JP2981044B2 - Digital automatic gain controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital automatic gain control (Automat) used for level adjustment in an audio signal processing device such as an audio mixing device or an audio storage device.
ic Gain Controll).

[0002]

2. Description of the Related Art FIG. 15 shows a conventional example for realizing a digital AGC device. Input signal x to be subjected to AGC
Is converted into a PCM code by an analog / digital converter (hereinafter, referred to as an A / D converter) 1 and input into the AGC loop AL of the digital AGC system in the form of a PCM code. It has been made multiplies the a _g a AGC output y. That is, y = A _g × x. The output y is branched and input to the AGC coefficient generator 5 via the level detection circuit 3 and the averaging circuit 4. Thus AGC coefficient A _g, by round the AGC loop AL, determined for each output y.

FIG. 16 shows an example of a hardware realizing the AGC loop AL in FIG. This figure is the Japanese Patent Publication No. 63-1908.
No. 9 is disclosed. However, in FIG. 16, the averaging circuit 4 needs to be partially corrected. Delay circuit T
_If the input of ₁ is not taken from the output of the multiplier 43, the averaging circuit 4 cannot average the continuous signal sequence. Therefore, FIG. 16 is modified as shown in FIG.
The operation of FIG. 17 will be described in detail.

[0004] An input audio signal is provided to the circuit in the form of a PCM code. PCM input signal x is first inputted to a multiplier 2, wherein, in the multiplications between the AGC coefficient A _g AG
A C output signal y is calculated. The output signal y also branches into the AGC loop and is applied to the level detection circuit 3. That is, the signal is input to the squaring circuit 31 and the adder 32.
However, it is input to the adder 32 as a negative input. Are separately reference value D _r is applied to the adder 32,
As negative input when the output signal y exceeds D _r, when y is below D _r in the reverse applied to the averaging circuit 4 as a positive input.

[0005] In the averaging circuit 4, first, a constant weight (weight coefficient a) is given by the multiplier 41. The output of the multiplier 41 is input to the subsequently adder 42, the addition of the value stored in the delay circuit T ₁ is performed. Then adder 4
The output of the output of the two multipliers 43 in it is multiplied by a weighting factor b becomes the output of the averaging circuit 4 simultaneously multiplier 43 is stored in the delay circuit T _1. At this time, b for the multiplier 43 is a weighting coefficient whose range is 0 <b <1. The delay time of the delay circuit T ₁ takes equal to the sample period. The relationship between the input signal A _in (i) and the output signal A _out (i) for the i-th sample timing of the averaging circuit 4 is represented by the following equation.

[0006]

(Equation 1)

Therefore, if an average output is obtained from the input of the past m samples, the output A _out (i) is calculated by the following equation.

[0008]

(Equation 2)

Thereafter, the addition is performed by the adder 51 of the AGC coefficient generating section, and the AGC coefficient _Ag is determined and generated. Here, also according to 1.0 weight for the adder 51
In the present invention, it is not necessary to limit to this value.
Absent.

In FIG. 15, the input signal is set to PCM
Although the A / D converter 1 is used for encoding, it is possible to perform AGC if the input signal is a PCM code, and the A / D converter 1 is not always necessary.

[0011]

A conventional digital type A
The AGC loop in the GC system is configured as shown in FIG. 16. If this is used for adjusting the level of an audio signal, the following problem appears. That is, since the PCM code of the sound to be subjected to the level control has a property that its amplitude range is wide and fluctuates greatly in a short time, it is calculated even if it is averaged according to the configuration of FIG. it is inevitable that the gain factor a _g causes a slight change in a short time. However, for voices, if the volume is frequently compensated too much, there is a problem that an unnatural feeling is felt and a natural conversation cannot be realized.

Therefore, when the average value of the input signal fluctuates within a certain range, it is desirable that the gain coefficient value be as constant as possible ignoring the fluctuation. However, in the conventional AGC circuit, when the average value of the input signal fluctuates, the gain value calculated based on the fluctuation also fluctuates, and it is necessary to keep the value at a stable value. There was a problem that it was not possible.

