, the signal held by the switch control signal transmitter 27b is reset so that it is always "0". On the other hand, the logical sum calculation unit 27b
In the reflection coefficient determination unit 17, it was confirmed that the signal line 101 to which the signal of "1" is transmitted and the level monitoring unit A20 when the reflection coefficient is within the predetermined value range have a sharp level increase. Sometimes, a logical sum of the signals of the signal line 104 to which a signal of "1" of 1 msec is transmitted is calculated, and the information is passed to the switch control signal transmission section 27b. When the received signal is "1", the filter control signal sending unit 27b sends a control signal from the signal line 106 for the filter to select an initial value, and until a reset signal is received from the signal line 105. Keep holding.

The above configurations may be slightly changed, and the configurations of FIGS. 6 and 7 may be combined. That is, the reflection coefficient calculated by the synthesis filter adaptive unit 5 may be used to monitor the level of the input signal, and whether the auditory weighting filter 8 is used or not may be selected based on the combination. FIG.
It is a block diagram of a configuration of an encoding device in this case. In the figure, 1 to 9, 17, 19, 20, 23 are the same as the elements in FIG.

The operation of the coding apparatus shown in FIG. 17 is basically the same as that in the second embodiment, except that the place where the reflection coefficient is extracted is changed from the auditory weighting filter adaptive unit 7 to the synthesis filter adaptive unit 5. . Therefore, detailed description of the operation is omitted.

The configurations of FIGS. 6 and 12 may be combined. That is, the reflection coefficient calculated by the synthesis filter adaptive unit 5 may be used, or the level of the input signal may be monitored, and the combination may be determined to select the signal that has not passed the auditory weighting filter 8. FIG. 18 is a configuration block diagram of the encoding device in this case. In the figure, 1 to 9, 17, 19, 21, and 24 are the same as the elements in FIG.

The operation of the coding apparatus shown in FIG. 18 is basically the same as that of the second embodiment, except that the place where the reflection coefficient is extracted is changed from the auditory weighting filter adaptive unit 7 to the synthesis filter adaptive unit 5. . Therefore, detailed description of the operation is omitted.

Embodiment 3 FIG. In this embodiment, a data multiplexing unit 22 is added to the encoding device of the first embodiment shown in FIG. 1, and its configuration block diagram is shown in FIG. In the figure, reference numerals 1 to 9 and 17 to 19 are the same as those of the encoding apparatus in FIG. Reference numeral 22 is a data multiplexing unit.

The operation of the coding apparatus of FIG. 19 is basically the same as that of the apparatus of the first embodiment, but the reflection coefficient determining unit 1
In the data multiplexer 22, a signal is sent together with codebook index information, and a mechanism for transmitting the multiplexed signal to the decoder side is added. This eliminates the need for the reflection coefficient determination unit in the decoding device and, in some cases, signal level monitoring. In this configuration, it is equivalent to that the switching signal of the selection switch which is used or not used of the auditory weighting filter is given to the data multiplexing unit. This switching signal is given to the data multiplexer 22 and transmitted to the decoding device as encoded data. The operation other than the above is the same as that of the first embodiment, and thus the description thereof is omitted.

It is also possible to combine the configurations of FIGS. 6 and 19 by slightly changing the above configuration. That is, the data multiplexing unit 22 may be added, and its configuration block diagram is shown in FIG. In the figure, 1 to 9 and 17 to 19 are the same elements as those of the encoding device of FIG. Reference numeral 22 is a data multiplexing unit.

The operation of the coding apparatus shown in FIG. 20 is basically the same as that of the apparatus having the configuration shown in FIG. 19, in which the signal sent from the reflection coefficient determination section 17 is multiplexed with the codebook index information in the data multiplexing section 22. , A mechanism for transmitting to the decoder side is added. Therefore, the operation description is
Omit it.

