JPH08122424A - Adaptive array - Google Patents

Abstract

PURPOSE: To realize the reception of a substantially intact target signal, using small number of sensors, even if the error of actual incoming direction of the target signal from a predicted direction is large. CONSTITUTION: In each space interruption filter circuit constituting space interruption filter circuit group 12, a subtraction circuit 6m subtracts the output signal of a leak adaptive filter 5m(m=0, 1,..., M-1) employing an output signal of a space pass filter 3 as a reference signal from the signal of each sensor delayed through a delay circuit 4m. The subtraction result is outputted from the space interruption filter circuit and employed as an error signal for updating the coefficients of the leak adaptive filter 5m. Consequently, the target signal is removed sufficiently from the output of the space interruption filter circuit and deterioration of the target signal is suppressed in the output 11 of a multiinput canceler circuit 11.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive array device that spatially selectively receives a signal using a plurality of sensors.

[0002]

2. Description of the Related Art In order to receive only a specific signal from a plurality of signal sources, as an application of adaptive array technology, a voice enhancement device using an adaptive microphone array, a wireless transmitting / receiving device using an adaptive antenna array, etc. are known. .

Here, a conventional example will be described by taking an adaptive microphone array that removes ambient noise and emphasizes voice to pick up the sound.

The microphone array forms a spatial filter by filtering the signals picked up by a plurality of microphones and then adding them. This spatial filter attenuates the ambient noise and receives only the signal, that is, the target signal, which has arrived from the direction defined in advance. The adaptive microphone array is a microphone array that adaptively changes spatial filter characteristics. As a configuration of the adaptive microphone array, “generalized sidelobe canceller” (IEE, Transactions on Antennas and Propagation (IEEE, Tr
transactions on Antennas an
d Propagation), