JP4428247B2 - Loudspeaker - Google Patents

Description

  The present invention relates to a loudspeaker equipped with a howling canceller for preventing howling.

  The loudspeaker amplifies the audio signal input from the microphone and inputs it to the speaker. In this loudspeaker, a closed loop is formed in the path from the speaker to the microphone, and the sound is output from the speaker and input to the microphone. It is well known that howling occurs by repeatedly amplifying the feedback signal.

In order to prevent such howling, it has been conventionally proposed to use a howling canceller for a loudspeaker that generates a simulated signal that simulates a feedback sound signal using an adaptive filter and removes the simulated signal from an input signal from a microphone. (See Non-Patent Document 1). In this howling canceller, a signal input to the speaker is input to the delay unit, and this signal is delayed by the delay unit by a delay time corresponding to the delay time from the speaker to the microphone. By performing a convolution operation on the delayed signal with a filter coefficient, a transfer function of the feedback sound from the speaker to the microphone is simulated by the adaptive filter, and a simulated signal simulating the feedback sound signal is generated. Then, the residual signal obtained by removing the simulation signal from the input signal from the microphone by the adding unit is output to the amplifying unit, whereby the feedback sound component is removed. Since the residual signal amplified by the amplifier is input to the speaker and emitted from the speaker, howling is prevented. Here, the residual signal is input to the adaptive filter as a reference signal, and the filter coefficient is corrected using a known adaptive algorithm (for example, LMS (Least Mean Square) algorithm) so that the residual signal is minimized. The As a result, even when there is a change in the acoustic conditions for the sound field of the microphone and the speaker, the adaptive filter is adapted to follow the change in the transfer function from the microphone to the speaker, and the simulated signal is brought closer to the feedback sound signal. Prevention of howling.
Inazumi, Imai, Konishi: “Preventing howling in loudspeakers using the LMS algorithm”, Proc. 417-418 (1991, 3)

  In a conventional howling canceller using an adaptive filter, when the transfer function of the feedback transfer path from the microphone to the speaker changes significantly, such as when the microphone placement position moves greatly, the filter coefficient is changed to the changed transfer function. There is a problem that it takes a long time to identify. That is, the howling canceller using the adaptive filter identifies the target transfer function by gradually correcting the filter coefficients as described above. For this reason, when the target transfer function changes greatly, it becomes difficult to quickly adapt the filter coefficient to the transfer function that has changed greatly.

  In addition, if the transfer function of the feedback transfer path has changed significantly and the transfer function and the filter coefficient of the adaptive filter are significantly different, the simulated signal generated by the adaptive filter is significantly different from the feedback sound signal. Become. Even if a simulated signal that is significantly different from the feedback sound signal is subtracted from the input signal from the microphone, not only the feedback sound component can be accurately removed, but also the sound signal component (for example, the speaker's voice) that is the target of loudening. In some cases, the noise component is removed or the noise component is added. As a result, inconveniences may arise in order to properly amplify the audio signal to be amplified.

  Accordingly, an object of the present invention is to provide a loudspeaker that can adapt an adaptive filter more quickly and effectively prevent the occurrence of the inconveniences described above in order to solve the above problems. Yes.

  In order to solve the above problems, the present invention employs the following means.

The present invention relates to a loudspeaker connected to a speaker and a microphone, a loudspeaker circuit unit that amplifies an input audio signal and inputs the amplified signal to the speaker, and a filter that simulates a transfer function of a feedback transfer system from the speaker to the microphone An adaptive filter that self-sets the coefficient, and delays the audio signal input to the speaker by a delay time that simulates the delay time of the feedback transmission system and inputs the delayed signal to the adaptive filter, and the simulated signal processed by the adaptive filter A howling canceller that outputs a residual signal subtracted from an audio signal input from the microphone to the loudspeaker circuit unit and a change level of a transfer function of the feedback transfer system are detected, and when the detected change level is greater than a predetermined level And a howling canceller control unit for initializing the adaptive filter. .

  According to the above configuration, when a signal is input to the microphone, the simulation signal is subtracted from the input signal by the howling canceller and supplied to the loudspeaker circuit unit as a residual signal, amplified by the loudspeaker circuit unit, and supplied to the speaker. The sound is input and emitted from the speaker. Here, the simulation signal is processed by an adaptive filter. The adaptive filter self-sets a filter coefficient that simulates the feedback transmission system from the speaker to the microphone based on the audio signal and the residual signal input to the speaker. Therefore, the simulated signal is transmitted from the speaker to the microphone by the feedback transmission system. The feedback sound that is fed back is simulated, and howling is prevented by subtracting the simulated signal from the input signal.

  Here, depending on the degree of change level of the transfer function of the feedback transfer system, it may take a longer time to identify the adaptive filter in the initialized state in order to identify the adaptive filter following the change of the transfer function. . In addition, the adaptive filter in the initialized state may have a higher degree of adaptation to the transfer function after the change than the adaptive filter before the transfer function changes.

  In the above configuration, the howling canceller control unit detects the change level of the transfer function of the feedback transfer system, and the adaptive filter is initialized when the detected change level is greater than a predetermined level. For this reason, when the detected change level is larger than the predetermined level, the adaptive filter in the initialized state has a higher degree of adaptation to the transfer function after the change than the adaptive filter before the transfer function changes. By setting, the adaptive filter can be identified more quickly by the transfer function changed compared to the conventional technique in which the adaptive filter is not initialized according to the change level of the transfer function. In addition, during the period until the adaptive filter identifies the transfer function after the transfer function of the feedback transfer system is changed, it is possible to prevent the simulated signal from being significantly different from the feedback sound signal as compared with the prior art. As a result, in such a period, it is possible to prevent inconveniences such as removal of a sound signal component to be loudened when a simulated signal is removed from a sound signal input from a microphone or adding a noise component on the contrary. Become.