It is also possible to combine the configurations of FIGS. 7 and 19 by slightly changing the above configuration. A block diagram of the configuration in this case is shown in FIG. The operation of the coding apparatus of FIG. 21 is also basically the same as that of the apparatus of the configuration of FIG. 19, and the signal sent from the reflection coefficient determination unit 17 is multiplexed with the codebook index information in the data multiplexing unit 22, and the decoder It is the one that adds a mechanism to transmit to the side.

Further, the configurations shown in FIGS. 12 and 19 may be combined by slightly changing the above configuration. A configuration block diagram in this case is shown in FIG. The operation of the encoding device shown in FIG.
Basically, it is the same as the device of the configuration of FIG. 19, except that the signal sent from the reflection coefficient determination unit 17 is multiplexed with the codebook index information in the data multiplexing unit 22 and is transmitted to the decoder side. is there.

Further, the configurations shown in FIGS. 17 and 19 may be combined by slightly changing the above configuration. A configuration block diagram in this case is shown in FIG. The operation of the encoder of FIG. 23 is also basically the same as that of the device of the configuration of FIG. 19, and the signal sent from the reflection coefficient determination unit 17 is multiplexed with the codebook index information in the data multiplexer 22, and the decoder side It is the one that adds a mechanism to transmit to.

Furthermore, the configurations of FIGS. 18 and 19 may be combined. A configuration block diagram in this case is shown in FIG. The operation of the coding apparatus of FIG. 24 is also basically the same as that of the apparatus of the configuration of FIG. 19, and the signal sent from the reflection coefficient determination unit 17 is multiplexed with the codebook index information in the data multiplexing unit 22, and the decoder It is the one that adds a mechanism to transmit to the side.

Embodiment 4 The decoding device will be described.
FIG. 25 is a configuration block diagram of a decoding device corresponding to the coding device according to the first embodiment. In the figure, 1 to 5 and 10 are the same elements as the decoder shown in FIG. 36 shown as a conventional apparatus, and 17 to 19 are the same as those described in the first embodiment, so description will be omitted. .

The operation of the decoding apparatus shown in FIG. 25 is basically the same as that of the conventional decoder shown in FIG. 36, and the auditory weighting filter 8 is used by using the reflection coefficient described in the first embodiment.
The method of switching between use and non-use is applied to the decoding device. Thus, when the DTMF signal is input, the decoding process is performed without passing through the post filter 10.

The operation of the decoding device of this configuration is the same as the conventional one with respect to the decoding operation itself. However, DTMF
The signal is detected based on the reflection coefficient calculated by the synthesis filter adaptive unit 5. Changeover switch (use changeover means) with the same method and reference as the reflection coefficient detection in the voice encoding device
The control of 19 is performed. Further, similarly to the voice encoding device, instead of selecting whether to use the post filter 10 or not, the initial value or the current value may be selected as the filter coefficient of the post filter 10. That is, the initial value of the post filter coefficient is a value that is not frequency-weighted, and if the value is passed through the post filter 10, it is equivalent to the result that weighting is not performed. FIG.
6 is a configuration diagram in that case.

Embodiment 5 A signal level monitoring unit A20 may be added to the decoding apparatus according to the fourth embodiment of the present invention, and a switch control unit B23 may be provided in place of the switch control unit A18. A configuration block diagram in this case is shown in FIG.
FIG. In the figure, 1 to 5, 10, 17 to 19
Are the same as the elements of the decoding apparatus of FIG. 25 shown in the fourth embodiment, and 20 and 23 are the same as those of FIG. 8 and FIG.

The operation of the decoding apparatus shown in FIG. 27 is basically the same as that of the decoding apparatus having the configuration shown in FIG. 25, but a mechanism is adopted in which the use / non-use of the post filter 10 is used together with the signal level monitoring. It was done. According to this configuration, the rising edge of the DTMF signal which is a burst signal is detected to confirm the input of the DTMF signal, and in that case, the signal that the post filter 10 does not pass is selected. Further, when the reflection coefficient at the time of the audio signal is out of the range of the specified value, it is determined that the audio is input, and the post filter 10 is switched. Also, instead of selecting whether to use the post filter 10 or not, the selection of the initial value and the current value of the filter coefficient of the post filter 10 may be replaced as in the fourth embodiment.

