JP4743018B2 - Howling removal device - Google Patents

Description

  The present invention relates to a howling removing apparatus for removing a howling component of an acoustic signal generated by feedback from an acoustic signal output device such as a speaker to an acoustic signal input device such as a microphone.

In a conventionally known howling removal apparatus, a plurality of notch filters are included, and each of the plurality of notch filters is assigned to each of a plurality of frequencies (howling frequencies) causing howling, thereby being included in an acoustic signal. In some cases, a plurality of howling components can be removed (see Patent Document 1 below).
Japanese Patent No. 3235925

  In a howling removing apparatus of the type having a notch filter as shown in the above-mentioned Patent Document 1, if a notch filter is mounted infinitely, a howling corresponding to the howling every time new howling occurs. As long as the frequency is assigned to the notch filter, all the howling components contained in the acoustic signal can be removed. However, in actuality, whether the notch filter is constituted by an analog circuit or a signal processing operation of a digital circuit, there are limited resources, so the number of mounting points of the notch filter becomes a finite number. . For this reason, a method of efficiently applying a finite number of notch filters to howling removal becomes one problem. In the prior art disclosed in Patent Document 1, the limitation of notch filter resources has not been taken into consideration. For example, all of a finite number of notch filters have already been applied to remove howling components. In the state, when a new howling component is detected, no effective method has been shown for the problem of how to remove the new howling component.

  The present invention has been made in view of the above points, and in a howling removal apparatus having a finite number of notch filters, by efficiently assigning a plurality of howling components to the plurality of notch filters, the present invention is efficient and effective. It is an object of the present invention to provide a howling removing apparatus capable of performing a general howling removal.

The present invention relates to a plurality of notch filters, a howling detection means for detecting a howling frequency included in an externally input acoustic signal, and a howling frequency detected by the howling detection means. Assigning means, time information indicating an application time for the howling frequency assigned by the assigning means, and center frequency, gain and Q parameters set corresponding to the assigned howling frequency, Storage means for storing each notch filter, and weight values used as criteria for selecting a notch filter to be reassigned when all of the plurality of notch filters have already been assigned to howling frequencies, each notch filter, stored in said storage means Application time, the center frequency, and weight calculation means for calculating, based on the gain and Q, by releasing one notch filter selected on the basis of the weight values calculated by the weight calculating means, with respect to the freed notch filter It is a howling removal apparatus provided with the reallocation means which enables reallocation of another howling frequency.

In addition, the present invention is a program executed for operating a computer as a howling removal apparatus having a plurality of notch filters, and detects a howling frequency included in an acoustic signal input from the outside to the computer. How to detect, how to assign the detected howling frequency to any of the plurality of notch filters, time information indicating an application time for the howling frequency assigned by the assignment procedure, and the assigned howling A storing procedure for storing each parameter of the center frequency, gain and Q set corresponding to the frequency for each notch filter, and when all of the plurality of notch filters have already been assigned to the howling frequency, As a criterion for selecting the assignment target 1 a weight value, each notch filter, apply time stored by said storage procedure, center frequency, and weight calculation step of calculating, based on the gain and Q, which are selected on the basis of the weight values calculated by the weight calculation procedure with This is a program for releasing one notch filter and executing a reassignment procedure that makes it possible to reassign another howling frequency to the released notch filter.

According to the present invention, in a howling removal apparatus equipped with a finite number of notch filters, when all the notch filters have already been assigned to the howling frequency, the time information indicating the application time for the assigned howling frequency, The center frequency, gain and Q parameters set corresponding to the assigned howling frequency are stored in the storage means for each notch filter, and a notch filter to be reassigned by the weight calculation means is stored. a weight value to be used as criteria for selecting for each notch filter, the application time stored in the storage unit, the center frequency, the configuration of calculating on the basis of the gain and Q, time applying a weighting value for each filter, and the center Excellent evaluation based on frequency, gain and Q parameters Achieve the results. Then, by selecting a filter to be automatically released based on the weight value, releasing the selected filter, and reassigning the newly detected howling component to the filter, the current howling removal status As a result, it is possible to reassign the newly detected howling component to the notch filter efficiently and effectively.

