JP5857216B2 - Automatic gain controller - Google Patents

Description

  The present invention relates to an automatic gain control device that controls a gain when an input signal is amplified according to the level of the input signal.

  As a conventional automatic gain control device, for example, there is one described in Patent Document 1. The conventional example described in Patent Document 1 includes a buffer that temporarily stores an input signal, an energy determination device that calculates an energy value of the input signal, a gain determination device that calculates a gain based on the energy value, and a buffer. A gain device for amplifying a stored input signal with the gain.

  In this conventional example, the energy determination device determines whether the state of the input signal is the beginning of the message (voice), ongoing, end, and no signal (or ambient noise), and the gain determination device The gain according to each state is determined.

Japanese National Patent Publication No. 3-502278

  By the way, in the above conventional example, when the energy determination device determines the state of the input signal, the calculation of the energy value of the input signal requires an operation time of several tens of milliseconds to several hundreds of milliseconds. It is difficult to follow. For example, when the input signal (voice) suddenly increases, the output signal is excessively amplified because the gain of the gain device does not change until the energy determination device and the gain determination device follow the change of the input signal. There is a risk of it.

  The present invention has been made in view of the above problems, and an object of the present invention is to reduce the influence of fluctuations in the input signal on the output signal as compared with the conventional example.

  An automatic gain control device according to the present invention calculates a signal level calculation unit that calculates a signal level of an input signal, and a gain for amplifying the signal based on the signal level calculated by the signal level calculation unit. A gain calculation unit; an amplification unit that amplifies the signal with the gain calculated by the gain calculation unit; and the gain calculated by the gain calculation unit is equal to or greater than a predetermined threshold value greater than 1. A fluctuation determining unit that determines that the input signal has a tendency to increase when the absolute value of the signal level of the signal after being amplified by the unit is greater than or equal to a predetermined upper limit, and an increase in the fluctuation determining unit And a gain control unit configured to correct a gain when the amplification unit amplifies the signal to be equal to or less than the threshold value when it is determined that there is a tendency change.

In this automatic gain control device, the input signal is an acoustic signal, and the signal level calculation unit calculates the signal level of the acoustic signal by averaging the absolute value of the signal level of the acoustic signal over time. is configured, the a voice section detection unit which detects a speech section which contains speech component in an acoustic signal, the acoustic in a shorter time than the time at which the signal level calculating unit for averaging the absolute value of the signal level time A second signal level calculation unit that calculates the signal level by averaging the absolute value of the signal level of the signal over time , and the variation determination unit is a predetermined value that is less than 1 by the gain calculated by the gain calculation unit There is a state in which a voice segment is detected by the voice segment detector and the signal level calculated by the second signal level calculator is less than or equal to a predetermined threshold. When the sound signal continues for more than a time, it is determined that the acoustic signal has a decreasing tendency, and the gain control unit amplifies the signal when the fluctuation determining unit determines that there is a decreasing tendency fluctuation. It is preferable to correct the gain at the time to the second threshold value or more.

  In this automatic gain control device, it is preferable that the signal level calculation unit time-averages the absolute value of the signal level of the input signal.

  In the automatic gain control device, the signal level calculation unit and the second signal level calculation unit are configured to detect a signal level when the voice segment detection unit detects a voice segment, and the voice segment detection unit detects a voice segment. It is preferable to calculate the signal level of the acoustic signal from the difference from the signal level when not.

  In this automatic gain control device, the gain control unit maintains a gain at the time when the amplification unit amplifies the signal at a predetermined fixed value for a predetermined time when the variation determination unit determines that there is variation. It is preferable.

  In this automatic gain control apparatus, it is preferable that the gain control unit gradually decreases the gain when the amplification unit amplifies the signal over a certain time when the variation determination unit determines that there is variation. .

  In this automatic gain control device, it is preferable that the signal level calculation unit shortens the time when the time averaging is performed when the fluctuation determination unit determines that there is a fluctuation of an increasing tendency or a decreasing tendency.

  In this automatic gain control device, the input signal is an acoustic signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the signal level calculation unit includes the audio segment It is preferable to shorten the time for averaging the time from when the detection unit starts to detect a voice interval until a predetermined time elapses.

  In this automatic gain control device, the input signal is an acoustic signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the signal level calculation unit includes the audio segment It is preferable to calculate the signal level only in the voice section detected by the detection unit.

  In this automatic gain control device, the input signal is an acoustic signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the amplification unit includes the audio segment detection unit It is preferable that the gain is fixed to 1 in a non-speech interval that is not detected in step (1).

  In the automatic gain control apparatus, the gain calculation unit stores a plurality of types of values for the gain, and is selected from the plurality of types of values according to the signal level calculated by the signal level calculation unit. It is preferable to set the gain as the gain.

  In this automatic gain control device, the gain calculation unit stores a reference value of the signal level, and a value obtained by dividing the reference value by the signal level calculated by the signal level calculation unit is used as the gain. Is preferred.

  In this automatic gain control device, it is preferable that the respective units are initialized when a state in which no speech section is detected by the speech section detection unit exceeds a predetermined upper limit time.

  The automatic gain control device of the present invention has an effect that the influence of fluctuations in the input signal on the output signal can be reduced as compared with the conventional example.

It is a block diagram which shows Embodiment 1 of this invention. It is a wave form diagram for operation explanation same as the above. It is explanatory drawing for demonstrating operation | movement of the gain control part in the same as the above. It is a block diagram which shows Embodiment 2 of this invention. It is a wave form diagram for operation explanation same as the above. It is explanatory drawing for demonstrating operation | movement of the gain control part in the same as the above. It is explanatory drawing for demonstrating operation | movement of the gain control part in the same as the above. It is a wave form diagram for demonstrating operation | movement of the signal level calculation part in the same as the above. It is a wave form diagram for demonstrating operation | movement of the signal level calculation part in the same as the above. It is a wave form diagram for demonstrating operation | movement of the signal level calculation part in the same as the above. It is a block diagram which shows the modification same as the above.

  Hereinafter, an embodiment in which the technical idea of the present invention is applied to an automatic gain control device used in a system or device (for example, an interphone system or a telephone) that handles acoustic signals will be described in detail. However, the type of input signal is not limited to an acoustic signal, and the technical idea of the present invention can be applied to all automatic gain control devices that amplify signals other than acoustic signals. In each embodiment, an input signal converted into a digital signal is handled, but the input signal can be handled as an analog signal.

(Embodiment 1)
In the present embodiment, as shown in FIG. 1, in addition to an amplification unit 3 that amplifies an input acoustic signal (hereinafter referred to as an input signal), a signal level calculation unit 1, a gain calculation unit 2, a variation determination unit 4, A gain control unit 5 is provided.

  The signal level calculation unit 1 calculates the signal level of the input signal by averaging the absolute value of the signal level of the input signal over time. For example, when a sample value obtained by sampling an input signal at a constant sampling period is x (n) and the number of samples is M, a value calculated by the following Expression 1 or 2 is defined as a signal level of the input signal ( (See the two-dot broken line in FIG. 2). If the signal level L (n) of the input signal is calculated in this way, the amount of calculation in the signal level calculation unit 1 can be reduced.

  The gain calculation unit 2 calculates a gain G (gain) for amplifying the output signal to an appropriate signal level according to the signal level calculated by the signal level calculation unit 1. For example, the gain calculation unit 2 stores a plurality of types of values for the gain G (for example, five types of values G1> G2> G3> G4> G5), and the signal level calculated by the signal level calculation unit 1 Accordingly, an appropriate value may be selected from a plurality of types of values Gi (i = 1, 2, 3, 4, 5). In this way, it is possible to reduce the amount of calculation necessary for the gain calculation unit 2 to calculate the gain G.

  Alternatively, the gain calculation unit 2 stores the reference value Ls of the signal level, and a value obtained by dividing the reference value Ls by the signal level L (n) calculated by the signal level calculation unit 1 (= Ls / L (n) ) May be the gain G (see FIG. 3). In this way, the gain G can be calculated more finely according to the signal level L (n) of the input signal, compared with the case where the gain G is selected from a plurality of predetermined values Gi. it can.

  The amplifying unit 3 amplifies the input signal by multiplying the input signal by the gain G calculated by the gain calculating unit 2. That is, if the gain G is a value greater than 1, the signal level of the output signal is higher than the signal level of the input signal (the volume of the acoustic signal is large), and if the gain G is a value smaller than 1, The signal level of the output signal is lower than the signal level of the input signal (the volume of the acoustic signal is low). When the gain G is 1, the signal levels of the input signal and the output signal are equal.

  The fluctuation determination unit 4 determines whether or not the input signal has abrupt fluctuations (abrupt increase in signal level), and outputs the determination result (fluctuation determination flag) to the gain control unit 5. The fluctuation determination unit 4 increases if the gain G calculated by the gain calculation unit 1 is greater than or equal to a predetermined threshold Gu that is greater than 1 and the absolute value of the signal level of the output signal is greater than or equal to a predetermined upper limit value Xu. It is determined that there is a trend change. If the fluctuation determination unit 4 determines that the input signal has a fluctuation in an increasing tendency, it sets a fluctuation determination flag (sets the value of the fluctuation determination flag to 1). When it is determined that the input signal does not have an increasing tendency, the variation determination unit 4 does not set the variation determination flag (sets the value of the variation determination flag to 0).

  Here, the threshold value Gu of the gain G is desirably set to a value equivalent to a value necessary for amplification to a desired signal level (volume) when the signal level (volume) of the input signal is small. . The upper limit value Xu for the signal level of the output signal is preferably set to a value higher than the maximum value of the appropriate signal level (volume) of the output signal.

  The gain control unit 5 sends a gain control signal to the amplification unit 3 when the variation determination flag is set, and does not send a gain control signal to the amplification unit 3 when the variation determination flag is not set. This gain control signal instructs the amplifying unit 3 to correct the gain G to the threshold value Gu or less.

  The amplification unit 3 amplifies the input signal by the gain G calculated by the gain calculation unit 2 when the gain control signal is not received, and when the gain control signal is received, the gain (threshold value) indicated by the gain control signal is received. The input signal is amplified with a correction value less than or equal to Gu.

  Thus, according to the present embodiment, when the fluctuation determination unit 4 determines that there is a change in the increasing tendency, the gain control unit 5 corrects the gain G of the amplification unit 3 to be equal to or less than the threshold value Gu. Output of an excessively amplified output signal can be suppressed when the signal increases rapidly. As a result, it is possible to reduce the influence of fluctuations in the input signal on the output signal as compared with the conventional example.

(Embodiment 2)
The present embodiment has the same configuration as that of the first embodiment except that the second signal level calculation unit 6 and the voice section detection unit 7 are provided as shown in FIG.

  The second signal level calculation unit 6 calculates the signal level of the input signal in a time shorter than the time during which the signal level calculation unit 1 calculates the signal level L (n). For example, the number of samples M in Equation 1 or Equation 2 described in Embodiment 1 may be a numerical value that is smaller in the second signal level calculation unit 6 than in the signal level calculation unit 1.

  The voice section detection unit 7 detects a section (sound section) in which a speech component is included in the input signal (acoustic signal). When a voice section is detected, a voice section determination flag is set to detect the voice section. When there is no (non-speech segment), the speech segment determination flag is not raised. In this speech section detection unit 7, the ratio of the long-term average value of the input signal from several seconds to several tens of seconds and the short-time average value of the input signal from several hundred milliseconds to several seconds (short-time average value / long-time average value). ), And if the ratio is equal to or greater than a certain value, it is determined as a speech segment. However, since such a method for detecting a speech section is well known in the art, a detailed description thereof will be omitted. Further, the speech section detection unit 7 may detect the speech section by another detection method.