It is also possible to slightly change the above configuration and add a signal level monitoring unit B21 to the decoding apparatus of the fourth embodiment and provide a switch control unit C24 instead of the switch control unit A18. FIG. 28 is a configuration block diagram in this case. In the figure, 1 to 5, 10, 17 to 1
9 is the same as the element of the decoding device of FIG.
Since 1 and 24 are the same as those in FIGS. 13 and 14, the description thereof will be omitted.

The operation of the decoding apparatus of FIG. 28 is basically the same as that of the decoding apparatus of FIG. 25 shown in the fourth embodiment, and the power of FIGS. 13 and 14 explained in the second embodiment is added to this. Post filter 1 with reset signal by calculation and level monitoring
Use of 0 is controlled.

Embodiment 6 FIG. In the present embodiment, a data separation unit is added to the decoding device of the fourth embodiment. That is, the selection signal is received from the encoding device side without providing the determination unit for examining the reflection coefficient in the decoding device. The structure in this case is shown in FIG. In the figure, 1 to 5,1
0, 18, and 19 are the same as the elements of the decoding device of FIG. 25 shown in the fourth embodiment. 16 is a data separation unit. The operation of the encoding device of FIG. 29 is basically the same as that of the decoding device of FIG.
The received signal sent from the data separation unit 16 is separated into encoded data and filter information, and the extracted filter information as a selection signal of use / non-use of the post filter 10 is sent to the switch control unit A18 described above. , The switch 19 is controlled.

Embodiment 7 FIG. A speech coding apparatus according to another embodiment of the present invention will be described. The device of the present embodiment is
The reflection coefficient determination unit 17, the switch control unit B23, the switch 19, and the switch 25 are added to the encoding device that captures the pitch information as part of the encoded data and transmits the encoded information, as shown in the second conventional technique. This is an encoding device that can perform an encoding process when a DTMF signal that is a non-voice signal is input without passing through the auditory weighting filter 8 and the long-period prediction synthesis filter 14 that synthesizes a signal using pitch information. FIG. 30 shows a block diagram of the configuration. In the figure, reference numerals 1, 2, 7 to 9 and 11 to 15 are the same as those of the encoder shown in FIG. Reference numeral 17 is a reflection coefficient determination unit, 19 and 25 are switches, and 23 is a switch control unit B.

The operation of the coding apparatus shown in FIG. 30 is basically the same as the operation of the apparatus having the configuration shown in FIG. 37, and the mechanism used in the first embodiment described in detail above. When the DTMF signal is input, the coding process is performed without passing both the auditory weighting filter 8 and the long-period prediction synthesis filter 14.

Note that, instead of selecting whether or not to use the auditory weighting filter 8, a mechanism for selecting whether or not to use an initial value as a filter coefficient of the auditory weighting filter 8 may be used as a mechanism. The same as 1.

Eighth Embodiment The present embodiment is the same as the encoding device shown in FIG. 30 of the seventh embodiment and the signal level monitoring unit B.
21 and a switch control unit C24 are combined. Even in this way, when the DTMF signal which is a non-voice signal is input, it is possible to select the non-use of both the auditory weighting filter and the long period prediction synthesis filter. A configuration block diagram in this case is shown in FIG. In the figure,
1, 2, 7 to 9, and 11 to 17 are the same as the elements of the encoding device of FIG. Reference numerals 19 and 25 are switches, 21 is a signal level monitoring unit B, and 24 is a switch control unit C.

The operation of the encoder of FIG. 31 is basically the same as the operation of the encoder of the configuration of FIG. As described in detail above, the switch 19 is selected based on the combination of the reflection coefficient and the audio signal level, and when the DTMF signal is input, the auditory weighting filter 8 and the long period prediction synthesis filter 14 are not passed through the encoding process. Is to do.