Further, since the sound signal includes both the sound time and the silence time, if the level control of the silence time is controlled to be the same level as the sound time, the sound time is reduced. In most cases, when the output signal is switched to the sound period, the gain coefficient value is excessively large, and the output is saturated and distorted.

Therefore, the present inventors have already applied for an invention that the gain coefficient value is not changed when the average value of the input signal does not fluctuate beyond a certain value. This is to solve the problem in that a gain control circuit is provided with a quantization circuit having a hysteresis characteristic to stabilize a gain coefficient value and does not excessively react to a minute change in an average value of an input signal. Because there is no coping when the input level changes suddenly, if the input level suddenly increases or decreases, the user will feel that the artificial coefficient is suddenly received when updating the gain coefficient. There's a problem.

If the variation of the average value of the input signal falls within a certain range and the difference between the output target value and the actual output value is close to the discrimination value for a predetermined time, the quantization at that time is used. The value performs the same quantization as the value that has a difference almost equal to the quantization width with respect to the input to the actual quantizer,
There is a problem that outputting this quantized value cannot be said to be performing optimal quantization.

The present invention has been made in order to solve the above-described problems, and is intended for an input such as an audio signal whose amplitude range is wide and whose value greatly changes in a short time. A digital automatic gain control device that can perform a natural level adjustment that does not make the user feel that the dynamic gain control is being performed, stabilizes the gain coefficient, and optimizes the coefficient value. The purpose is to gain.

[0017]

A digital automatic gain control device according to a first aspect of the present invention is characterized in that a digital automatic gain control device for controlling the output level of an audio signal, which is characterized by a wide input level variation range. In the control device, for example, a digital type automatic gain control loop constituting the digital type automatic gain control device is provided with a quantizer having a hysteresis characteristic, and further provided with a fade-in circuit to provide a suddenly exceeding a quantization identification value. If the input level suddenly increases for a large input change, the gain coefficient is quickly reduced to a value that does not cause saturation distortion of the output signal at the next gain update, and conversely, if the input decreases The value to be amplified at the time of one gain update is determined in advance, and the gain coefficient value is increased stepwise over several times.

A digital automatic gain control device according to a second aspect of the present invention is a digital automatic gain control device for controlling the output level of an audio signal, which is characterized in that the input level varies widely. Equipped with a quantizer with hysteresis characteristics in the digital automatic gain control loop that composes the automatic gain control device, and quantizes the difference between the output target value and the actual output, while changing the quantization value. When the same quantized value is output for a predetermined time or more from the time when the quantization is performed, the quantization value and the identification value are reset based on the average of the input values to the quantization circuit for the predetermined time.

A digital automatic gain control device according to a third aspect of the present invention is provided with a voice detector in the digital automatic gain control device having the above-described configuration. It is characterized in that it is performed using an audio signal input during a sound period.

[0020]

In the digital automatic gain control device according to the first aspect of the present invention, the quantization of the average value of the difference between the output target value and the actual output in the digital automatic gain control loop is performed by a quantization circuit having a hysteresis characteristic. Therefore, even if the difference between the output target value and the actual output average value fluctuates near the discrimination value due to the fluctuation of the input signal level, the quantization value is not changed one by one.

In the digital automatic gain control device according to the second invention, when the same quantized value is continuously output for a predetermined time or more after the quantized value is changed, the input to the quantizer after the predetermined time is output. Since the discrimination value and the quantization value are reset on the basis of this, the gain coefficient takes the optimum value corresponding to the fluctuation of the average value of the input signal, and the discrimination value is separated from the actual output at equal intervals up and down. Becomes more stable.

The digital automatic gain control device according to the third aspect of the present invention uses a voice detector to check for the presence or absence of voice, and performs gain control only by referring to the signal level of the voiced time, thereby providing a silent time. In, the calculation of the quantization level of the noise amount is not performed, the corresponding gain coefficient is not generated, and the noise amount can be amplified by an excessive gain in the silence time. There is no possibility that the sound signal is amplified by the excessive gain calculated in the silent period of the above and no saturation distortion occurs.