  Similar to the first embodiment, the fluctuation determination unit 4 determines whether or not the input signal has abrupt fluctuation (abrupt increase in signal level), and the determination result (fluctuation determination flag) is sent to the gain control unit 5. Output. Further, the fluctuation determination unit 4 in the present embodiment also determines whether or not the signal level has a sharp decrease (rapid fluctuation) in the input signal, and the determination result (fluctuation determination flag) is sent to the gain control unit 5. Output. For example, the fluctuation determination flag may be composed of 2 bits, the lower bit may be a flag indicating the presence or absence of a change in increasing tendency, and the upper bit may be a flag indicating the presence or absence of a fluctuation in decreasing tendency. That is, the fluctuation determination flag is “01” when it is determined that there is an increase tendency fluctuation, the fluctuation determination flag is “10” when it is determined that there is a decrease tendency fluctuation, and the fluctuation determination when it is determined that there is no fluctuation. The flag is “00”.

  The fluctuation determination unit 4 is a signal in which the gain G calculated by the gain calculation unit 1 is less than or equal to the second threshold value Gd that is less than 1, the voice segment determination flag is set, and the signal calculated by the second signal level calculation unit 6 If the state where the level is equal to or lower than the predetermined threshold value Lu continues for the predetermined time T1 or longer, it is determined that there is a change in the decreasing tendency. Note that the threshold value Lu is preferably set to a value smaller than the appropriate signal level (volume) of the input signal, and the time T1 is preferably set to about several hundred milliseconds. The second threshold Gd is desirably set to a value equivalent to a value necessary for amplification (attenuation) to a desired signal level (volume) when the signal level (volume) of the input signal is large. .

  The gain control unit 5 sets a gain control signal for correcting the gain G to be equal to or less than the threshold Gu when the value of the fluctuation determination flag is “01”, and sets the gain G when the value of the fluctuation determination flag is “10”. Control signals to be corrected to be equal to or greater than 2 threshold values Gd are respectively sent to the amplifying unit 3. Note that when the value of the variation determination flag is “00”, the gain control signal is not transmitted.

  The amplification unit 3 amplifies the input signal by the gain G calculated by the gain calculation unit 2 when the gain control signal is not received, and when the gain control signal is received, the gain (threshold value) indicated by the gain control signal is received. The input signal is amplified with a correction value equal to or less than the value Gu or a correction value equal to or greater than the second threshold Gd.

  Thus, according to the present embodiment, it is possible to suppress output of an excessively amplified output signal when the input signal rapidly increases as in the first embodiment. Further, in the present embodiment, when the fluctuation determining unit 4 determines that there is a fluctuation in the decreasing tendency, the gain control unit 5 corrects the gain G of the amplifying unit 3 to be equal to or higher than the second threshold value Gd. In this case, the gain G can be immediately increased and appropriately amplified without erroneously reducing (suppressing) the signal level. As a result, it is possible to reduce the influence of fluctuations in the input signal on the output signal as compared with the conventional example.

  Here, the signal level calculation unit 1 and the second signal level calculation unit 6 detect the signal level S when the speech segment detection unit 7 detects the speech segment and the speech segment detection unit 7 detects the speech segment. The signal level of the input signal may be calculated from the difference (S−N) from the signal level N when there is no signal (see FIG. 5). In this way, even when stationary noise is superimposed on the input signal, there is no need to change the upper limit value Xu for the output signal.

  Further, in the non-speech section that is not detected by the speech section detector 7, if the amplifying unit 3 fixes the gain G to 1, when the stationary noise is superimposed on the input signal, the noise component of the non-speech section is Amplification can be suppressed.

  Furthermore, it is preferable that the units 1 to 7 are initialized when the state in which the speech section is not detected by the speech section detection unit 7 exceeds a predetermined upper limit time (several tens of seconds to several minutes). That is, when the voice is not input beyond the upper limit time, it is possible to prevent the input signal from being amplified with an inappropriate gain in the next voice section by initializing the processing in the automatic gain control device.