In the above embodiment, the extraction of the reflection coefficient is L
It was performed in the PC analysis unit 11. The extraction of the reflection coefficient may be changed to the auditory weighting filter adaptor 7, and a block diagram of the configuration in this case is shown in FIG. The operation of the encoding apparatus of FIG. 32 is basically the same as the operation of the apparatus of the configuration of FIG. 30, and the reflection coefficient as a signal is obtained from the auditory weighting filter adaptive unit 7. Subsequent operations are the same as those of the above encoding device.

It is also possible to combine the devices having the configurations of FIGS. 31 and 32 with a slight change in the configuration. That is, in the configuration of FIG. 31, the extraction of the reflection coefficient in the LPC analysis unit 11 is changed to the auditory weighting filter adaptive unit 7. A configuration block diagram in this case is shown in FIG. FIG.
The operation of the coding apparatus of No. 3 is basically the same as the operation of the apparatus of the configuration of FIG. 31, but the reflection coefficient is obtained from the auditory weighting filter adaptive unit 7. After that, the operation is the same as that of the encoding device.

Embodiment 9 A speech decoding apparatus corresponding to the coding apparatus according to the seventh embodiment will be described. In the present embodiment, a switch control unit A18, a switch 19, a switch 25, and a data separation unit 16 are added to a decoding device that receives the pitch information shown in the second conventional technique from the encoding device side. The block diagram of the configuration is shown in FIG. In the figure, reference numerals 1, 2, 10, 13, and 14 are the same elements as those of the decoder of FIG. Reference numerals 19 and 25 are switches, and 23 is a switch control unit A.

The operation of the decoding apparatus of FIG. 34 is the same as that of the decoder of FIG. 38 in the decoding part, but the received coded data is separated by the data separation section 16 and the extracted long cycle prediction synthesis is performed. The switch control unit A18 receives the filter information, which is a selection signal indicating whether the filter 14 and the post filter 10 are used or not, and the switch 1 is operated based on the information.
9. The switch 24 is controlled to determine whether it is used or not, and a decoding process is performed. Also, instead of selecting whether to use the post filter 10 or not, a mechanism for selecting an initial value or a current value as the filter coefficient of the post filter 10 may be used, which is similar to the devices of other embodiments.

[0079]

As described above, according to the present invention, the reflection coefficient determining means and the use switching means are provided. Therefore, when a DTMF signal which is a non-voice signal is input, a audibility correction filter or a post filter, A signal that does not pass through the long-period predictive synthesis filter is selected to prevent distortion that occurs when the signal passes through these elements, thereby improving the transmission characteristics of the signal.

[Brief description of drawings]

FIG. 1 is a configuration block diagram of a speech coding apparatus according to Embodiment 1 of the present invention.

FIG. 2 is a flow chart of the Levinson-Durbin method.

FIG. 3 is a detailed configuration block diagram of a reflection coefficient determination unit according to the first embodiment of the present invention.

FIG. 4 is a detailed configuration block diagram of a switch control unit A according to the first embodiment of the present invention.

FIG. 5 is a configuration block diagram of another speech encoding apparatus according to the first embodiment.

FIG. 6 is a configuration block diagram of another speech encoding apparatus according to the first embodiment.

[Fig. 7] Fig. 7 is a configuration block diagram of a speech encoding apparatus in Embodiment 2 of the present invention.

8 is a detailed configuration block diagram of a signal level monitoring unit A of FIG.

9 is a detailed configuration block diagram of a switch control unit B in FIG.

[Fig. 10] Fig. 10 is a configuration block diagram of another speech encoding apparatus according to the second embodiment.

11 is a detailed configuration block diagram of a filter control unit A of FIG.

FIG. 12 is a configuration block diagram of another speech encoding apparatus according to the second embodiment.

13 is a detailed configuration block diagram of a signal level monitoring unit B of FIG.

14 is a detailed configuration block diagram of a switch control unit C in FIG.

FIG. 15 is a configuration block diagram of another speech encoding apparatus according to the second embodiment.