  The initialization of the adaptive filter is, for example, setting the first tap setting to “1” or an integer smaller than “1” and setting the remaining taps to “0”, or setting all taps to “0”. And so on. In the configuration in which the first tap is set to “1” or an integer smaller than “1” and the remaining taps are set to “0”, in the adaptive filter in the initialized state, the sound signal input to the speaker is delayed after being delayed by the delay unit. Is subtracted from the audio signal input from. In the configuration in which all the taps are set to “0”, the adaptive filter in the initialized state does not remove the feedback sound component from the sound signal input from the microphone. Further, the filter coefficient may be initialized with an initial value calculated based on an empirical value in advance.

(1) In the above loudspeaker, the howling canceller control unit detects a signal level between a residual signal obtained by removing a simulated signal from the input signal and an audio signal input from the microphone, and an audio signal input from the microphone When the residual signal is larger than a predetermined level, the adaptive filter is initialized by assuming that the change level is larger than the predetermined level.

  According to the above configuration, the signal level of the residual signal and the audio signal input from the microphone is detected by the howling canceller control unit, and when the residual signal exceeds the predetermined level than the audio signal input from the microphone, It is determined that the change level is greater than the predetermined level. Here, the state in which the residual signal is larger than the audio signal input from the microphone is a state in which a noise component is added by removing the simulated signal from the audio signal input from the microphone. In such a state, the degree of mismatch between the adaptive filter and the transfer function of the feedback transfer system is greater than the degree of mismatch between the adaptive filter and the transfer function in the initial state. According to the above-described configuration, it is possible to accurately determine that the state is in such a state by a simple process of simply comparing the signal levels of the audio signal input from the microphone and the residual signal. “When the residual signal is larger than a predetermined level than the sound signal input from the microphone” means, for example, when the residual signal is larger than the sound signal input from the microphone or when the residual signal is larger than the sound signal input from the microphone. This is the case when the residual signal is larger than several times.

(2) In the above loudspeaker, the howling canceller control unit detects a correction amount for calculating a filter coefficient of the adaptive filter, and when the correction amount is larger than a threshold value, the change level is larger than a predetermined level. The adaptive filter is initialized as follows. The correction amount is calculated based on the residual signal and the audio signal input to the speaker, and is calculated as a large value when the residual signal value is large. As described above, it is possible to detect that the noise component is added by removing the simulation signal from the voice signal input from the microphone according to the value of the residual signal. For this reason, the degree of mismatch between the adaptive filter and the transfer function of the feedback transfer system can be reduced by simply processing the comparison of the correction amount with the threshold value by setting the threshold value to the value of the correction amount in such a state. It is possible to accurately determine that the adaptive filter and the transfer function in the initial state are in a state greater than the degree of mismatch.

(3) In the above loudspeaker, when the adaptive filter is initialized, the gain for amplifying the residual signal by the loudspeaker circuit unit is smaller than when the adaptive filter is not initialized. It further has a setting part. According to this configuration, when the adaptive filter is initialized, the gain that amplifies the residual signal by the loudspeaker circuit unit is made smaller by the gain control unit than when the adaptive filter is not initialized. When the adaptive filter is initialized, the simulation signal becomes 0 or an initial value. For this reason, even if such a simulation signal is removed from the audio signal input from the microphone, the residual signal is likely to include a feedback sound signal, and howling is likely to occur. According to the configuration of the present invention, in such a case, the gain is made smaller than that in the case where the adaptive filter is not initialized, so that howling can be effectively prevented.

(4) In the above loudspeaker, a switching unit that is connected to a plurality of sets of the microphone and the speaker and selectively turns on / off a set of the microphone and the speaker, and the switching A switching pattern by the switching unit, a switching pattern by the switching unit, and a table storage unit that stores a table for registering the delay time corresponding to the switching pattern, and the howling canceller control unit is controlled by the switching unit. When the switching pattern is changed, the adaptive filter is initialized, the delay time corresponding to the switching pattern of the switching unit is acquired from the table, and the howling canceller is set to the acquired delay time. It is characterized by that.

  According to the above configuration, a pair that is connected to a plurality of sets of microphones and speakers and that turns on input / output from the plurality of pairs of microphones and speakers is selectively switched by the switching unit. In a configuration in which multiple pairs of microphones and speakers are connected, and the pair in which input / output is turned on is switched from this pair, the transfer function of the feedback transfer system changes until it is difficult to quickly identify the adaptive filter. Or the delay time changes. The change in the delay time can be predicted as long as the arrangement of the speaker and the microphone does not change.

Here, in the table storage unit, a table for registering the switching pattern by the switching unit and the delay time corresponding to the switching pattern is stored in advance, and the delay time corresponding to the change of the delay time predicted in advance is registered in this table Is possible.

  When the switching pattern is changed by the switching unit, the delay time corresponding to the switching pattern of the switching unit is acquired from the table, and the howling canceller is set to the acquired condition. For this reason, it becomes possible to make a howling canceller respond | correspond rapidly to the change of the delay time after switching pattern change. This makes it possible to quickly match the timing at which the feedback sound signal is input to the howling canceller and the timing at which the simulated signal corresponding to the feedback sound signal is subtracted from the sound signal input by the howling canceller. Can be effectively prevented.

  Furthermore, when the switching pattern is changed, the adaptive filter is initialized by the howling canceller control unit, so that the adaptive filter can be adapted more quickly compared to the conventional technology that does not initialize the adaptive filter according to the change of the switching pattern. It becomes possible to identify the filter. In addition, after the change of the switching pattern, the audio signal component to be amplified may be removed from the audio signal input from the microphone during the period until the adaptive filter adapts to the change in the transfer function of the feedback transfer system caused by this change. However, it is possible to prevent inconveniences such as adding a noise component.