[0023]

[Embodiment 1] FIG. 1 shows an embodiment of a digital AGC apparatus using the present invention. Reference numerals 2, 3 and 4 are the same as those of the conventional apparatus shown in FIGS. 15, 16 and 17, respectively. A quantization circuit 6 and a fade-in circuit 7 are newly provided in the AGC loop.

FIG. 2 is a block diagram of the quantization circuit 6, which comprises the following. Reference numerals 61 and 62 denote quantizers that realize quantization having a hysteresis characteristic using the two. 63 is a subtractor for subtracting the output value of the pre-quantization circuit from the input signal level, 64 is a code extractor, 65 is a selector circuit, 66 is an inverse quantizer, and T ₂ is a memory for storing the previous output value of the quantization circuit. Is a delay circuit that plays the role of.

FIG. 3 is a block diagram of the fade-in circuit 7. Reference numeral 71 denotes a selector circuit, and reference numeral 72 denotes a comparator which compares an input of the fade-in circuit with an output value obtained from the input by a predetermined sampling time T in the past. . T ₃ serves as a memory, and a delay circuit for delaying by a predetermined sample time T;
Reference numeral 3 denotes a step intensifier for adding a preset value. The fade-in circuit 7 has a configuration in which an output signal of the quantization circuit 6 is input and the output is applied to the AGC coefficient generator 5.

Next, the operation will be described in detail. The digital AGC apparatus according to the present invention inputs an input signal x as a PCM code to an AGC loop as shown in FIG. 1 and sequentially performs digital processing, similarly to the conventional digital AGC system. . Until a signal is output from the level detection circuit 3, a conventional example shown in FIG.
7 is performed in the same manner as described above. The difference between the output target value and the actual output is calculated by the averaging circuit 4 as in the conventional example.
The values output from the averaging circuit 4 are successively input to the quantization circuit 6, where the values are quantized, and then processed by the fade-in circuit 7.

The operation of the quantization circuit 6 will now be described with reference to the drawings. FIG. 4 shows the quantization characteristics of the quantization circuit used in the present invention. For convenience, the signals are sequentially inputted to the quantization circuit L _n, and _{L n + 1, L n +} 2, the output result of this O _n, and _{O n + 1, O n +} 2.

Now, if the input signal is L _n = 3.27, L _{n + 1}
= 4.30, L _{n + 2} = 3.83, the input L _{n is} subjected to quantization corresponding to the point A in FIG. 4, and the output O _n = 3. Next, when L _{n + 1} = 4.30 is input,
Output characteristics, respectively, in FIG 5 (a), is input to the quantizer 61 and 62 as shown in (b), is from the quantizer 61 "4", from the quantizer 62 "4" The value is output. At the same time, the input L _{n + 1} is also input to the subtractor 63.

On the other hand, the output O _n = 3 of the quantization circuit 6 is stored in the delay circuit T ₂ , and the input L
_{n + 1} and the subtraction between the output of the delay circuit T ₂ 4.30-3 =
1.30. Therefore, the code extractor 64 extracts a positive code. The selector circuit 65 is connected to the quantizer 61 when the sign is positive, and is connected to the quantizer 62 when the sign is negative.
Is to be connected. Now that a positive code has been input, it is connected to the quantizer 61, and quantization corresponding to the point B in FIG. 4 is performed, and On _{+ 1} = 4. This value is simultaneously inputted to the AGC coefficient generator also input to the inverse quantization circuit 66 is input to the delay circuit T ₂ after inverse quantization is performed. Here, for simplicity, quantization and inverse quantization have slope 1
Is assumed to be linear quantization. Therefore, the delay circuit T ₂ "4" is stored.

Next, when L _{n + 2} is input, the quantizer 61
Outputs “3”, and the quantizer 62 outputs “4”. Operation 3.83-4 = −0.17 in the subtractor 63
Is performed and the sign becomes negative, the signal is connected to the quantizer 62 in the selector circuit 65, and is quantized corresponding to the point C in FIG. 4, and the output On _{+ 2} = 4 of the quantization circuit 6 is obtained.