  By the way, the gain control unit 5 preferably maintains the gain G of the amplification unit 3 at a predetermined fixed value (correction value) for a certain time T2 when the variation determination unit 4 determines that there is variation. For example, if the gain G before the change is G1, the gain G is set to the fixed value G3 (< G1) is fixed. After a certain time T2 has elapsed (after time t + T2), the value is changed to a value (for example, G4) corresponding to the signal level of the input signal (see FIG. 6). However, the fixed value G3 at this time is preferably set to 1. Moreover, it is preferable that the fixed time T2 is set to about the time until the signal level calculated by the signal level calculation unit 1 follows the input signal after the change. By doing so, the signal level of the input signal is equal to the signal level of the output signal until the input signal level calculation unit 1 follows the input signal after fluctuation, and the input signal is excessively increased by the amplification unit 3. Amplification can be avoided. Note that the gain control unit 5 does not immediately switch the gain G to the fixed value G3 when the variation determination unit 4 determines that there is a fluctuation in the increasing tendency, but is fixed over a fixed time T2 as shown in FIG. It may be gradually decreased to the value G3. If the gain G is gradually reduced in this way, it is possible to alleviate sudden fluctuations in the signal level of the output signal, and to reduce the sense of discomfort with respect to changes in the volume of the output signal. Although detailed description is omitted, the gain control unit 5 also sets the gain G of the amplifying unit 3 to a predetermined fixed value (correction value) for a certain time T2 even when the fluctuation determining unit 4 determines that there is a fluctuation in the decreasing tendency. Or may be gradually reduced to a fixed value G3 over a certain time T2.

  Here, when it is determined by the fluctuation determination unit 4 that there is a fluctuation of an increasing tendency or a decreasing tendency (time t in FIG. 8), the time when the signal level calculation part 1 performs time averaging to calculate the signal level is shortened. It is preferable to reduce the number of samples M. In this way, the signal level calculated by the signal level calculation unit 1 (see the broken line in FIG. 8) from the time point (time t) when the signal level of the input signal suddenly fluctuates (increases in FIG. 8) is actual. Time to follow the input signal can be shortened.

  Also, the time during which the signal level calculation unit 1 performs time averaging to calculate the signal level from the time point (time t) when the speech segment detection unit 7 starts to detect the speech segment until a predetermined time T4 elapses. Is preferably shortened (sample number M is decreased) (see FIG. 9). Note that the average time for shortening is preferably less than several milliseconds, and the fixed time T4 is preferably about several hundred milliseconds to several seconds. In this way, the signal level (see the broken line in FIG. 9) calculated by the signal level calculation unit 1 from the time (time t) when the input signal including the audio component is input is the actual input signal (audio component). Time to follow up can be shortened.

  Furthermore, it is preferable that the signal level calculation unit 1 calculates the signal level only in the voice section detected by the voice section detection unit 7. That is, as shown by a two-dot broken line in FIG. 10, the signal level calculation unit 1 calculates the signal level in the voice interval, and the signal level calculation unit 1 does not calculate the signal level in the non-voice interval, You can substitute it. Thereby, compared with the case where the signal level calculation unit 1 calculates the signal level in the non-speech section, it is possible to suppress a decrease in the signal level in the non-speech section.

  If only the fluctuation of the decreasing tendency of the input signal is to be corrected, the gain controller 5 corrects the gain G of the amplifier 3 without referring to the signal level of the output signal as shown in FIG. Also good.