  According to the present invention, the howling canceller control unit detects the change level of the transfer function of the feedback transfer system, and the adaptive filter is initialized when the detected change level is greater than the predetermined level. Here, depending on the degree of the change level of the transfer function in the feedback transfer system, in order to identify the adaptive filter immediately before the transfer function changes following the change of the transfer function, the adaptive filter in the initialized state is identified. May take a long time. In addition, compared with the adaptive filter immediately before the transfer function changes, the adaptive filter in the initialized state may have a higher degree of adaptation to the transfer function after the change. According to the present invention, when the detected change level is greater than a predetermined level, the adaptive filter in the initialized state has a higher degree of adaptation to the transfer function after the change than the adaptive filter before the transfer function changes. By matching with the large case, the adaptive filter can be identified more quickly.

  In addition, it effectively prevents removal of a simulated signal that is significantly different from the feedback sound signal from the input signal from the microphone during the period from when the transfer function changes until the adaptive filter identifies the change in the transfer function. be able to. For this reason, there is an effect that inconveniences such as removal of a voice signal component (for example, a speaker's voice) that is a target of loudspeaking or addition of a noise component instead of a voice signal input from a microphone occur. Can be prevented.

  Embodiments of the present invention will be described below with reference to the drawings. The present embodiment is characterized in that the adaptive filter of the howling canceller is returned to the initial state when the transfer function of the feedback transfer system from the speaker to the microphone greatly changes. When the level of change in the transfer function of the feedback transfer system from the speaker to the microphone increases, the degree of coincidence (degree of adaptation) of the transfer function in the initial state of the adaptive filter changes compared to the adaptive filter immediately before the change of the transfer function. May be big. In such a case, by returning the adaptive filter to the initialization state, the transfer function can be adapted to the changed transfer function more quickly than the conventional technology that does not employ the process of returning the adaptive filter to the initial state according to the change level. It is.

  In addition, during the period from when the transfer function changes until the adaptive filter identifies the change of the transfer function, a simulated signal that is significantly different from the feedback sound signal is removed from the audio signal (microphone input signal) input from the microphone. The inconvenience caused by this is effectively prevented. In other words, although this affects the calculation of the adaptive filter coefficient after the change of the adaptive filter coefficient before the transfer function of the feedback transfer system changes significantly, in this embodiment, the processing is performed with the adaptive filter returned to the initial state. It is effective to remove the simulated signal from the microphone input signal to eliminate the inconvenience such as removing the voice signal component (for example, the speaker's voice) that is the target of loudening or adding a noise component instead. To prevent.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to FIGS. 1 and 2. FIG. 1 is a block diagram showing a schematic configuration of a loudspeaker 1 according to the first embodiment. A loudspeaker 1 is connected to a microphone 2 and a speaker 3 and includes a loudspeaker processing unit 4, a howling canceller control unit 5, and a gain control unit 6.

  The microphone 2 inputs sound from outside the apparatus as a microphone input signal. The microphone input signal is A / D converted by an A / D (Analog / Digital) converter (not shown). This A / D converted signal is output to the loudspeaker processing unit 4. The speaker 3 converts the analog audio signal input from the loudspeaker 1 into sound and emits the sound.

  The loudspeaker processing unit 4 includes an amplifying unit 41 and a howling canceller 42. The amplification unit 41 corresponds to the loudspeaker circuit unit of the present invention, and performs signal processing such as digital amplification of a microphone input signal input via an A / D converter (not shown) -howling canceller 42. The audio signal (x) subjected to signal processing is input to the howling canceller 42 and the speaker 3. The howling canceller 42 includes a delay unit 421, an adaptive filter 422, and an addition unit 423.

  The delay unit 421 and the adaptive filter 422 simulate the feedback transmission system 100 that is an acoustic transmission path from the speaker 3 to the microphone 2. That is, the delay unit 421 simulates the delay time τ of the feedback sound through the feedback transfer system 100, and the adaptive filter 422 simulates a transfer function that is a sound propagation characteristic of the feedback transfer system 100.

  Specifically, the delay unit 421 delays the input audio signal x (k) by the delay time τ, and outputs the delayed audio signal x (k−τ) to the adaptive filter 422. The adaptive filter 422 includes a digital filter 422a (typically a FIR filter: Finite Impulse Response Filter) and a filter coefficient calculation unit 422b that determines a filter coefficient of the digital filter 422a. The digital filter 422a generates a simulated signal do (k) by performing a convolution operation on the input audio signal x (k−τ) with a filter coefficient. The digital filter 422a outputs the generated simulation signal do (k) to the adding unit 423.

  The filter coefficient calculation unit 422b estimates the transfer function of the feedback transfer system 100 based on the residual signal that is the signal output from the addition unit 423 and the audio signal x (k−τ) input from the delay unit 421. The filter coefficient is self-set by correcting the filter coefficient of the digital filter 422a in accordance with (simulating) the estimated transfer function. This correction is performed so that the residual signal becomes as small as possible, and is performed at predetermined time intervals (for example, every several tens of microseconds to several seconds).

  The estimation of the transfer function of the feedback transfer system 100 and the correction of the filter coefficient are performed using an adaptive algorithm. Specifically, the correction amount calculated from the residual signal and the audio signal x (k−τ) using the adaptive algorithm is multiplied by the step size, and the adaptive filter coefficient before update is multiplied by the forgetting coefficient. Are added to calculate a filter coefficient. The filter coefficient is corrected to the calculated value. As an adaptive algorithm, for example, a learning identification method, an LMS method, a projection method, an RLS method, or the like can be applied. In this way, the audio signal x (k) is supplied to the delay unit 421-adaptive filter 422 in parallel with the output from the speaker 3, so that the simulated signal do (k) output from the adaptive filter 422. Can be approximated to the feedback sound signal d (k).