The above is summarized as follows. Input Ln = 3.27 Ln + 1 = 4.30 L
n + 2 = 3.83 output On = 3 On + 1 = 4 O
n + 2 = 4

This example shows that once the output becomes "4", the output holds "4" even if the input becomes 3.83. That is, in order for the output to change from "3" to "4", the input value must be equal to or greater than the identification value 4.20 according to the upward arrow in FIG. 4, and conversely, for the output to change from "4" to "3", According to the downward arrow, the input value must be equal to or less than the identification value 3.80.

If a quantization circuit having a hysteresis characteristic is used as described above, once the quantization value changes, the same value is output until a value exceeding a point having a reverse arrow is input. Even if the input value to the quantization circuit fluctuates near the discrimination value, a constant and stable quantization value can be obtained. As a result, a constant and stable gain coefficient can be generated based on the quantized value.

Further, the same quantization value is output as compared with the case where the identification value is fixed, and the control of the gain coefficient value is stabilized.

Next, the operation of the fade-in circuit 7 will be described with reference to FIGS. Assuming that the value quantized by the quantization circuit 6 is Q _(t) , the quantized value Q _(t) is first input to the selector circuit 71 and the comparator 72. An output value Q _(tT) obtained in the past for a predetermined sample time T of the fade-in circuit 7 is simultaneously added to the comparator 72. The two values are compared, and an instruction corresponding to FIG. Give to 71. That is, when Q _(t) -Q _(tT) <0, it is considered that a large value is suddenly input, and the selector circuit 71 selects the contact point (a) in order to prevent the output from causing a saturation distortion. When Q _(t) -Q _(tT) = 0, it is considered that the fluctuation of the input level is within the dead zone. At this time, the selector circuit 71 also selects the contact (a). When Q _(t) −Q _(tT) > 0, it is considered that a small value is suddenly input, and the selector circuit 71 selects the contact (b) and changes the contact width to Q _(tT) by a preset step width S. The sum is used as the output.

For example, if Q _(t) = 4 and Q _(tT) = 3, then Q _(t) -Q _(tT) > 0, and the selector circuit selects the contact (b). Here, the step intensifier 73 sets S =
Assuming that the output is set to increase at a step width of 0.2, the output is 3.2, 3..
4, 3.6, 3.8, and 4.0 in this order.

If it is assumed here that Q _(t) = 4 is directly input to the AGC coefficient generator 5 without such step increase, Q _(tT) = 3 to Q _(t) = 4 , The gain coefficient increases, and as a result, the output level sharply increases, and it is inevitable that the level adjustment has been performed artificially. Therefore, as shown in FIG. 6C, when the input level fluctuates in a decreasing direction,
By increasing the output stepwise by the fade-in circuit, it is possible to have a natural conversation without feeling excessive or insufficient volume in a telephone line where an audio signal is input, for example.

As described above, when the input level suddenly increases, the gain coefficient rapidly decreases to a value at which the output signal does not cause saturation distortion. On the contrary, when the input level fluctuates in the decreasing direction, When the voice signal is input by gradually increasing the gain coefficient according to the operation of the fade-in circuit without abruptly increasing the gain coefficient, it is possible to have a natural conversation without feeling excessive or insufficient volume. I can do it.

As described above, the value output from the fade-in circuit 7 is continuously input to the AGC coefficient generation circuit 5 to generate a gain coefficient through a predetermined process and adjust the level of the input signal x. As a result, for an excessive input, the gain is quickly reduced to eliminate the saturation distortion, and when the input level decreases, the gain coefficient is updated in a stepwise manner at every predetermined sampling time T, so that the fade-in effect is exhibited. In addition, comfortable voice communication can be performed without feeling that an artificial level adjustment has been performed.

As described above, in the above-described embodiment, in the digital automatic gain control device which performs a predetermined automatic gain control using the PCM code as an input, the digital automatic gain control loop constituting the digital automatic gain control device is included in the digital automatic gain control loop. Equipped with a quantization circuit that quantizes the difference between the target value and the actual output,
Even if the input level of the quantization circuit fluctuates near the quantization identification value by giving a hysteresis characteristic to this,
Unless the fluctuation width is larger than a predetermined value, the same value is output, and in a control method in which the gain value is not corrected excessively frequently, when the input level is rapidly increased, the input level is quickly responded to. Modify the gain factor value to the value
A digital automatic gain control device characterized in that the gain coefficient is increased stepwise by a predetermined value even if the input level is rapidly reduced, and the gain coefficient is corrected in a plurality of times to perform a fade-in operation. Was explained.

