Sound amplification device, howling suppression device and howling suppression method

Abstract

The present invention relates to a sound amplification device, a howling suppression device, and a howling suppression method. The sound amplification device includes: a speaker that generates a reproduced sound; a microphone that generates a microphone collected sound signal; a first filter that generates a pseudo echo signal based on a speaker drive signal; and an echo cancellation unit that obtains a difference between the microphone collected sound signal and the pseudo echo signal to generate an echo cancellation signal. The sound amplification device includes: a second filter that whitens an input sound included in the speaker drive signal; a third filter that whitens an input sound included in the microphone collected sound signal; a first adaptive filter that uses the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal to estimate the propagation characteristics from the speaker to the microphone; a component that repeatedly updates the filter coefficient of the first filter based on the filter coefficient identified by the first adaptive filter; and a frequency shift unit that performs a frequency shift on the echo cancellation signal to generate a speaker drive signal.

Description

Technical Field

[0001] The present invention relates to a sound amplification device, a howling suppression device, and a howling suppression method, and more particularly to an improvement in a sound amplification device in which a speaker and a microphone are arranged in the same acoustic space. Background Art

[0002] In a sound amplification device in which a speaker and a microphone are arranged in the same acoustic space, the reproduced sound output from the speaker reaches the microphone and becomes an echo sound. It is known that howling occurs when the loop gain of such an echo sound exceeds 1.

[0003] In an existing sound amplification device, there is known a sound amplification device having a notch filter in a signal processing unit to which a speaker and a microphone are connected to suppress an echo sound. A notch filter is a band-stop filter that blocks a signal from passing through in a narrow frequency band. If the frequency of the echo sound is within the blocking band of the notch filter, the echo sound can be suppressed and howling can be prevented. However, for other frequency bands, the echo sound cannot be blocked and howling cannot be prevented. In addition, if the blocking bandwidth is increased, there is a problem that the gain of the sound amplification device is reduced.

[0004] In addition, in an existing sound amplification device, there is known a sound amplification device that uses a feedback canceller to suppress an echo sound (for example, Patent Documents 1 and 2).

[0005] Figure 3 FIG. 18 is a diagram showing a structural example of an existing sound amplification device 200 including a feedback canceller 30. The speaker 1 inputs a speaker drive signal u(n) via a D / A converter 11 and a speaker amplifier 12, and outputs the reproduced sound to the acoustic space 4. The microphone 2 collects the sound in the acoustic space and generates a microphone sound collection signal y(n) via a microphone amplifier 21 and an A / D converter 22.

[0006] The speaker 1 and the microphone 2 are arranged in the same acoustic space. Therefore, the microphone sound collection signal y(n) is composed of an echo sound x(n) that loops from the speaker 1 to the microphone 2 and other input sounds v(n).

[0007] The feedback canceller 30 is composed of a first filter 301 and an echo cancellation unit 302. The first filter 301 obtains a pseudo echo signal e(n) = W * u(n) from the speaker drive signal u(n) using filter coefficients W, and the echo cancellation unit 302 subtracts the pseudo echo signal e(n) from the microphone sound collection signal y(n) to obtain an echo cancellation signal d(n). The echo cancellation signal d(n) is delayed in a delay unit 33 and becomes a new speaker drive signal u(n).

[0008] If the filter coefficient W is consistent with the propagation characteristic Wo from the speaker 1 to the microphone 2, the feedback canceller 30 can suppress the echo sound x(n) included in the microphone sound collection signal y(n), thereby preventing the generation of howling. Therefore, it is important to obtain the filter coefficient W that is consistent with the propagation characteristic Wo.

[0009] Prior art documents

[0010] Patent documents

[0011] Patent document 1: International Publication No. WO2010 / 106820;

[0012] Patent document 2: European Patent Application Publication No. 1675374 Gazette. Summary of the invention

[0013] Problems to be solved by the invention

[0014] The propagation characteristic Wo also changes with the passage of time during the use of the sound amplification device 200. Therefore, it is necessary to repeatedly estimate the propagation characteristic Wo using an application filter and repeatedly update the filter coefficient W of the feedback canceller 30 based on the identified filter coefficient.