  In addition, the present invention is not only configured and implemented as an apparatus invention, but may be configured and implemented in the form of a software program executed to cause a computer to operate as a howling removal apparatus according to the present invention.

  Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 1 is a block diagram for explaining the configuration of a howling removing apparatus according to an embodiment of the present invention. Reference numeral 1 denotes a howling removing apparatus according to this embodiment. This howling removal apparatus 1 may be configured as a single apparatus, or may be incorporated as one function in an acoustic apparatus such as a mixer. The howling removal apparatus 1 is roughly divided into an acoustic signal processing unit 2 that performs howling removal processing on an externally input acoustic signal, a control unit 3 that includes a microcomputer including a CPU 10, and a user's operation input. It is comprised from the panel part 4 containing user interfaces, such as various operation element groups 21 and the indicator 22.

  As is well known, howling means that an acoustic signal reproduced from the speaker 7 is input to the microphone 5 again, amplified by the amplifier 6, output from the speaker 7, and input to the microphone 5 again. This is a phenomenon that occurs when Therefore, the howling removal apparatus 1 is inserted between the microphone 5 and the amplifier 8 as shown in FIG. 1, and removes the howling component of the acoustic signal input from the microphone 5 to remove the howling component. An acoustic signal can be reproduced from the speaker 7. The input stage of the howling removal apparatus 1 is provided with an input gain 8 for adjusting the gain of the acoustic signal input to the howling removal apparatus 1, and the acoustic signal output from the howling removal apparatus 1 is provided in the output stage. An output gain 9 for adjusting the gain is provided.

  The acoustic signal processing unit 2 includes a howling detection unit 30 and a plurality of predetermined notch filters 40. As is well known, the notch filter 40 is a filter having a characteristic of abruptly attenuating a specific frequency component set as a center frequency, and by setting a frequency (howling frequency) causing howling as the center frequency. A howling component corresponding to the howling frequency contained in the acoustic signal to be processed can be removed. Since the acoustic signal processing unit 2 includes a plurality of predetermined notch filters 40, the plurality of notch filters 40 can be assigned to each of the plurality of howling components included in the acoustic signal to be processed. The plurality of howling components can be removed. According to the present invention, as will be described in detail later, a finite number of notch filters 40 mounted in the acoustic signal processing unit 2 are efficiently and effectively allocated to a plurality of howling components included in the acoustic signal. Therefore, it is possible to perform appropriate howling removal.

  FIG. 2 is a block diagram showing an electrical hardware configuration of the howling removal apparatus 1. As shown in FIG. 2, the howling removal apparatus 1 includes a CPU 10, a ROM 11, a RAM 12, a signal processing circuit (DSP) 13, an acoustic interface (acoustic I / F) 14, a communication interface (communication I / F) 15, and an operator detection circuit. 16 and a display control circuit 17, and the devices are connected via a communication bus 18. A microcomputer including the CPU 10, the ROM 11, and the RAM 12 corresponds to the control unit 3 shown in FIG. The CPU 10 executes various control programs stored in the ROM 11 or the RAM 12, and controls the overall operation of the howling removal apparatus 1, that is, the signal processing control in the acoustic signal processing unit 2 and the panel unit 4 in FIG. Responsible for the control.

  An acoustic signal input / output device (acoustic signal I / O) 19 such as a microphone 5 and a speaker 7 is connected to the howling removal apparatus 1 via an acoustic I / F 14. The acoustic I / F 14 includes an AD converter and a DA converter, converts an analog acoustic signal supplied from the acoustic signal I / O 19 into a digital acoustic signal, and takes it into the howling removal apparatus 1. The digital audio signal subjected to signal processing is converted into an analog audio signal and output to the audio signal I / O 19. Further, by connecting a personal computer (external device 20), for example, via the communication I / F 15, the howling removal apparatus 1 may be remotely controlled in the personal computer. As another example of the external device 20, an audio device such as a mixer may be connected via the communication I / F 15.