Embodiment 2 FIG. In the above embodiment, FIG.
The quantization based on the input / output conversion in accordance with the straight line Y = X-0.5 shown in FIG. 7A is set as shown in FIG. 7B, and the input / output characteristics of the quantization circuit 6 are set to realize this as stable as possible. As shown in FIG. 7C, when the input / output conversion follows an arbitrary curve or straight line, the following is performed. For example, when the input / output conversion follows the curve as shown in FIG. 8A, the input / output characteristics of the quantization circuit are as shown in FIG. 8B, and the characteristics of the inverse quantizer are as shown in FIG. If this is the case, the same effects as in the above embodiment can be obtained.

Embodiment 3 FIG. Further, in the above embodiment, a feedback system in which control is performed based on an output signal is employed, but the same effect can be obtained by using a feedforward system as shown in FIG. In FIG. 9, 33 is an absolute value circuit, and 44 is a peak hold circuit.

Embodiment 4 FIG. Further, the same effect as that of the above embodiment can be obtained even if the automatic gain control device of the present invention is executed by software processing centering on a processor.

Embodiment 5 FIG. In the above embodiment, the quantization circuit 6 having the hysteresis characteristic has been described. However, the quantization circuit may not have the hysteresis characteristic. For example, a quantization circuit for performing simple quantization as shown in FIG. 7B or 8A may be used. This is because the identification value exists as long as the quantization is performed, and when an input signal that straddles the identification value is input, the fade-in circuit 7 described in the present invention can be applied.

Embodiment 6 FIG. In the above embodiment, the squaring circuit 31 is used for calculating the level. However, the level detection may be performed using an absolute value circuit, and the averaging may be performed by the averaging circuit 4. The same effect as the example can be obtained.

Embodiment 7 FIG. Hereinafter, an embodiment of the second invention will be described with reference to the drawings. FIG. 10 shows an embodiment of a digital AGC apparatus according to the present invention. In the figure, 2,
Reference numerals 3 and 4 are the same as those of the conventional device shown in FIGS. 15 and 16, respectively.
It is provided with a quantization circuit 6, a voice detector 9 and a selector circuit 8.

Here, the quantization circuit 6 shown in FIG.
The configuration shown in FIG. In FIG. 11, reference numerals 61 and 62 denote quantizers which realize quantization having a hysteresis characteristic by using these two. 63 is a subtractor for subtracting the output value of the pre-quantization circuit from the input signal level, 64 is a code extractor, 65 is a selector circuit, 66 is an inverse quantizer, and T ₂ is a memory for storing the previous output value of the quantization circuit. A delay circuit serving as
Reference numeral 68 denotes a selector circuit which receives the outputs of the selector circuit 65 and the averaging circuit 4 as inputs.

The AGC coefficient generator 5 shown in FIG.
Has the configuration shown in FIG. In FIG.
51 is an adder, 52 is a selector circuit, 53 is a comparator which compares output values at three consecutive times and gives an instruction to each of the selector circuits 52, 67 and 68 based on the result.
54 and 55 delay circuit for delaying storing audio signals by a predetermined sampling time T ₃ and T ₄ 56 is a step increase or decrease device for adding a preset value.

In FIG. 10, the voice detector 9 receives the output signal y of the digital AGC loop as an input, and the output is applied to the selector circuit 8. The selector circuit 8 is equipped between the averaging circuit 4 and level detecting circuit 3, the output of the level detecting circuit 3 in one terminal, the output of the delay circuit T ₂ are the other terminal is connected. This delay circuit T
As _second delay time is equal to the sample period, the output of the delay circuit T ₂ are a front output of the selector circuit 8.

[0050] The quantization characteristic of the quantization circuit 6 shown in FIG. 11 is similar to that shown in Figure 4 as described in Example 1, description thereof is omitted here.