16 is a detailed configuration block diagram of a filter control unit B of FIG.

FIG. 17 is a configuration block diagram of another speech encoding apparatus according to the second embodiment.

FIG. 18 is a configuration block diagram of another speech encoding apparatus according to the second embodiment.

FIG. 19 is a configuration block diagram of a speech coding apparatus according to Embodiment 3 of the present invention.

[Fig. 20] Fig. 20 is a configuration block diagram of another speech encoding apparatus according to the third embodiment.

FIG. 21 is a configuration block diagram of another speech encoding apparatus according to the third embodiment.

[Fig. 22] Fig. 22 is a configuration block diagram of another speech encoding apparatus according to the third embodiment.

[Fig. 23] Fig. 23 is a configuration block diagram of another speech encoding device according to the third embodiment.

[Fig. 24] Fig. 24 is a configuration block diagram of another speech encoding device according to the third embodiment.

FIG. 25 is a configuration block diagram of a speech decoding apparatus according to Embodiment 4 of the present invention.

FIG. 26 is a configuration block diagram of another speech decoding apparatus according to the fourth embodiment.

FIG. 27 is a block diagram of a speech decoding apparatus according to Embodiment 5 of the present invention.

FIG. 28 is a block diagram of another speech decoding apparatus according to the fifth embodiment.

FIG. 29 is a block diagram of a speech decoding apparatus according to Embodiment 6 of the present invention.

[Fig. 30] Fig. 30 is a configuration block diagram of a speech coding apparatus in Embodiment 7 of the present invention.

[Fig. 31] Fig. 31 is a configuration block diagram of a speech coding apparatus in Embodiment 8 of the present invention.

[Fig. 32] Fig. 32 is a configuration block diagram of another audio encoding device according to the eighth embodiment.

[Fig. 33] Fig. 33 is a configuration block diagram of another speech encoding device according to the eighth embodiment.

[Fig. 34] Fig. 34 is a configuration block diagram of a speech decoding device in a ninth embodiment of the present invention.

FIG. 35 is a block diagram showing a first prior art speech encoder.

FIG. 36 is a block diagram showing a first prior art speech decoder.

FIG. 37 is a block diagram showing a second prior art speech coder.

FIG. 38 is a block diagram showing a second prior art speech decoder.

[Explanation of symbols]

1 excitation VQ codebook, 2 gain adjuster, 3 synthesis filter, 4 gain adaptor, 5 synthesis filter adaptor,
6 signal adder, 7 auditory weighting filter adaptor, 8
Auditory weighting filter, 9 least squares error calculator, 1
0 post filter, 11 LPC analysis unit, 12 data multiplexing unit, 13 short cycle prediction synthesis filter, 14 long cycle prediction synthesis filter, 15 signal addition unit, 16 data separation unit, 17 reflection coefficient determination unit, 17a reflection coefficient distribution unit, 17b k1 determination unit, 17ck2 determination unit, 17d
Comprehensive determination unit, 18 switch control unit A, 18a signal determination unit, 18b switch control signal transmission unit, 19 switch, 20 signal level monitoring unit A, 20a buffer A,
20b Buffer B, 20c Power calculation unit A, 20d
Power calculation unit B, 20e Signal power comparison unit, 21
Signal level monitor B, 21a buffer A, 21b buffer B, 21c power calculator A, 21d power calculator B, 21e level monitor, 21f signal power comparator, 21g reset signal transmitter, 22 data multiplexer, 23 switch Control unit B, 23a Reflection coefficient monitoring unit, 23b Logical sum calculation unit, 23c Reset signal sending unit, 23d Signal holding unit, 23e Switch control signal sending unit, 24 Switch control unit C, 24a Logical sum calculating unit, 24b Signal holding unit , 24c switch control signal transmitter, 25 switch, 100 signal line, 101 signal line, 102 signal line, 103 signal line, 104 signal line, 105 signal line.