  An operator group 21 arranged in the panel unit 4 (see FIG. 1) is connected via the operator detection circuit 16, and the user uses the operator group 21 to give instructions to start howling detection processing and various users. Operation input such as setting can be performed. Various operation inputs made by the user using the operation element group 21 are detected by the operation element detection circuit 16, and the CPU 10 controls the operation of the howling removal apparatus 1 according to the detected operation input. Further, a display 22 arranged on the panel unit 4 (see FIG. 1) is connected via the display control circuit 17. The display 22 is for displaying a character string or the like indicating various data or an operating state to the user based on the control of the CPU 10, and may be configured by, for example, an LED display or an LCD liquid crystal display.

  The DSP 13 implements each function (howling detection unit 30 and predetermined plural notch filters 40) of the acoustic signal processing unit 2 shown in FIG. 1 by executing signal processing based on control by the CPU 10. That is, the DSP 13 performs howling detection processing (operation of the howling detection unit 30 shown in FIG. 1) on the digital audio signal input via the audio I / F 14 and, based on the result of the howling detection processing, A process of removing howling components contained in the digital acoustic signal by the notch filter 40 is performed.

  Each function of the acoustic signal processing unit 2 will be briefly described. The howling detection unit 30 performs signal analysis of the input acoustic signal, and detects the frequency of the howling component (howling frequency) included in the signal and the amplitude level of the howling frequency. It should be noted that conventionally known howling detection algorithms such as pitch detection, FFT (fast Fourier transform), and BPF (bandpass filter) may be applied as an algorithm for realizing the operation of the howling detection unit 30. The detection result (howling frequency and amplitude level) in the howling detection unit 30 is output to the control unit 3.

  The control unit 3 assigns one howling component detected by the howling detection unit 30 to one notch filter 40. The allocation of the detected howling component to the notch filter is managed by a notch filter management table 41 shown in FIG. The notch filter management table 41 is provided in an appropriate memory such as the RAM 12 included in the control unit 3 (see FIG. 1), and a predetermined plurality of “i” (“i”) provided in the acoustic signal processing unit 2 is provided. A plurality of “i” filter management tables are prepared corresponding to each of the notch filters 40 (“is a positive integer). As shown in FIG. 3, the filter management table 41 corresponding to one notch filter includes parameter setting values of gain, center frequency, and Q, which are parameters constituting the notch filter, and the last update time of the filter management table 41. Is stored. That is, the control unit 3 stores, in any one notch filter management table 41, the center frequency, gain, and Q parameter setting values based on the howling frequency and level of the howling component detected by the howling detection unit 30. Thus, the detected howling component is newly assigned to the notch filter 40 corresponding to the notch filter management table 41.

  The last update time stored in the notch filter management table 41 is the time when the filter management table 41 is updated, that is, the time when a certain howling component is newly assigned to a certain notch filter. It is a parameter which shows. Here, as the “time” information, data measured by a timer (clock) included in the control unit 3 (microcomputer including the CPU 10) may be used. Since the last update time is held in the filter management table 41, the control unit 3 can obtain the application time (elapsed time) from when the notch filter 40 is assigned to the howling frequency to be processed to the present time. it can. For example, the application time can be obtained for the filter 40 by subtracting the last update time of the filter management table 41 from the current time, or each notch can be obtained by comparing the last update times of the filter management tables 41. The length of the application time for each filter 40 can be determined. Note that the filter management table 41 is not limited to the configuration that holds the last update time, but may be information that can obtain the application time after the notch filter 40 is assigned to the howling component to be processed. For example, other appropriate time information such as data representing the application time by a count value accumulated every predetermined period from the last update time may be held.

  The control unit 3 reflects the settings of the parameters (gain, center frequency, Q value) stored in the filter management table 41 on the DSP 13 so that the notch has a filter characteristic based on the parameters stored in the filter management table 41. A filter is applied to the detected howling component. Thereby, in the acoustic signal processing unit 2 (DSP 13), the detected howling component included in the acoustic signal can be removed.

  FIG. 4A is a description example of the notch filter management table 41. As an example, it is assumed that four notch filters 40 (filters N1 to N4) are provided in the howling removal apparatus 1, and the filters N1 to N1 are provided. It is assumed that four howling components H1 to H4 detected for N4 are assigned. In FIG. 4A, the filter management tables T1 to T4 corresponding to the filters N1 to N4 include gains, center frequencies, and Q values corresponding to the howling frequencies and levels of the howling components H1 to H4 to be removed. Each parameter and the last update time of each filter management table T1 to T4 are stored. In this example, the last update time is data described by “hour: minute: second” as shown in the figure. FIG. 4B schematically shows the filter characteristics of the filters N1 to N4 and the corresponding howling components H1 to H4. The horizontal axis represents frequency and the vertical axis represents amplitude level.