[0015] In addition, there is a strong correlation between the input sound v(n) and the speaker drive signal u(n). Therefore, in the estimation of the propagation characteristic Wo using an adaptive filter, there is a problem of generating a bias error due to the above-mentioned correlation. That is, there is a problem that it is difficult to identify an appropriate filter coefficient W based on the speaker drive signal u(n) and the microphone sound collection signal y(n).

[0016] Therefore, it is considered to perform linear prediction of the input sound v(n), whiten the input sound v(n) included in the speaker drive signal u(n) and the microphone sound collection signal y(n) respectively, make the input sound v(n) and the speaker drive signal u(n) uncorrelated, and then perform the estimation of the propagation characteristic Wo using an adaptive filter. Thus, an appropriate filter coefficient W can be identified.

[0017] However, in the case of adopting this method, for example, in the case where the input sound v(n) is composed of only a single frequency component such as a sine wave, it is not possible to use a filter to whiten the input sound v(n). As a result, there is a problem as follows: a correlation is generated between the input sound v(n) and the speaker drive signal u(n), and a bias error is generated in the identified filter coefficient W. That is, there is a problem that howling cannot be stably suppressed.

[0018] The present invention has been completed in view of the above circumstances, and an object thereof is to provide a sound amplification device that suppresses howling without reducing the gain. In addition, an object is to provide a sound amplification device that stably suppresses howling. Furthermore, an object is to provide a howling suppression device and a howling suppression method that can be applied to such a sound amplification device.

[0019] Solution for solving the problem

[0020] The sound amplification device according to the first embodiment of the present invention includes: a speaker that generates a reproduced sound based on a speaker drive signal and outputs it to an acoustic space; a microphone that collects the reproduced sound and an input sound from the acoustic space and generates a microphone collected sound signal; a first filter that generates a pseudo echo signal corresponding to an echo sound that reaches the microphone in the reproduced sound based on the speaker drive signal; and an echo cancellation unit that obtains a difference between the microphone collected sound signal and the pseudo echo signal and generates an echo cancellation signal, wherein the sound amplification device includes: a second filter that whitens the input sound included in the speaker drive signal; a third filter that whitens the input sound included in the microphone collected sound signal; a first adaptive filter that uses the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal to estimate the propagation characteristics from the speaker to the microphone in the acoustic space; a component that repeatedly updates the filter coefficient of the first filter based on the filter coefficient identified by the first adaptive filter; and a time-varying processing unit that performs time-varying processing on the echo cancellation signal to generate the speaker drive signal.

[0021] By adopting such a configuration, the first adaptive filter can estimate the propagation characteristics from the speaker to the microphone in the acoustic space, and update the filter coefficient of the first filter based on the filter coefficient identified by the first adaptive filter. Therefore, even when the propagation characteristics change during the use of the sound amplification device, howling suppression following the change can be performed.

[0022] In addition, the first adaptive filter uses the speaker drive signal that whitens the input sound as a reference signal and the microphone collected sound signal that whitens the input sound as a desired signal to estimate the propagation characteristics from the speaker to the microphone in the acoustic space. Therefore, it is possible to suppress the bias error caused by the correlation between the speaker drive signal and the input sound, and thus identify more appropriate filter coefficients. Therefore, howling suppression can be effectively performed.

[0023] In addition, the time-varying processing unit performs time-varying processing on the echo cancellation signal to generate a speaker drive signal. Thus, even when the input sound consists of only specific frequency components and the speaker drive signal cannot be made uncorrelated with the input sound, for example, the correlation between the speaker drive signal and the input sound can be reduced, thereby suppressing the generation of bias errors. Therefore, howling suppression can be stably performed.