Claims (13)

[Claims] 1. A perceptual weighting filter for increasing a perceptual quality to change a frequency characteristic by obtaining a filter coefficient showing a characteristic of the signal in encoding a signal in a voice band, and a reflection obtained by analyzing an input signal. A reflection coefficient determining means for comparing the coefficient with a set value, and a use switching means for using the auditory weighting filter if the reflection coefficient is out of a specified range and not using the auditory weighting filter for within the specified range. A provided audio encoding device. 2. The speech coding apparatus according to claim 1, wherein the reflection coefficient is obtained by an auditory weighting filter adaptation section that obtains an auditory weighting filter coefficient indicating a characteristic of the input signal. 3. The speech coding apparatus according to claim 1, wherein the reflection coefficient is obtained by a synthesis filter adaptation unit that obtains a coefficient when a vector candidate from the codebook is passed through the synthesis filter. 4. A signal level monitoring means for monitoring a change in the level of the input signal is added, and the use switching means switches the use of the auditory weighting filter by a combination of the reflection coefficient and the change in the level of the input signal. The speech coding apparatus according to claim 1, wherein: 5. The voice according to claim 1, further comprising data multiplexing means for multiplexing a reflection coefficient obtained by analyzing an input signal and a vector code selected from a codebook for transmission. Encoding device. 6. The use switching means switches between use and non-use of the auditory weighting filter and switches between a current value and an initial value of the filter coefficient of the auditory weighting filter. The speech coding apparatus according to claim 1, wherein 7. A post filter for improving the perceptual quality of an output signal from a synthesis filter unit for decoding a code vector from a codebook corresponding to the reception code in decoding of a reception code, and the synthesis filter unit. When the reflection coefficient is outside the specified range, the post filter is used if the synthesis filter adaptation means for giving the coefficient, the reflection coefficient determination means for comparing the reflection coefficient obtained by the synthesis filter adaptation means with the set value, A voice decoding device provided with a use switching means that does not use the above post filter within the specified range. 8. A signal level monitoring means for monitoring a change in the level of the output signal from the synthesis filter section is added, and the use switching means switches the use of the post filter by a combination of the reflection coefficient and the change in the level of the output signal. The speech decoding apparatus according to claim 7, wherein the speech decoding apparatus is performed. 9. In place of the reflection coefficient determination means, a data separation means for demultiplexing the encoded data and the reflection coefficient from the reception encoded data is added, and the reflection coefficient of the reflection coefficient obtained by the separation by the data separation means is added. The speech decoding apparatus according to claim 7, wherein the use of the post filter is switched according to the value. 10. An LPC (Linear Prediction) is applied to the above-mentioned signal in the encoding of a voice band signal.
Coding) Reflection coefficient determination means for comparing the reflection coefficient obtained by the LPC analysis section with a set value, and perceptual weighting for weighting vector candidates based on the codebook with perceptual weighting filter coefficients obtained by analyzing the input signal. A filter, a long-period prediction synthesis filter that synthesizes a signal using pitch information, and a use switching unit that switches between use and non-use of the auditory weighting filter and the long-period prediction synthesis filter based on the result determined by the reflection coefficient determination unit. A speech coding apparatus equipped with. 11. A signal level monitoring means for monitoring the level of an input signal is added, and a result of the judgment made by the reflection coefficient judging means and the value of the voice level obtained by the signal level monitoring means are used for the auditory weighting filter. 11. Use and non-use can be switched.
A speech encoding device according to claim 1. 12. The speech coding apparatus according to claim 10, wherein the reflection coefficient is obtained from an auditory weighting filter adaptation section that obtains an auditory weighting filter coefficient instead of being obtained from the LPC analysis section. 13. Demultiplexing means for demultiplexing the received coded data and the reflection coefficient from a received signal in decoding the received code, and a synthesis filter section for decoding a code vector from a codebook corresponding to the received coded data. A speech decoding apparatus comprising a post filter for enhancing the perceptual quality of an output signal from the device and a use switching means for switching between use and non-use of the post filter by the reflection coefficient obtained by the data separating means.