In FIG. 10, the presence / absence of voice is checked using the voice detector 9 and the gain control is performed only by referring to the signal level in the voiced time. Therefore, the quantization level of the noise amount is calculated in the voiceless time. Is not performed, the corresponding gain factor is not generated, and the amount of noise can be amplified by an excessive gain in silence time, or the excessive noise calculated in the silence time immediately before switching to the sound time can be obtained. There is no case where the sound signal is amplified by the gain to cause saturation distortion.

Next, the operation will be described in detail. The digital AGC apparatus according to the present embodiment inputs an input signal x as a PCM code to an AGC loop as shown in FIG. 10 and sequentially performs digital processing, similarly to the conventional digital AGC system. is there. Until the level detection circuit 3 outputs a signal, the same processing as that described with reference to FIG. 17 as a conventional example is performed.

The output signal y of the digital AGC loop is input to the voice detection circuit 9, and the detection result is sent to the selector circuit 8. The selector circuit 8 receives the output signal of the level detection circuit 3 and the previous output value of the selector circuit 8 in the delay circuit T.
_When the sound detection result is sound, the terminal C is connected to the terminal C. When the sound detection result is no sound, the terminal C is connected to the terminal D. That is, the output signal of the level detection circuit 3 is selected as the output signal of the selector circuit 8 again when the call voice is sound, and the previous output value of the selector circuit 8 is selected again when the call voice is silent. Subsequently, the output of the selector circuit 8 is input to the delay circuit T ₂ and the averaging circuit 4. The difference between the output target value and the actual output is averaged by conventional as well as averaging circuit 4, by subsequently processing the input Figure 1 1 to the quantization circuit 6 having the input-output characteristics as shown in FIG. 4 Quantization is performed.

However, it is assumed that the state in which the same quantized value is output after the quantized value is changed continues for a predetermined time or more, that is, the determination formula of the comparator 53 in the AGC coefficient generator 5 is shown in FIG. In the case corresponding to the state 4 shown in the figure, the switch of the selector circuit 67 is connected to the terminal ii to realize the gain control operation corresponding to the state 4, and the averaging circuit 4
Output L _{n + 2} is selected instead of the O _{n + 2} as the output of the quantization circuit 6. At the same time, the switch b of the selector circuit 68 is connected, and the identification values and the quantization values of the quantizers 61 and 62 are reset based on the output value of the averaging circuit 4.

After the quantization process has been performed as described above, the output is subsequently input to the AGC coefficient generator 5 shown in FIG. 12 and first to the selector circuit 52 and the comparator 53. At the same time, two gain values Q _(tT) and Q _(t−2T) output in the past by the predetermined sample time T of the output value of the AGC coefficient generator 5 by the delay circuits T ₃ and T ₄ are output to the comparator 53. Has been entered. The three values are compared, and an instruction corresponding to the determination formula is given to the selector circuit 52 and the selector circuits 67 and 68 in the quantization circuit 6 according to FIG. That is, when Q and Q _(tT) are equal, it is considered that the input level is in a stable state, and the selector circuit 52 is connected to the connection terminal A so that Q _(t) = Q (= Q _{( tT )} ).
Is different from Q _{( tT )} , a connection terminal B is connected, and a value obtained by increasing or decreasing the previous output Q _{( tT )} by a step width S preset in the step _adjuster 56 is defined as Q _(t) .

For the selector circuit 67, Q = Q _{( tT )}
Only when ≠ Q _{( t−2T )} , is connected to terminal ii, and Q _(t) is the output of averaging circuit 4. Also, the selector circuit 68 is connected to b only when Q = Q _{( tT )} ≠ Q _{( t−2T )} and becomes the output of the averaging circuit 4. As described above, the gain coefficient becomes the most suitable value by monitoring the fluctuation state of the input level and resetting the identification value and the quantization value of the quantization circuit when a certain condition is satisfied. Is located at the center of the pair of upper and lower identification values, so that the control can be stabilized.

Embodiment 8 FIG. Although the squaring circuit 31 is employed in the seventh embodiment for calculating the level, the level detection is performed using an absolute value circuit and the averaging circuit 4 performs averaging instead. The same effects as in the embodiment can be obtained.