[0024] The sound amplification device according to the second embodiment of the present invention further includes, in addition to the above structure: a second adaptive filter that identifies filter coefficients for whitening the input sound with reference to the echo cancellation signal; and a component that repeatedly updates the filter coefficients of the second filter and the third filter based on the filter coefficients identified by the second adaptive filter.

[0025] By adopting such a structure, the filter coefficients of the second filter and the third filter can be repeatedly updated based on the filter coefficients identified by the second adaptive filter. Therefore, even when the input sound changes, howling suppression that follows the change can be performed.

[0026] The sound amplification device according to the third embodiment of the present invention is configured, in addition to the above structure, such that the time-varying processing unit is a frequency shift processing unit that generates the speaker drive signal by shifting the frequency of the echo cancellation signal.

[0027] The sound amplification device according to the fourth embodiment of the present invention is configured, in addition to the above structure, such that the time-varying processing unit is a phase shift processing unit that generates the speaker drive signal by shifting the phase of the echo cancellation signal.

[0028] The sound amplification device according to the fifth embodiment of the present invention uses a first filter to generate a pseudo echo signal corresponding to the echo sound that propagates from the speaker to the microphone via the acoustic space based on the speaker drive signal, and removes the echo sound from the microphone sound collection signal based on the pseudo echo signal to generate the speaker drive signal. The microphone sound collection signal is generated by collecting the echo sound and the input sound from the acoustic space. The howling suppression device includes: a second filter that whitens the input sound included in the speaker drive signal; a third filter that whitens the input sound included in the microphone sound collection signal; a first adaptive filter that uses the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal to estimate the propagation characteristics from the speaker to the microphone in the acoustic space; a component that repeatedly updates the filter coefficients of the first filter based on the filter coefficients identified by the first adaptive filter; and a time-varying processing unit that performs time-varying processing on the echo cancellation signal to generate the speaker drive signal.

[0029] The sound amplification device according to the sixth embodiment of the present invention further includes, in addition to the above structure: a second adaptive filter that identifies filter coefficients for whitening the input sound with reference to the echo cancellation signal; and a component that repeatedly updates the filter coefficients of the second filter and the third filter based on the filter coefficients identified by the second adaptive filter.

[0030] The howling suppression method according to the seventh embodiment of the present invention uses a first filter to generate a pseudo echo signal corresponding to the echo sound propagated from the speaker to the microphone via the acoustic space based on the speaker drive signal, removes the echo sound from the microphone sound collection signal based on the pseudo echo signal, and generates the speaker drive signal. The microphone sound collection signal is generated by collecting the echo sound and the input sound from the acoustic space. The howling suppression method includes: a step of whitening the input sound included in the speaker drive signal using a second filter; a step of whitening the input sound included in the microphone sound collection signal using a third filter; a step of using a first adaptive filter to estimate the propagation characteristics from the speaker to the microphone in the acoustic space, with the output signal of the second filter as the reference signal and the output signal of the third filter as the desired signal; a step of repeatedly updating the filter coefficients of the first filter based on the filter coefficients identified by the first adaptive filter; and a step of performing time-varying processing on the echo cancellation signal to generate the speaker drive signal.

[0031] The howling suppression method according to the eighth embodiment of the present invention further includes, in addition to the above structure: a step of using a second adaptive filter to identify filter coefficients for whitening the input sound with reference to the echo cancellation signal; and a step of repeatedly updating the filter coefficients of the second filter and the third filter based on the filter coefficients identified by the second adaptive filter.