  The adder 423 receives the simulated signal do (k) and the microphone input signal. The adder 423 outputs a residual signal obtained by removing the simulated signal do (k) from the microphone input signal to the amplifier 41 and outputs the residual signal as a residual signal to the filter coefficient calculator 422b. As a result, the feedback sound component can be removed from the microphone input signal, and howling can be prevented.

  The howling canceller control unit 5 is constituted by, for example, a microcomputer and receives a microphone input signal and a residual signal. The howling canceller control unit 5 executes the change level determination program to determine whether the change level of the transfer function of the feedback transfer system 100 is greater than a predetermined level using the microphone input signal and the residual signal. A process (determination process) is executed.

  For example, in the determination process, the signal level of the microphone input signal and the residual signal is detected, and after performing the process of comparing the two signal levels, the signal level of the residual signal becomes a predetermined level from the signal level of the microphone input signal. It is determined whether or not the signal is larger than the microphone input signal (for example, whether the signal is larger than the microphone input signal, or whether the signal is larger than several times (three times) the microphone input signal). When the signal level of the residual signal is higher than the signal level of the microphone input signal by a predetermined level, it is determined that the change level of the transfer function of the feedback transfer system 100 is higher than the predetermined level. This predetermined level is a level indicating that the change level of the feedback transfer system 100 has changed so that the degree of coincidence with the transfer function after the change of the adaptive filter 422 is larger in the initial state than in the state immediately before the change. It is.

  The howling canceller control unit 5 executes a process (initialization process) for setting the adaptive filter 422 to an initial state when the determination process determines that the level of change in the transfer function of the feedback transfer system 100 exceeds a predetermined level. . The initial state is that the adaptive filter 422 is reset. “Reset” sets the first tap in the digital filter 422a to “1” or an integer smaller than “1”, sets the remaining taps to “0”, and sets the value of the pre-update filter coefficient in the filter coefficient calculation unit 422b. This is a state set to 0. Instead of resetting in the initial state, a filter coefficient set in a normally considered transfer function may be set as an initial value, and a preset may be returned to the initial value. In the preset, the value of the filter coefficient is returned to the initial value, and the value of the pre-update filter coefficient is returned to the initial value.

  When the determination process determines that the change level of the transfer function of the feedback transfer system 100 exceeds the predetermined level, the degree of coincidence with the transfer function after the change of the adaptive filter 422 is equal to the initial state. This is a case where the change level of the transfer function changes as the degree of coincidence becomes greater than immediately before the change. In such a case, since the adaptive filter 422 is returned to the initialized state, the adaptive filter 422 can be quickly adapted to the transfer function after the change as compared with the prior art that does not return to the initialized state. Further, during the period until the adaptive filter 422 identifies the changed transfer function, the simulated signal do (k) that is significantly different from the feedback sound signal d (k) is subtracted from the microphone input signal, so It is possible to effectively prevent inconveniences such as the removal of even the sound component or the addition of a noise component.

  The howling canceller control unit 5 notifies the gain control unit 6 of the execution of the initialization process when executing the initialization process. The howling canceller control unit 5 performs frequency analysis on the residual signal, and detects a specific frequency estimated to be a feedback sound component from the analysis result. The howling canceller control unit 5 notifies the gain control unit 6 of the detected specific frequency along with the execution of the initialization process.

  The gain control unit 6 is composed of, for example, a microcomputer, and corresponds to the gain setting unit of the present invention, and controls the gain of the signal level amplified by the amplification unit 41. The gain control unit 6 corresponds to the gain setting unit of the present invention, and executes a gain change processing program so that the amplification unit 41 amplifies the signal when the howling canceller control unit 5 is notified of the execution of the initialization process. A process of changing the level gain (gain changing process) is executed. In the gain changing process, for a specific frequency estimated to be a feedback sound component, the gain for amplifying the signal level is reduced to a predetermined level, and after a predetermined time when it is determined that the identification of the adaptive filter 422 has progressed thereafter. To normal level. The predetermined level at which the gain is lowered is a level at which the feedback sound does not cause discomfort to the listener (for example, a level smaller than the speaker's real voice). At the same time, the predetermined level is preferably a level that does not impair the sound quality of the audio signal to be amplified. The predetermined time for which it is determined that the identification of the adaptive filter 422 has progressed is set in advance based on experience values. When the level of the residual signal detected by the howling canceller control unit 5 becomes smaller than a preset threshold value, the gain control unit 6 determines that the identification of the adaptive filter 422 has progressed, and the gain is increased to the normal level. You can return it. When the adaptive filter 422 is initialized, the audio signal input to the howling canceller 42 is output as it is as the simulated signal do (k), and thus the feedback sound signal d (k) is sufficiently removed by the howling canceller 42. However, since it is amplified by the amplifying unit 41, howling is likely to occur. In such a case, howling is prevented by the gain control unit 6 executing the gain changing process.

  FIG. 2 is a flowchart showing an example of determination processing, initialization processing, and gain change processing executed by the loudspeaker 1 shown in FIG. This processing is executed when the main power source of the loudspeaker 1 and the power source of the microphone 2 are turned on. First, the howling canceller control unit 5 starts detecting the change level of the transfer function of the feedback transfer system 100 (S1). Specifically, howling canceller control unit 5 detects the change level of the transfer function of feedback transfer system 100 by comparing the signal levels of the microphone input signal and the residual signal.