Embodiment 9 FIG. Also in the seventh embodiment, FIG.
The quantization based on the input / output conversion according to the straight line Y = X-0.5 shown in FIG. 7A is set as shown in FIG. 7B, and the input / output of the quantization circuit 6 is implemented in order to stabilize the quantization as much as possible. Although the characteristics are as shown in FIG. 7C, when the input / output conversion follows an arbitrary curve or straight line, the following is performed.

For example, when the input / output conversion follows a curve as shown in FIG. 8A, the input / output characteristics of the quantizer are as shown in FIG.
If the characteristics of the inverse quantizer are as shown in FIG. 8C, the same effect as in the above embodiment can be obtained.

Embodiment 10 FIG. Further, the present invention adopts the feedback scheme for controlling the output signal, as shown in FIG. 1 4, raised similar effects by using a feed-forward system.
1 4, 33 the absolute value circuit, 44 is a peak hold circuit.

Embodiment 11 FIG. Further, the same effect as that of the above embodiment can be obtained even if the automatic gain control device of the present invention is executed by software processing centering on a processor.

[0062]

As described above, according to the first aspect, the AG
By equipping the C-loop with a quantization circuit with hysteresis characteristics, even if a small change in the signal sequentially input to the quantization circuit occurs near the identification value, the identification once the quantization output changes. Unless a value exceeding the other identification value that forms a pair with the value is input, the same value continues to be output from the quantization circuit without returning to the original quantized output, so that the gain coefficient calculated based on this quantized value And stable control can be performed. Also, when the input level increases rapidly, the gain coefficient is rapidly reduced to avoid causing output saturation distortion, and conversely, when the input level decreases, the gain coefficient is gradually increased in steps. By doing so, there is an effect that a natural conversation can be performed without feeling excessive or insufficient volume on a telephone line where a voice signal or the like is input, for example.

According to the second aspect of the present invention, by providing a quantization circuit having a hysteresis characteristic in the AGC loop, a minute change in a signal sequentially input to the quantization circuit is reduced in the vicinity of the identification value. Even if it occurs, once the quantization output changes, the same value continues to be output from the quantization circuit without returning to the original quantization output unless a value exceeding the other identification value that pairs with the identification value is input, unless it is input. In addition, it is possible to prevent the gain coefficient calculated based on the quantization value from excessively fluctuating, and to perform stable control. However Zhou as the difference between the quantized value very close value to the identification value input to the quantizer is more equal to the quantization width
However , according to the present invention, when the same quantization value is output for a predetermined time or longer, the identification value and the quantization value are reset with, for example, the output value of the averaging circuit. The gain coefficient can be re- determined and the control can be made more stable.

According to the third aspect of the present invention, the presence / absence of voice of the input signal is detected by the voice detector, and the immediately preceding voiced signal level is used instead of the input signal during the silent period.
Even if the state is switched from silence to speech, smooth gain control can be performed, and it is possible to have a natural conversation without feeling excessive or insufficient volume on a telephone line where an audio signal is input, for example. Effective.

[Brief description of the drawings]

FIG. 1 is a diagram showing an embodiment of a digital AGC apparatus using the present invention.

FIG. 2 is a diagram showing a configuration example of a quantization circuit 6 provided in an embodiment of a digital AGC apparatus using the present invention.

FIG. 3 is a diagram showing a configuration example of a fade-in circuit 7 provided in an embodiment of a digital AGC device using the present invention.

FIG. 4 is a diagram showing quantization characteristics of a quantization circuit 6 provided in an embodiment of a digital AGC device using the present invention.

FIG. 5 is a diagram showing input / output characteristics of two types of quantizers 61 and 62 included in a quantization circuit 6 provided in an embodiment of a digital AGC device using the present invention.

6 is a diagram showing a determination formula of a selector circuit 71 in the fade-in circuit 7 shown in FIG.

FIGS. 7A and 7B show input / output characteristics in the quantization circuit 6 of the present embodiment, FIG. 7A shows input / output conversion characteristics, FIG.
(C) is a diagram showing input / output characteristics of the circuit.