  When the howling detection unit 30 newly detects the howling frequency H5 in the state shown in FIGS. 4A and 4B, any one of the notch filters N1 to N4 is applied to the newly detected howling frequency H5. I want to assign a new one. However, the howling components H1 to H4 have already been assigned to the four notch filters N1 to N4. Therefore, the control unit 3 releases at least one of these filters N1 to N4, and reassigns (reassigns) the newly detected filter to the newly detected howling frequency band H5. You must do the processing. At this time, the howling frequency that is currently being removed should not be affected as much as possible. Therefore, the control unit 3 calculates the “weight” for each filter 40 as a criterion for selecting the filter 40 to be released, and compares the calculated “weight” for each filter 40. The filter 40 to be released is controlled to be selected.

As an example, the “weight” for each filter 40 is based on the weight element based on the last update time and the center frequency among the parameters stored in each filter management table 41 as shown in the following formula (1). It can be calculated by multiplying each weight element of the weight element and the weight element based on the gain and Q. Note that the greater the value of the “weight”, the more preferentially the filter is targeted for release.
Filter weight = weight of last update time * weight of center frequency * weight of gain and Q (Expression 1)
In this specification, the symbol “*” indicates multiplication.

  FIG. 5A shows a weighting element based on the last update time, FIG. 5B shows a weighting element based on gain and Q, and FIG. 5C shows a table configuration example for obtaining each weighting element based on the center frequency. ing. In the description of this embodiment, a table is used as an example of a calculation method for each weight element. However, the present invention is not limited to this, and each weight element is calculated using an arithmetic expression corresponding to the table characteristics described below. May be.

  In (a), the value of the weight element based on the last update time shown on the vertical axis (shown on the vertical axis) is the application time (in seconds) after the notch filter 40 shown on the horizontal axis in FIG. It has a logarithmic function characteristic that gradually increases with the increase of. As a result, it is possible to more preferentially select a filter with an old last update time as a release target. The filter with the older last update time can be considered less likely to affect the current howling occurrence status (i.e., the current howling removal status) if it is released. It is meaningful to select an older filter as a release target with higher priority.

  In FIG. 5B, the weighting element based on the gain and Q is a value obtained by a calculation formula that multiplies the gain value by the reciprocal of Q, which is a value that correlates with the range to be filtered. . The range to be filtered is centered on the center frequency determined according to the magnitude of the gain and the frequency bandwidth according to the value of Q, as is clear from the filter characteristics of the notch filter shown in FIG. The calculation formula shown in Fig. 5 (b) is not an expression for obtaining a value that accurately corresponds to the range to be filtered, but even a value that correlates to the range to be filtered can be obtained. In the illustrated calculation formula, the gain value is multiplied by the reciprocal of Q because the larger the set value of the parameter “Q” is, the narrower the frequency bandwidth is. (The numerical value of Q and the frequency bandwidth are inversely proportional.) The weighting factor based on this gain and Q increases as the range to which the filter is applied increases, that is, Therefore, a filter that has not been sharply narrowed, that is, a filter with a low howling frequency removal accuracy, is more preferentially released. It is possible to leave what is sharply applied only to the howling frequency, so that the effect of howling cancellation by the notch filter or the frequency band not related to the howling component is not affected as much as possible. It is beneficial in that.

  In FIG. 5C, the horizontal axis indicates the set value (Hz unit) of the center frequency, and the vertical axis indicates the value of the weight element based on the center frequency. As shown in (c), the weighting factor based on the center frequency is set to a large value near a low frequency or a high frequency where the auditory sensitivity is low, and by reducing the value of the weighting factor for a relatively high frequency range of the auditory sensitivity, The filter is preferentially released from the filter assigned to the frequency range with low auditory sensitivity. As a result, the hearing sensitivity is relatively high, and therefore, howling components for which howling removal is more important can be prevented from being affected.