  Then, the howling canceller control unit 5 determines whether or not the detected change level of the transfer function is greater than a predetermined level (S2). Specifically, it is determined whether the signal level of the residual signal is higher than a predetermined level from the signal level of the microphone input signal. If the determination result is affirmative, it is determined that the change level is higher than the predetermined level. . By executing step S2, it is possible to accurately determine whether or not the change level is greater than the predetermined level using a simple process that simply compares the signal level of the microphone input signal and the signal level of the residual signal. Here, when the residual signal is larger than the microphone input signal, the feedback sound signal d (k) and the simulation signal do (k) are largely inconsistent, and instead, the surplus signal for the simulation signal do (k). (Noise component) is added, and the adaptive filter 422 has a greater degree of mismatch with the transfer function than in the initial state. By executing step S2, it is possible to accurately determine whether or not the state is in such a state by a simple process that merely determines whether or not the residual signal is larger than a predetermined level from the microphone input signal.

  If it is not determined that the change level is greater than the predetermined level (NO in S2), the howling canceller control unit 5 returns to step S1 and compares the signal level for the new signal acquired. On the other hand, if it is determined that the change level is greater than the predetermined level (YES in S2), howling canceller control unit 5 initializes adaptive filter 422 (S3). Specifically, howling canceller control unit 5 outputs a reset signal to adaptive filter 422. When the reset signal is input to the adaptive filter 422, the first tap of the digital filter 422a is set to “1” or an integer smaller than “1”, and the remaining taps are set to “0”. The pre-update filter coefficient of the calculation unit 422b is set to zero. When the howling canceller control unit 5 transmits the reset signal, the howling canceller control unit 5 notifies the gain control unit 6 of the transmission of the reset signal.

  When the gain control unit 6 is notified of the transmission of the reset signal, the gain control unit 6 starts executing the gain changing process (S4). Specifically, in the gain changing process, in the audio signal input to the amplifying unit 41, the gain for amplifying the signal level of the specific frequency estimated to correspond to the feedback sound component by the amplifying unit 41 is set to a level smaller than the normal level. After that, it returns to the normal level as the predetermined time elapses. By executing step S4, it is possible to effectively prevent the occurrence of howling due to the adaptive filter 422 being in the initialized state. When the adaptive filter 422 is in the initial state, the simulation signal do (k) does not simulate the feedback sound signal d (k), and thus the feedback sound component cannot be sufficiently removed from the microphone input signal. For this reason, howling is likely to occur until the adaptation to the transfer function of the feedback transfer system 100 proceeds (the adaptive filter 422 simulates the transfer function) after the adaptive filter 422 is initialized. In step S2, since the gain for amplifying the signal level of the specific frequency estimated to correspond to the feedback sound component is made smaller than the normal level during the period in this state, howling can be effectively prevented.

  And the gain control part 6 returns this process to step S1. This process ends when the main power source of the loudspeaker 1 or the microphone 2 is turned off, and continues to be executed until the power is turned off.

  According to the above configuration, whether or not the change level of the feedback transmission system 100 is greater than the predetermined level is determined by executing the determination process by the howling canceller control unit 5. The case where the change level of the feedback transfer system 100 is determined to be greater than a predetermined level by the determination process is such that the degree of coincidence with the transfer function after the change of the adaptive filter 422 becomes larger in the initial state than in the state immediately before the change. This is a case where the change level of the feedback transmission system 100 changes. In this case, since the adaptive filter 422 is returned to the initial state by the howling canceller control unit 5, in the first embodiment, the adaptive filter 422 is quickly transmitted after the change compared to the conventional technique that does not return to the initial state. It can be adapted to functions. Further, during the period until the adaptive filter 422 adapts to the change in the transfer function of the feedback transfer system 100 after the change, the simulated signal do (k) significantly different from the feedback sound signal d (k) is subtracted from the microphone input signal. Thus, it is possible to effectively prevent the occurrence of inconveniences such as removal of the speech component that is the target of loudening or the addition of a surplus signal (noise component) corresponding to the simulated signal do (k).

(Second Embodiment)
A loudspeaker 1A according to a second embodiment of the present invention will be described below with reference to FIG. In the first embodiment, the howling canceller control unit 5 detects the change level of the transfer function of the feedback transfer system 100 by comparing the signal levels of the microphone input signal and the residual signal. In the embodiment, the howling canceller control unit 5A detects the correction amount calculated by the filter coefficient calculation unit 422b as indicated by a one-dot chain line arrow in FIG. 1, and returns when the correction amount exceeds a preset threshold value. It is determined that the change level of the transmission system 100 is higher than the predetermined level (or higher than the predetermined level). As described above, the correction amount is calculated using an adaptive algorithm based on the residual signal and the audio signal x (k−τ), and is greatly calculated when the signal level of the residual signal is large. As described above, when the signal level of the residual signal is large compared to the microphone input signal, the simulated signal do (k) is significantly different from the feedback sound signal d (k). k) is added to the microphone input signal as a surplus signal, and the adaptive filter 422 is in a state where the degree of mismatch with the transfer function is larger than in the initial state. In the second embodiment, the value of the correction amount obtained by calculating the value of the correction amount when estimated to be in such a state from the experience value is set as the threshold value.

  For this reason, in the second embodiment, when the change level of the transfer function of the feedback transfer system 100 is larger than the predetermined level (or higher than the predetermined level) using a simple process of simply comparing the correction amount with the threshold value. Can be accurately determined. Since the definition of the predetermined level and the configuration of the loudspeaker 1A other than the howling canceller control unit 5A are the same as those in the first embodiment, description thereof will be omitted.

(Third embodiment)
The loudspeaker 1B according to the third embodiment of the present invention will be described below with reference to FIGS. In the third embodiment, a plurality of microphones 20 and speakers 30 are connected to the loudspeaker 1B, and a howling canceller corresponding to the switching of the set to be turned on / off among these and the change of the switching pattern. The setting of 420 is changed from the first and second embodiments.

  FIG. 3 is a block diagram schematically showing the configuration of a loudspeaker 1B according to the third embodiment. FIG. 4 is a diagram illustrating an example of the delay time table T. As illustrated in FIG.