[0032] Advantageous Effects of the Invention

[0033] According to the present invention, it is possible to provide a sound amplification device that suppresses howling without reducing the gain. In addition, it is possible to provide a sound amplification device that stably suppresses howling. Furthermore, it is possible to provide a howling suppression device and a howling suppression method that can be applied to such a sound amplification device. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a block diagram showing a structural example of a sound amplification device 100 according to an embodiment of the present invention;

[0035] Figure 2 is a block diagram showing Figure 1 an example of the detailed structure of the howling suppression device 3;

[0036] Figure 3 FIG. 3 is a diagram showing a structural example of a conventional sound amplification device 200 including a feedback canceller 30. DETAILED DESCRIPTION

[0037] Figure 1 FIG. 4 is a block diagram showing a structural example of a sound amplification device 100 according to an embodiment of the present invention. The sound amplification device 100 includes a speaker 1, a D / A converter 11, a speaker amplifier 12, a microphone 2, a microphone amplifier 21, an A / D converter 22, and a howling suppression device 3.

[0038] The speaker 1 is a device that converts a speaker drive signal u(n) into sound and outputs it to the acoustic space 4. The speaker drive signal u(n) is a digital signal at discrete time n and is composed of sound information output from the speaker 1. The speaker drive signal u(n) is converted into an analog sound signal in the D / A converter 11, amplified in the speaker amplifier 12, and then converted into sound by the speaker 1 and output as reproduced sound to the acoustic space 4.

[0039] The microphone 2 is a device that converts sound collected in the acoustic space 4 into a microphone sound collection signal y(n). The microphone sound collection signal y(n) is a digital sound signal at discrete time n and is composed of sound information collected by the microphone 2. The analog sound signal output from the microphone 2 is amplified by the microphone amplifier 21 and converted into a digital signal by the A / D converter 22 to become the microphone sound collection signal y(n).

[0040] The speaker 1 and the microphone 2 are arranged in the same acoustic space 4. Therefore, in the microphone sound collection signal y(n), the reproduced sound output from the speaker 1 and reaching the microphone 2 after propagating in the acoustic space 4 is included. That is, the microphone sound collection signal y(n) includes an echo sound x(n) and an input sound v(n). The echo sound x(n) corresponds to the reproduced sound that bypasses from the speaker 1 to the microphone 2, and the input sound v(n) corresponds to the sound generated in the acoustic space other than the reproduced sound. The input sound v(n) is the sound that the sound amplification device 100 should amplify and is an additive interference signal applied to the echo sound x(n).

[0041] If the propagation characteristic of the reproduced sound output from the speaker 1 as the echo sound x(n) to the acoustic space until it reaches the microphone 2 based on the speaker drive signal u(n) is Wo, the echo sound x(n) and the microphone sound collection signal y(n) can be expressed by the following formula:

[0042] 〔Equation 1〕

[0043]

[0044] "*" represents a convolution operation.

[0045] The howling suppression device 3 is a digital processing unit that suppresses the echo sound x(n) included in the microphone sound collection signal y(n) to suppress howling. For example, a DSP (Digital Signal Processor) can be used. The howling suppression device 3 generates a new speaker drive signal u(n) based on the input microphone sound collection signal y(n).

[0046] Figure 2 is a block diagram showing Figure 1 an example of the detailed structure of the howling suppression device 3. The howling suppression device 3 is composed of a feedback canceller 30, a filter identification unit 31, a frequency offset unit 32, and a delay unit 33.

[0047] The feedback canceller 30 is a processing unit that suppresses the echo sound x(n) included in the microphone sound collection signal y(n) based on the speaker drive signal u(n) to generate an echo cancellation signal d(n), and is composed of a first filter 301 and an echo cancellation unit 302.

[0048] The first filter 301 is an FIR (Finite Impulse Response) filter that estimates the echo sound x(n), and obtains a pseudo echo signal e(n)=W*u(n) from the speaker drive signal u(n) using the filter coefficient W. The echo cancellation unit 302 is a differential operation processing unit that subtracts the pseudo echo signal e(n) from the microphone sound collection signal y(n), and generates an echo cancellation signal d(n)=y(n)-e(n) as the microphone sound collection signal y(n) in which the echo sound x(n) is suppressed.