  Note that the characteristics of the weight elements shown in FIGS. 5A to 5C are examples, and are not limited to the illustrated examples. For example, the characteristic of the weighting element based on the last update time shown in (a) is not limited to the logarithmic logarithmic function characteristic as shown in the figure, but other appropriate characteristics such as a linearly increasing characteristic with respect to the application time. You can have it. In addition, in (b), the formula for obtaining the weight element based on the gain and Q is not limited to the example shown in the figure, and any formula can be used as long as it obtains a value that correlates with the range to be filtered based on the gain and / or Q. It may be a thing. In this embodiment, an example in which one weight element is obtained based on both gain and Q has been described. However, a weight element based on gain and a weight element based on Q may be obtained respectively. Further, the characteristics of the weight element with respect to the center frequency shown in (c) are not limited to the illustrated example. In addition, the characteristics of the weighting elements shown as examples in FIGS. 5A to 5C may be arbitrarily set by the user.

  FIG. 6 is a table showing the results of calculating the weights of the filters N1 to N4 based on the notch filter management tables T1 to T4 shown in FIG. The weighting element based on the last update time, the weighting element based on the center frequency, and the weighting element based on the gain and Q are values obtained according to the table configuration shown in FIGS. Was obtained using the above formula (1). In the example shown in FIG. 6, since the weight of the filter N1 is the maximum, the filter N1 is selected as a release target.

As another method for calculating the weight of the entire filter, as shown in the following calculation formula (2), a weight element based on the last update time, a weight element based on the center frequency, and a weight element based on the gain and Q, respectively. Further, a weighting factor may be multiplied.
Filter weight = (weight of last update time * coefficient a) * (weight of center frequency * coefficient b) * (weight of gain and Q * coefficient c) (Expression 2)
In the above formula (2), the coefficient a to coefficient c multiplied by each weight element may be appropriately weighted depending on which parameter is important as a determination element to be released. In addition, the user may be able to arbitrarily set the weights of the coefficients a to c multiplied by the weight elements.

  In addition, regardless of which calculation method (1) or (2) is applied, the weighting element used for determining the release target (that is, used for calculation) is any one of the above three. May be selectively combined as appropriate, or the selective combination may be arbitrarily performed by the user. Further, the filter weight calculation method is not limited to the method of the above formula (1) or (2), as long as the weight for each filter can be evaluated using each weight element or an appropriate combination thereof. Good. In short, various variations of the weight calculation method for each filter are possible depending on the combination of weight elements for each parameter, the calculation method, the weighting of coefficients, and the like.

  FIG. 7 is a flowchart for explaining an example of the procedure of the main process for howling removal performed by the howling removal apparatus 1 according to this embodiment. The main process shown in FIG. 7 may be activated in response to a predetermined key operation by the user. For example, the operator 21 of the panel unit 3 is equipped with a “howling check button” for instructing howling detection and removal, and the user operates the “howling check button” to activate the main process. It may be possible. Or you may make it start the said main process automatically for every predetermined period (for example, once per minute etc.).

  In step S1, various setting (environment) information related to the operation of the howling removal apparatus 1 is acquired. Here, each set value stored in all the notch filter management tables 41 is acquired, and the current operation state is grasped. Further, in the howling removal apparatus 1 according to this embodiment, the user uses the operation element 21 of the panel unit 3 to either (1) permit the resetting of all the current notch filters 40 or (2) the notch It is possible to set whether to permit reassignment of a new howling component to the filter. In this embodiment, in the acquisition of the setting information in step S1, information regarding the two user settings is acquired.

  In step S2, based on the user setting information (1) acquired in step S1, it is determined whether to reset the settings of all notch filters 40 that have already been assigned to howling components. If the resetting of the notch filter 40 is permitted by the user setting (yes in step S2), the settings of all the notch filters 40 are reset in step S3. As a result, the reset notch filter 40 can be newly assigned to a howling component newly detected by howling detection to be described later. Therefore, it is necessary to perform filter release processing for reassignment according to this embodiment. Disappear. If the resetting of the notch filter 40 is not permitted by the user setting (no in step S2), the notch filter 40 in step S3 is not reset, and the currently assigned notch filter 40 setting is maintained. be able to.