  A pair of microphones 20A, 20B and speakers 30A, 30B is connected to the loudspeaker 1B, and one of them is selectively turned on. That is, any one of the microphone 20A-speaker 30A group, the microphone 20A-speaker 30B group, the microphone 20B-speaker 30A group, and the microphone 20B-speaker 30B group is selectively turned on. In the following description, when the microphones 20A and 20B and the speakers 30A and 30B are not distinguished from each other, they are described as the microphone 20 and the speaker 30. When they are distinguished from each other, additional notes A and B are added.

  The loudspeaker 1B includes a loudspeaker processing unit 40, a howling canceller control unit 50, a gain control unit 6, a switching control unit 7, and a storage unit 8. The loudspeaker processing unit 40 includes an amplifying unit 410 and a howling canceller 420. The amplifying unit 410 amplifies the audio signal input from the microphones 20B and 20A and outputs the amplified audio signal to the speakers 30B and 30A. Since the amplification unit 410 has the same configuration as the amplification unit 41, the description of the configuration is omitted.

  The howling canceller 420 prevents howling by removing the feedback sound component between the microphones 20A and 20B and the speakers 30A and 30B. That is, when the combination of the microphone 20A and the speaker 30A is turned on, the feedback sound component output from the speaker 30A and input to the microphone 20A via the feedback transmission system 100A is derived from the audio signal input from the microphone 20A. Removed. When the pair of the microphone 20A and the speaker 30B is turned on, an audio signal input from the microphone 20A is a feedback sound component output from the speaker 30B and input to the microphone 20A via a feedback transmission system (not shown). Removed from. When the pair of the microphone 20B and the speaker 30A is turned on, the sound signal input from the microphone 20B is the feedback sound component output from the speaker 30A and input to the microphone 20B via the feedback transmission system (not shown). Removed from. When the pair of the microphone 20B and the speaker 30B is turned on, the feedback sound component output from the speaker 30B and input from the microphone 20B via the feedback transmission system 100B is removed from the audio signal input from the microphone 20B. The

  Since the configuration and signal processing of the howling canceller 420 are the same as those of the howling canceller 42, the description of the configuration and signal processing is omitted.

  The switching control unit 7 is constituted by, for example, a microcomputer or the like, and executes a switching program to switch on / off setting (switching pattern) between the microphones 20A and 20B and the speakers 30A and 30B (switching pattern change). Process). In the switching pattern change process, for example, the switching pattern is set so that one of the pair of microphones 20A, 20B and speakers 30A, 30B is selectively turned on according to the switching setting received from the user by an operation unit (not shown). Is done. Here, when the switching pattern is changed, the distance between the speaker 30 that is turned on and the microphone 20 changes greatly, so that the transfer function for transmitting the feedback sound and the delay time τ change greatly.

  For example, when the microphone 20A and the speaker 30A are turned on, the transfer function of the feedback sound input to the microphone 20A is for the feedback transfer system 100A. When the switching pattern is changed and the microphone 20B is turned on instead of the microphone 20A, since the feedback sound output from the speaker 30A is input to the microphone 20B, the distance of the transmission route of the feedback sound is greatly increased. Be changed. For this reason, the transfer function and the delay time τ of the feedback sound component input to the howling canceller 420 change.

  The storage unit 8 is composed of, for example, a rewritable memory such as a flash memory or a hard disk, and stores a delay time table T as shown in FIG. 4 in advance. The delay time table T is a table for registering the switching pattern switched by the switching control unit 7 and the delay time τ (conditions of the howling canceller 420) of the delay unit 421 corresponding to the switching pattern. As described above, when the distance until the feedback sound is transmitted after the switching pattern is changed, the delay time τ of the feedback sound input to the microphone 20 changes. The delay time τ of the delay unit 421 corresponding to the change of the switching pattern is registered in the delay time table T in association with the switching pattern. The delay time table T may be stored from the shipping stage when the distance between the speaker 30 and the microphone 20 is fixed, and is input by the operator when the distance is not fixed.

  The howling canceller control unit 50 executes the determination process and the initialization process as in the first and second embodiments, and executes the pattern change corresponding initialization process when there is a change in the switching pattern. As described above, when the switching control unit 7 changes the on / off switching pattern, the transfer function of the feedback sound signal d (k) input to the howling canceller 420 in response to the switching pattern change. And there is a change in the delay time τ. The pattern change corresponding initialization process initializes the adaptive filter 422 of the howling canceller 420 to which the changed feedback sound signal d (k) is input, and executes the process of changing the delay time τ of the delay unit 421. It is processing. In the processing for changing the delay time τ, the howling canceller control unit 50 refers to the delay time table T, the delay time τ corresponding to the change of the switching pattern is acquired, and the delay time of the delay unit 421 is added to the acquired delay time τ. This is a process for changing τ. Here, the initialization process of the adaptive filter 422 is synonymous with the first embodiment. Note that when the initialization process is executed for the adaptive filter 422, the howling canceller control unit 50 notifies the gain control unit 6 of the execution of the initialization process as in the first embodiment. The gain controller 6 that has received this notification executes a gain changing process in the same manner as in the first embodiment.

  When the switching pattern is changed, there is a high possibility that the degree of coincidence between the transfer function of the feedback sound component after this change and the adaptive filter 422 is larger in the initial state than in the state immediately before the change. For this reason, when the switching pattern is changed as described above, the adaptive filter 422 is initialized, so that the adaptive filter 422 can be quickly adapted to the change of the switching pattern as compared with the conventional technique in which the adaptive filter 422 is not initialized. Filter 422 can be adapted to the changed transfer function. In addition, when the change of the switching pattern is performed, the adaptive filter initialization process is performed, so that the feedback sound is returned during the period from when the transfer function changes until the adaptive filter 422 identifies the change of the transfer function. An inconvenience (for example, speech signal component (for example, a speaker's voice)) to be amplified is removed by removing a simulated signal do (k) significantly different from the signal d (k) from the microphone input signal. On the other hand, a surplus signal equivalent to the simulation signal do (k) is effectively prevented).