DESCRIPTION OF SYMBOLS 1 Signal level calculation part 2 Gain calculation part 3 Amplification part 4 Fluctuation determination part 5 Gain control part

Claims (13)

  A signal level calculation unit that calculates a signal level of an input signal, a gain calculation unit that calculates a gain when the signal is amplified based on the signal level calculated by the signal level calculation unit, and the gain calculation unit An amplifying unit for amplifying the signal with the gain calculated in (1), and the signal after the gain calculated by the gain calculating unit is equal to or greater than a predetermined threshold value greater than 1 and amplified by the amplifying unit. When the absolute value of the signal level is greater than or equal to a predetermined upper limit value, the fluctuation determination unit that determines that the input signal has a tendency to increase, and the fluctuation determination unit determines that there is a fluctuation in the increase tendency An automatic gain control device comprising: a gain control unit configured to correct a gain when the amplification unit amplifies the signal to be equal to or less than the threshold value. The input signal is an acoustic signal, and the signal level calculation unit is configured to calculate the signal level of the acoustic signal by averaging the absolute value of the signal level of the acoustic signal over time. An audio section detection unit for detecting an audio section in which an audio component is included in the signal, and an absolute value of the signal level of the acoustic signal in a time shorter than the time when the signal level calculation unit averages the absolute value of the signal level over time A second signal level calculation unit that calculates the signal level of the acoustic signal by averaging the time averages, and the variation determination unit has a predetermined second threshold that the gain calculated by the gain calculation unit is less than 1. When a voice section is detected by the voice section detector and the signal level calculated by the second signal level calculator is below a predetermined threshold for a predetermined time or longer. Determining that the acoustic signal has a decreasing tendency, and the gain control unit determines a gain when the amplifying unit amplifies the signal when the fluctuation determining unit determines that there is a decreasing tendency fluctuation. 2. The automatic gain control apparatus according to claim 1, wherein the automatic gain control apparatus corrects the second threshold value or more. The signal level calculation unit, the automatic gain control device according to claim 1, wherein the averaging the absolute value of the signal level of the signal input time.   The signal level calculation unit and the second signal level calculation unit are a signal level when the speech segment detection unit detects a speech segment and a signal level when the speech segment detection unit does not detect a speech segment. 3. The automatic gain control device according to claim 2, wherein a signal level of the acoustic signal is calculated from a difference between the automatic gain control and the sound signal.   The gain control unit maintains a gain when the amplification unit amplifies the signal at a predetermined fixed value for a predetermined time when the variation determination unit determines that there is variation. 3. The automatic gain control device according to 1 or 2.   3. The gain control unit according to claim 1, wherein when the fluctuation determination unit determines that there is a fluctuation, the gain control unit gradually decreases the gain when the amplification unit amplifies the signal over a predetermined time. The automatic gain control device described.   4. The automatic gain control device according to claim 3, wherein the signal level calculation unit shortens the time when the time averaging is performed when the variation determination unit determines that the increase tendency or the decrease tendency varies. .   The input signal is an audio signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the signal level calculation unit detects the audio segment by the audio segment detection unit. 4. The automatic gain control apparatus according to claim 3, wherein the time when the time is averaged is shortened after a certain period of time has elapsed since the start of the automatic gain control.   The input signal is an acoustic signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the signal level calculation unit is an audio detected by the audio segment detection unit 5. The automatic gain control apparatus according to claim 3, wherein the signal level is calculated only in a section.   The input signal is an acoustic signal, and includes an audio segment detection unit that detects an audio segment in which an audio component is included in the acoustic signal, and the amplification unit is in a non-audio segment that is not detected by the audio segment detection unit. 10. The automatic gain control apparatus according to claim 1, wherein the gain is fixed to 1.   The gain calculation unit stores a plurality of types of values for the gain, and a value selected from the plurality of types of values according to the signal level calculated by the signal level calculation unit is used as the gain. The automatic gain control device according to any one of claims 1 to 10, wherein   The gain calculation unit stores a reference value of the signal level, and the gain is a value obtained by dividing the reference value by the signal level calculated by the signal level calculation unit. The automatic gain control device according to any one of 10 to 10.   The automatic gain control according to any one of claims 1 to 12, wherein each of the units is initialized when a state in which a speech segment is not detected by the speech segment detection unit exceeds a predetermined upper limit time. apparatus.