  In step S4, howling detection processing is performed. The howling detection process is a process executed by the howling detection unit 30. As described above, conventionally known howling detection algorithms such as pitch detection, FFT (fast Fourier transform), and BPF (bandpass filter) are used. May apply.

  When a howling frequency is not found by the howling detection process at step S4 (no at step S5), a confirmation display (for example, howling frequency) that a howling frequency has not been found using the display unit 22 at step S6. And the like, and the main routine is terminated.

  When a howling frequency is found by the howling detection process in step S4 (yes in step S5), in step S7, various setting (environment) information (notch filter management) related to the operation of the howling removal apparatus 1 acquired in step S1. Based on the data in the table 41), it is checked whether all the notch filters 40 mounted in the howling removal apparatus 1 have already been assigned to the howling components. If there is an unassigned notch filter 40, step S7 is branched to no, and the process proceeds to step S11 described later. When the notch filter 40 is reset by the user setting in the step S3, since all the notch filters 40 are in an unassigned state, the step S7 is branched to no, and the process proceeds to a step S11 described later. .

  If all of the notch filters 40 mounted on the howling removal apparatus 1 have already been assigned to howling components (yes in step S7), in step S8, new howling components are reassigned to the assigned notch filter 40. Determine whether to allow. This reassignment permission is determined based on the user setting information (2) acquired in step S1. If reassignment of a new howling component to the assigned notch filter 40 is permitted by the user setting (yes in step S8), “filter release processing” described later with reference to FIG. 8 is performed in step S10. If reassignment of the notch filter 40 is not permitted by the user setting (no in step S8), it is indicated in step S9 that all the notch filters 40 have been assigned to howling components using the display unit 22. Is displayed (for example, a character string notifying that the howling frequency has not been found), and the main routine is terminated without reassigning the notch filter 40. That is, the howling component newly detected in the howling detection process in step S4 is left without being removed. Thus, by allowing the user to set whether or not to permit reassignment, the current assignment state of the notch filter 40 can be maintained without performing reassignment of the notch filter 40 as desired by the user. Can be made.

  FIG. 8 is a flowchart showing an example of the procedure of “filter release processing”. In step S <b> 12, the “weight” for each filter 40 is calculated based on the parameter setting values stored in the filter management table 41 corresponding to all the assigned notch filters 40. As an example, the weight calculation method for each filter can be calculated by the above formula (1) using three weight elements obtained from the respective tables shown in FIGS. As described above, the weight calculation method for each filter can be variously varied depending on the combination of weight elements for each parameter, the calculation method, the salt of the coefficient, and the like.

  In step S13, based on the weight for each filter calculated in step S12, the notch filter 40 having the largest weight value (ie, the release target) is selected. In step S14, it is checked whether or not there are a plurality of filters to be released selected in step S13, that is, whether or not the weights of the plurality of filters have the same maximum value. When there are a plurality of filters to be released (yes in step S14), in step S15, the set values of the last update times of the plurality of filters selected as the release targets are compared, and the oldest update time is released. Select the filter. In this step S15, as an example, the final update time is referred to in determining the selection of the filter to be released. However, the present invention is not limited to this, and other parameters (center frequency, gain, Q, etc.) are referred to. Also good. In addition, the user may be able to set the type of parameter referred to the selection of the release target filter in step S15. In step S16, the notch filter 40 selected as the release target by the above processing is released, and the released filter 40 is put into an unallocated state.

  Returning to the main processing of FIG. 7, in step S11, the notch filter 40 released in step S16 of FIG. 8 is reassigned to the howling component newly detected in step S4. That is, in the filter management table 41 corresponding to the released notch filter 40, the parameter setting values of the center frequency, gain, and Q corresponding to the howling component newly detected in step S4 are stored, and the current time Is set to the last update time of the filter 40. Thereby, the released notch filter 40 can be reused for removing the newly detected howling component. If there is an unassigned notch filter 40 in step S7, an appropriate unassigned notch filter 40 is assigned to the howling component newly detected in step S4 in step S11. Assign it. Then, by applying the notch filter 40 assigned to the newly detected howling component (that is, reflecting the contents of the table 41 to the DSP 13), the newly detected howling component is removed. Can do.