  Further, when there is a change in the delay time τ of the feedback sound, if the setting of the delay time τ of the delay unit 421 is not changed so as to correspond to this change, the feedback sound signal d (k) input from the microphone 20 is changed. And the simulation signal do (k) simulating this are input to the adder 423 at different timings. If this timing is different, the feedback sound component cannot be removed from the audio signal input from the microphone 20. Therefore, when the switching pattern is changed by executing the delay time τ changing process, the delay time τ corresponding to the switching pattern can be quickly acquired by referring to the delay time table T. By changing the setting of the delay unit 421 to the acquired delay time τ, the setting of the delay unit 421 can be quickly changed corresponding to the change of the switching pattern.

  FIG. 5 is a flowchart showing determination processing, initialization processing, and switching pattern change initialization processing executed by the loudspeaker 1B shown in FIG. In this drawing, steps in which the same processing as the processing shown in FIG. 2 is executed are denoted by the same reference numerals and description thereof is omitted. In this process, after the start of detection of the change level of the transfer function (step S1), the howling canceller control unit 50 determines whether or not a level change event has occurred. In this level change event, it is determined that the change level of the transfer function detected by the howling canceller control unit 50 is higher than the predetermined level (or higher than the predetermined level), and the switching pattern is changed by the switching control unit 7. That is. When the switching control unit 7 changes the switching pattern, the switching control unit 7 notifies the howling canceller control unit 50 of the change, and the howling canceller control unit 50 determines that a level change event has occurred when the notification is made. .

  If it is not determined that a level change event has occurred (NO in S11), the howling canceller control unit 50 returns this processing to step S1, and detects the signal change level for the newly captured signal. If it is determined that a level change event has occurred (YES in S11), the howling canceller control unit 50 determines that the change level is greater than a predetermined level as in the first embodiment. Steps S3 and S4 are executed, and then the process returns to Step S1. If it is a level change event that the switching pattern change is notified, the howling canceller control unit 50 executes step S12 described later.

  In step S12, the howling canceller control unit 50 refers to the delay time table T and acquires the delay time τ corresponding to the changed switching pattern. Referring to FIG. 4, in the state where the pair of speaker 30A and microphone 20A is turned on, the process of this step is performed by taking as an example a case where the switching pattern is changed so that microphone 20B is turned on instead of microphone 20A. explain.

  In this example, the feedback sound signal d (k) that is simulated by the howling canceller 420 is the feedback sound that is output from the speaker 30A and returned to the microphone 20A, and is output from the speaker 30A and returned to the microphone 20B. The feedback sound signal d (k) is changed. Referring to FIG. 4, in the delay time table T, the delay time τ10 is registered as the delay time τ before the change of the switching pattern, but the delay time τ11 is registered after the change of the switching pattern. Accordingly, the howling canceller control unit 50 acquires the delay time τ11 of the delay unit 421A.

  Thereafter, the howling canceller control unit 50 changes the delay unit 421 to the acquired delay time τ (S13). In the example described above, the howling canceller control unit 50 changes the delay unit 421 from the delay time τ10 to the delay time τ11.

  Thereafter, as in the first embodiment, the initialization process is executed in step S3, the adaptive filter 422 is initialized, the gain change process is executed in step 4, and the gain is changed. It returns to step S1.

  In the above-described third embodiment, even in a configuration in which a plurality of sets of the microphone 20 and the speaker 30 are connected and the on / off setting is switched between the sets, the change in the delay time τ of the feedback sound due to the switching is performed. In response to this, the setting of the delay time τ of the delay unit 421 is changed by the howling canceller control unit 50. As a result, the input timing of the feedback sound signal d (k) and the simulated signal do (k) simulating the feedback sound signal d (k) to the adder 423 can be matched, thereby improving howling. Can be done well. Further, along with the change in the transfer function of the feedback sound due to the change of the switching pattern, the adaptive filter 422 simulating the transfer function before the change is initialized by the howling canceller control unit 50. As a result, the adaptive filter 422 can be quickly adapted to the changed transfer function as compared with the case where the adaptive filter 422 before the change is adapted following the change of the transfer function. Further, during the period from when the transfer function is changed by changing the switching pattern until the adaptive filter 422 identifies the change of the transfer function, a simulated signal do (k) significantly different from the feedback sound signal d (k) An inconvenience (for example, a speech signal component (for example, a speaker's speech)) to be amplified is removed, or a surplus signal corresponding to a simulated signal d (k) is converted into a microphone input signal. Etc.) are effectively prevented.

  Further, the delay time table T in which the delay time τ corresponding to the switching pattern is registered in advance is referred to by the howling canceller control unit 50, and the delay time τ corresponding to the switching pattern is acquired. A delay time τ of 421 is set. For this reason, the delay time τ after the change of the switching pattern can be quickly acquired. Accordingly, since the delay time τ can be changed quickly in response to the change of the switching pattern, howling can be prevented more effectively.

  The embodiment of the present invention can employ the following modifications.

  (1) In the first to third embodiments, the loudspeakers 1, 1A, 1B are configured to be externally attachable to the microphones 2, 20 and the speakers 3, 30, but are not limited thereto and are integrated. The structure which is provided may be sufficient.

  (2) In the first to third embodiments, the gain changing process is executed. However, the gain changing process is not necessarily executed.

  (3) In the first to third embodiments, it is determined whether “exceeds a predetermined level (greater than a predetermined level)” or “exceeds a threshold (greater than a threshold)”. However, it may be determined whether it is “above a predetermined level” or “above a threshold” including a predetermined level and a threshold. Further, the terms “greater than a predetermined level” and “exceeding a threshold value” in claims include the concepts of “above a predetermined level” and “above a threshold value”.