[0049] If the filter coefficient W is consistent with the propagation characteristic Wo from the speaker 1 to the microphone 2, the pseudo echo signal e(n) is consistent with the echo sound x(n), and only the input sound v(n) is output from the echo cancellation unit 302. That is, by suppressing the echo sound x(n) through the feedback canceller 30, the generation of howling can be prevented. Therefore, in order to prevent howling, it is important to make the filter coefficient W consistent with the propagation characteristic Wo.

[0050] The propagation characteristic Wo changes according to the change of the environment in the sound space, and also changes with the passage of time during the use of the amplification device 100. In order to follow such a change in the propagation characteristic Wo, the filter coefficient W identified by the filter identification unit 31 is used as the filter coefficient W of the feedback canceller 30. The filter identification unit 31 repeatedly identifies the filter coefficient W, and the first filter 301 uses the identified filter coefficient W to repeatedly perform the process of updating the filter coefficient.

[0051] The filter identification unit 31 is a processing unit that estimates the propagation characteristics Wo and identifies the filter coefficients W, and is composed of a second adaptive filter 310, second and third filters 311, 312, a first adaptive filter 313, and a subtraction unit 314.

[0052] The second adaptive filter 310 is a linear predictor that identifies the filter coefficients A that whiten the echo cancellation signal d(n). The reference signal of the second adaptive filter 310 is the echo cancellation signal d(n) output by the feedback canceller 30, and the error signal is its own output signal. Therefore, by obtaining the filter coefficients that minimize the error signal, the filter coefficients A that transform the echo cancellation signal d(n) into white noise can be identified. If it is assumed that the filter coefficients W of the first filter 301 are consistent with the propagation characteristics Wo, the echo cancellation signal d(n) is consistent with the input sound v(n), and the filter coefficients A become the coefficients of the whitening filter that whitens the input sound v(n).

[0053] Both the second filter 311 and the third filter 312 are FIR filters that use the filter coefficients A identified by the second adaptive filter 310 to whiten the input sound v(n). The second filter 311 inputs the speaker drive signal u(n) and generates a signal A*u(n) that whitens the input sound v(n) included in the speaker drive signal u(n). On the other hand, the third filter 312 inputs the microphone pickup signal y(n) and generates a signal A*y(n) that whitens the input sound v(n) included in the microphone pickup signal y(n).

[0054] In order to follow the change of the input sound v(n), the second adaptive filter 310 repeatedly performs the identification of the filter coefficients A, and the second filter 311 and the third filter 312 repeatedly perform the process of updating the filter coefficients using the identified filter coefficients A.

[0055] The first adaptive filter 313 is a device that estimates the propagation characteristics Wo and identifies the filter coefficients W. The reference signal of the first adaptive filter 313 is the output signal A*u(n) of the second filter 311, and the desired signal is the output signal A*y(n) of the third filter 312. The subtraction unit 314 is a differential operation processing unit that subtracts the output signal W*A*u(n) of the first adaptive filter 313 from the desired signal A*y(n), and the operation result is input to the first adaptive filter 313 as an error signal.

[0056] The desired signal of the first adaptive filter 313 is the microphone pickup signal A*y(n) that whitens the input sound v(n), and can be expressed by the following formula:

[0057] 〔Equation 2〕

[0058] .

[0059] That is, A*u(n) is supplied as a reference signal to the first adaptive filter 313, and Wo*{A*u(n)} + {A*v(n)} is supplied as a desired signal to the first adaptive filter 313. Both A*u(n) and A*v(n) are signals whitened using the filter coefficient A. Therefore, if the whitening of the input sound v(n) is completed completely, even if there is a strong correlation between u(n) and v(n), A*u(n) and A*v(n) are uncorrelated. Thus, the first adaptive filter 313 can suppress the bias error caused by the correlation between u(n) and v(n), and estimate the propagation characteristic Wo from the speaker 1 to the microphone 2, and identify the appropriate filter coefficient W.