  As described above, according to the present invention, when all of the notch filters 40 mounted in the howling removal apparatus 1 have already been assigned to howling components, by calculating the weight for each filter, Based on the weight of each filter, a filter to be released can be automatically selected, and a newly detected howling component can be reassigned to the selected filter. Therefore, it is possible to effectively remove a newly detected howling component without affecting the already removed howling component as much as possible. Also, since the weight for each filter is calculated based on the last update time (application time), center frequency, gain, and Q parameters for each filter, the weight for each filter can be evaluated in a composite manner. it can. Therefore, there is an excellent effect that the newly detected howling component can be reassigned to the notch filter efficiently and effectively without affecting the current howling removal state as much as possible.

  In the above-described embodiment, the example in which the howling removal apparatus according to the present invention is configured and implemented as the apparatus invention has been described. However, the present invention is not limited thereto, and the software program that causes the computer to operate as the howling removal apparatus according to the present invention You may comprise and implement this invention.

The functional block diagram of the structure of the howling removal apparatus which concerns on one Example of this invention. The block diagram which shows the electrical hardware constitutions of the howling removal apparatus which concerns on the same Example. The figure which shows the structural example of the filter management table of the howling removal apparatus which concerns on the same Example. It is one implementation example of the howling removal apparatus which concerns on the Example, (a) is a description example of filter management table T1-T4 corresponding to four filters N1-N4, (b) is each filter N1-N4. The figure which shows filter characteristics and the corresponding howling components H1-H4 in simulation. In the howling removal apparatus according to the embodiment, a table configuration example for obtaining weight elements for filter reassignment, wherein (a) is a weight element based on the last update time, and (b) is a weight element based on gain and Q. , (C) is a weighting element based on the center frequency. The weight for each filter calculated in the implementation example of FIG. The flowchart which shows an example of the procedure of the main process of the howling removal apparatus which concerns on the Example. The flowchart which shows an example of the procedure of the filter release process in the howling removal apparatus which concerns on the same Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Howling removal apparatus, 2 Acoustic signal processing part, 3 Control part, 4 Panel part, 5 Microphone, 6 Amplifier, 7 Speaker, 8 Input gain, 9 Output gain, 10 CPU, 11 ROM, 12 RAM, 13 Signal processing circuit, 14 acoustic interface, 15 communication interface, 16 operator detection circuit, 17 display control circuit, 18 communication bus, 19 acoustic signal input / output device, 20 external device, 21 operator, 22 display, 30 howling detection unit, 40 notch filter , 41 Filter management table

Claims (2)

Multiple notch filters;
Howling detection means for detecting howling frequency contained in an externally input acoustic signal;
Assigning means for assigning a howling frequency detected by the howling detection means to any of the plurality of notch filters;
Memory storing time information indicating the application time for the howling frequency assigned by the assigning means, and the center frequency, gain, and Q parameters set corresponding to the assigned howling frequency for each notch filter. Means,
When all of the plurality of notch filters are already assigned to the howling frequency, a weight value used as a reference for selecting a notch filter to be reassigned is stored in the storage unit for each notch filter. Weight calculation means for calculating based on the applied time, the center frequency, the gain and the Q ,
Reallocating means for releasing one notch filter selected on the basis of the weight value calculated by the weight calculating means so that another howling frequency can be reassigned to the released notch filter. Howling removal device. A program executed to cause a computer to operate as a howling removal device having a plurality of notch filters ,
A howling detection procedure for detecting a howling frequency included in an externally input acoustic signal;
An assigning procedure for assigning the detected howling frequency to any of the plurality of notch filters;
Memory storing time information indicating the application time for the howling frequency allocated by the allocation procedure, and the center frequency, gain, and Q parameters set corresponding to the allocated howling frequency for each notch filter. Procedure and
When all of the plurality of notch filters have already been assigned to the howling frequency, a weight value used as a reference for selecting a reassignment target is stored for each notch filter by the storage procedure according to the storage procedure. A weight calculation procedure for calculating based on frequency, gain and Q ;
A reassignment procedure for releasing one notch filter selected based on the weight value calculated by the weight calculation procedure, thereby enabling reassignment of another howling frequency to the released notch filter; A program to be executed .

Images