FIG. 8 is a diagram showing a quantization characteristic (b) and an inverse quantization characteristic (c) when the input / output conversion is set to an arbitrary curve (a).

FIG. 9 is a diagram showing one embodiment of the present invention using a feedforward system.

FIG. 10 is a configuration diagram showing one embodiment of a digital AGC device according to the present invention.

11 is a configuration diagram of the quantization circuit 6 of FIG.

FIG. 12 is a configuration diagram of an AGC coefficient generator 5 of FIG. 10;

FIG. 13 is a diagram showing a determination formula of a comparator 53 in the AGC coefficient generator 5 of FIG.

FIG. 14 is a configuration diagram showing another embodiment of the present invention using a feedforward system.

FIG. 15 is a configuration diagram of a conventional example for realizing a digital AGC.

FIG. 16 is a block diagram showing a conventional example in terms of hardware.

FIG. 17 is a block diagram partially modified from the conventional example shown in FIG. 16;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 A / D converter 2 Multiplier 3 Level detection circuit 4 Averaging circuit 5 AGC coefficient generation part 6 Quantization circuit 7 Fade-in circuit 8 Selector circuit 9 Audio detector 31 Square circuit for power extraction 32, 42, 51, 63 Adders 41, 43 Multipliers for weighting 52, 65, 67, 71 Selector circuit 53 Comparator54 Delay circuit T ₁ 55 delay circuit T _Two  56 step changer 61, 62 quantizer 64 code extractor 66 inverse quantizer 68 switch 72 comparator, 73 step increaser x input signal y output signal T₁ , T_Two , T_Three , T_Four Delay circuit L_n , L_{n + 1} , L_{n + 2} Continuous input to the quantization circuit 6
Signal level O_n , O_{n + 1} , O_{n + 2} L of the quantization circuit 6_n , L_{n + 1} ,
L_{n + 2} Quantization value for

Claims (3)

(57) [Claims] An input signal is input to an automatic gain control loop.
And the actual output from the automatic gain control loop is
Use the difference between the output target value and the actual output to achieve the output target value.
-Type automatic gain control device with automatic gain control
In the automatic gain control loop, the difference between the output target value and the actual output is input, and the input output target
Quantizes the difference between the value and the actual output to determine the quantized value and outputs it
And the difference between the input output target value and the actual output
If it is within the specified range, use the hysteresis characteristics.
Output the quantization value as a constant value
And the quantization value from the quantization circuit, and the level of the input signal is
When the number of signals increases, the quantum output by the quantization circuit
Value to the value at which the actual output does not cause saturation distortion.
If the level of the force signal decreases, the quantization value is changed to a predetermined value.
Fade-in times that are increased step by step and corrected
Path, and automatically by the quantized value corrected by the fade-in circuit.
Digital automatic control characterized by dynamic gain control
Gain control device. 2. An input signal is input to an automatic gain control loop.
And the actual output from the automatic gain control loop is
Use the difference between the output target value and the actual output to achieve the output target value.
-Type automatic gain control device with automatic gain control
In location, the difference between the output target value and the actual output to the automatic gain control loop
And a quantization circuit that quantizes the difference between the output target value and the actual output.
The difference between the value and the actual output is quantized using a predetermined
Determines the output value and outputs the input target
If the difference between the value and the actual output is within a certain range,
The quantization value can be kept constant by using the
And a quantizer for outputting the quantized value for a predetermined time or more.
If the value is a constant value, the output
Identification value used by the quantizer based on the difference between the standard value and the actual output
Resetting means for resetting the quantization value again and
And an automatic gain according to the quantization value determined by the quantization circuit.
The digital automatic gain system which to perform the control and feature
Control device. 3. The digital automatic gain control device according to claim 1, further comprising: detecting a time when an audio signal is present in the signal as a sound time, and detecting a time when there is no audio signal in the signal as a silent time. In the sound detector, the difference between the output target value of the sound period and the actual output is selected in the sound period detected by the sound detector and output to the quantization circuit. the digital automatic gain control device comprising a kite <br/> a selector circuit difference selects the output to the quantization circuit between the output target value and the actual output of the sound detected time.