[0060] The frequency shift unit 32 is a processing unit that frequency-shifts the echo cancellation signal d(n) by a predetermined offset amount. The echo cancellation signal d(n) is frequency-shifted in the frequency shift unit 32 and delayed in the delay unit 33 to become a new speaker drive signal u(n). When the input sound v(n) is a periodic signal such as a sine wave, frequency shifting is performed to suppress the generation of bias error.

[0061] If the input sound v(n) is a sine wave without frequency shifting, the second filter 311 and the third filter 312 cannot whiten the input sound v(n) completely. White noise is a signal having frequency components in the entire frequency band, while a sine wave is a signal having no frequency components outside a specific frequency band. Therefore, when the input sound v(n) is a sine wave, the second filter 311 and the third filter 312 cannot whiten the input sound v(n) to make A*u(n) and A*v(n) uncorrelated. As a result, corresponding to the correlation between A*u(n) and A*v(n), a bias error is generated in the filter coefficient W identified by the first adaptive filter 313.

[0062] In contrast, by providing the frequency shift unit 32 to frequency-shift the echo cancellation signal d(n) to generate the speaker drive signal u(n), even if the input sound v(n) is a sine wave, the input signals of the second filter 311 and the third filter 312 have multiple frequency components, and the correlation between A*u(n) and A*v(n) can be reduced.

[0063] In this case, even if it is impossible to make A*u(n) and A*v(n) completely uncorrelated, it is still possible to shift the frequency in each cycle, expand the frequency band of the input signals of the second filter 311 and the third filter 312, and reduce the bias error of the filter coefficient W. Therefore, even when the input sound v(n) consists only of, for example, the sound of an air conditioner and is composed of sine waves, it is possible to suppress the generation of bias error in the identification of the filter coefficient W, thereby stably suppressing howling.

[0064] It is considered that, when ensuring a sufficient offset, the frequency shift itself has an effect of suppressing howling. However, it is considered that if the offset amount of the frequency shift becomes large, it will bring a sense of disharmony to the listener.

[0065] Regarding this, it is considered that, when the frequency shift unit 32 aims to reduce the correlation between A*u(n) and A*v(n), the offset amount can be an extremely small value. Therefore, it is possible to prevent the generation of howling without bringing a large sense of disharmony to the listener staying in the acoustic space 4. According to the experiments of the inventor, it is known that the frequency shift can be, for example, a small offset amount of about 4 Hz. The offset amount of the frequency shift unit 32 can be, for example, 20 Hz or less, and more preferably 10 Hz or less.

[0066] Since the frequency shift unit 32 is used to reduce the correlation between A*u(n) and A*v(n), it can also be replaced by a processing unit that performs a time-varying process other than frequency shift. For example, a phase shift processing unit that performs phase shift can also be used instead of the frequency shift unit 32. The time-varying process is a process that represents the transfer function as a function of time.

[0067] Description of Reference Numerals

[0068] 100: Sound Amplifying Device

[0069] 1: Speaker

[0070] 2: Microphone

[0071] 3: Howling Suppression Device

[0072] 4: Acoustic Space

[0073] 11: D / A Converter

[0074] 12: Speaker Amplifier

[0075] 21: Microphone Amplifier

[0076] 22: A / D Converter

[0077] 30: Feedback Canceler

[0078] 31: Filter Identification Unit

[0079] 32: Frequency offset section

[0080] 33: Delay section

[0081] 301: First filter

[0082] 302: Echo cancellation section

[0083] 310: Second adaptive filter

[0084] 311: Second filter

[0085] 312: Third filter

[0086] 313: First adaptive filter

[0087] 314: Subtraction section

[0088] A, W: Filter coefficients

[0089] Wo: Propagation characteristics

[0090] d(n): Echo cancellation signal

[0091] e(n): Pseudo echo signal

[0092] u(n): Speaker drive signal

[0093] v(n): Input sound

[0094] x(n): Echo sound

[0095] y(n): Microphone pickup signal.