  (4) In the third embodiment, a total of four switching patterns can be set between the four sets of the microphones 20A and 20B and the speakers 30A and 30B. However, the present invention is not limited to this. The speaker 30 and the microphone 20 may be connected to the loudspeaker 1 </ b> B so that a plurality of switching patterns can be set between the plurality of sets.

  (5) In the first to third embodiments, the initialization of the adaptive filter 422a means that the first tap is set to “1” or an integer smaller than “1”, and the remaining taps are set to “0”. However, the present invention is not limited to this. For example, the value of the filter coefficient in the digital filter 422a is set to 0, and the value of the pre-update filter coefficient in the filter coefficient calculation unit 422b is set to 0. Also good. According to this configuration, the feedback sound component is not removed from the microphone input signal in the initialization state.

  (6) In the delay time table T, the delay time corresponding to the switching pattern is registered. In addition, the filter coefficient calculated based on the experience value in advance is registered corresponding to the switching pattern. It may be configured. In this configuration, when the switching pattern is changed, the howling canceller control unit 50 acquires the filter coefficient corresponding to the switching pattern by referring to the delay time table T, and sets the filter coefficient of the adaptive filter 422 as the initial value. The filter coefficient is corrected to the acquired filter coefficient. According to this configuration, since the filter coefficient is calculated in advance based on experience values, compared to the present embodiment in which the adaptive filter 42 is initialized, howling is likely to occur immediately after the switching pattern is changed. Is effectively prevented.

  (7) In the third embodiment, the switching control unit 7 switches which of the microphones 20A and 20B and the speakers 30A and 30B is turned on based on the switching setting received from the user by an operation unit (not shown). However, it is not limited to this configuration. For example, the signal levels between the microphones 20A and 20B may be compared, and the microphone 20 with the higher signal level may be turned on. In this case, which of the speakers 30A and 30B is turned on is set in advance according to the microphone 20 that is turned on, for example. That is, when the microphone 20A is turned on, the speaker 30A is turned on in advance.

It is a block diagram which shows schematic structure of the loudspeaker concerning 1st Embodiment. It is a flowchart which shows an example of the determination process, initialization process, and gain change process which the loudspeaker shown in FIG. 1 performs. It is a block diagram which shows roughly the structure of the loudspeaker concerning 3rd Embodiment. 4 is a diagram illustrating an example of a delay time table T. FIG. It is a flowchart which shows the determination process, initialization process, and switching pattern change corresponding initialization process which the loudspeaker shown in FIG. 3 performs.

Explanation of symbols

1, 1A, 1B-Loudspeaker 2, 20 (20A, 20B)-Microphone 3, 30 (30A, 30B)-Speaker 41, 410-Amplifier (speaker circuit) 42, 420-Howling canceller 5, 5A, 50 -Howling canceller control unit 6-Gain control unit (gain setting unit) 100-Feedback transmission system T-Delay time table

Claims (4)

In a loudspeaker connected to a speaker and a microphone,
A loudspeaker circuit that amplifies the input audio signal and inputs it to the speaker;
An adaptive filter that self-sets a filter coefficient that simulates a transfer function of a feedback transfer system from the speaker to the microphone, and delays an audio signal input to the speaker by a delay time that simulates a delay time of the feedback transfer system; A howling canceller that outputs to the loudspeaker circuit a residual signal that is input to the adaptive filter and subtracts the simulated signal processed by the adaptive filter from the audio signal input from the microphone;
A howling canceller control unit that detects a change level of a transfer function of the feedback transfer system and initializes the adaptive filter when the detected change level is greater than a predetermined level;
With
The howling canceller control unit detects a signal level of a residual signal obtained by removing a simulated signal from the input signal and a voice signal input from the microphone, and the residual signal is at a predetermined level than the voice signal input from the microphone. If beyond large, as the level of change is greater than a predetermined level, the adaptive filter expansion voice device wherein initializing a. In a loudspeaker connected to a speaker and a microphone,
A loudspeaker circuit that amplifies the input audio signal and inputs it to the speaker;
An adaptive filter that self-sets a filter coefficient that simulates a transfer function of a feedback transfer system from the speaker to the microphone, and delays an audio signal input to the speaker by a delay time that simulates a delay time of the feedback transfer system; A howling canceller that outputs to the loudspeaker circuit a residual signal that is input to the adaptive filter and subtracts the simulated signal processed by the adaptive filter from the audio signal input from the microphone;
A howling canceller control unit that detects a change level of a transfer function of the feedback transfer system and initializes the adaptive filter when the detected change level is greater than a predetermined level;
With
The howling canceller control unit detects a correction amount of the filter coefficient of the adaptive filter, and when the detected correction amount is greater than or equal to a threshold value or greater than the threshold value, the adaptive filter is initially determined that the change level is greater than a predetermined level. expansion voice equipment characterized in that the reduction. When the adaptive filter is initialized, it further includes a gain setting unit that reduces a gain for amplifying the residual signal by the loudspeaker circuit unit, compared to a case where the adaptive filter is not initialized. The loudspeaker according to claim 1 or 2 . Connected to a plurality of sets of the microphone and the speaker;
A switching unit that selectively switches a pair for turning on / off from a plurality of pairs of the microphone and the speaker;
A table storage unit for storing a table for registering a switching pattern by the switching unit and the delay time corresponding to the switching pattern;
The howling canceller control unit initializes the adaptive filter when the switching unit is changed by the switching unit, acquires a delay time corresponding to the switching pattern of the switching unit from the table, and The loudspeaker according to any one of claims 1 to 3 , wherein the howling canceller is set to the acquired delay time.

Images