Claims

1. An amplification device, comprising: A speaker that generates a reproduced sound based on a speaker drive signal and outputs it to an acoustic space; A microphone that collects the reproduced sound and an input sound from the acoustic space and generates a microphone collected sound signal; A first filter that generates a pseudo echo signal corresponding to an echo sound that reaches the microphone in the reproduced sound based on the speaker drive signal; And An echo cancellation unit that obtains a difference between the microphone collected sound signal and the pseudo echo signal and generates an echo cancellation signal, Characterized in that the amplification device comprises: A second filter that whitens the input sound included in the speaker drive signal; A third filter that whitens the input sound included in the microphone collected sound signal; A first adaptive filter that uses the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal to estimate the propagation characteristics from the speaker to the microphone in the acoustic space; A component that repeatedly updates the filter coefficient of the first filter based on the filter coefficient identified by the first adaptive filter; A time-varying processing unit that performs time-varying processing on the echo cancellation signal to generate the speaker drive signal; A second adaptive filter that refers to the echo cancellation signal and identifies a filter coefficient for whitening the input sound; and A component that repeatedly updates the filter coefficients of the second filter and the third filter based on the filter coefficient identified by the second adaptive filter.

2. The amplification device according to claim 1, Characterized in that The time-varying processing unit is a frequency shift processing unit that generates the speaker drive signal by shifting the frequency of the echo cancellation signal.

3. The amplification device according to claim 1, Characterized in that The time-varying processing unit is a phase shift processing unit that generates the speaker drive signal by shifting the phase of the echo cancellation signal.

4. A howling suppression device that uses a first filter to generate a pseudo echo signal corresponding to an echo sound that propagates from a speaker to a microphone via an acoustic space based on a speaker drive signal, removes the echo sound from a microphone collected sound signal based on the pseudo echo signal to generate an echo cancellation signal, and generates the speaker drive signal based on the echo cancellation signal, where the microphone collected sound signal is generated by collecting the echo sound and an input sound from the acoustic space, Characterized in that The howling suppression device comprises: A second filter that whitens the input sound included in the speaker drive signal; A third filter that whitens the input sound included in the microphone collected sound signal; A first adaptive filter that uses the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal to estimate the propagation characteristics from the speaker to the microphone in the acoustic space; A component that repeatedly updates the filter coefficient of the first filter based on the filter coefficient identified by the first adaptive filter; A time-varying processing unit that performs time-varying processing on the echo cancellation signal to generate the speaker drive signal; A second adaptive filter that identifies filter coefficients for whitening the input sound with reference to the echo cancellation signal; And A component that repeatedly updates the filter coefficients of the second filter and the third filter based on the filter coefficients identified by the second adaptive filter.

5. A howling suppression method that uses a first filter to generate a pseudo echo signal corresponding to an echo sound propagated from a speaker to a microphone via an acoustic space based on a speaker drive signal, removes the echo sound from a microphone sound collection signal based on the pseudo echo signal to generate an echo cancellation signal, and generates the speaker drive signal based on the echo cancellation signal. The microphone sound collection signal is generated by collecting the echo sound and the input sound from the acoustic space. Characterized in that The howling suppression method includes: A step of whitening the input sound included in the speaker drive signal using a second filter; A step of whitening the input sound included in the microphone sound collection signal using a third filter; A step of using a first adaptive filter to estimate the propagation characteristics from the speaker to the microphone in the acoustic space, with the output signal of the second filter as a reference signal and the output signal of the third filter as a desired signal; A step of repeatedly updating the filter coefficients of the first filter based on the filter coefficients identified by the first adaptive filter; A step of performing time-varying processing on the echo cancellation signal to generate the speaker drive signal; A step of using a second adaptive filter to identify filter coefficients for whitening the input sound with reference to the echo cancellation signal; And A step of repeatedly updating the filter coefficients of the second filter and the third filter based on the filter coefficients identified by the second adaptive filter.

Images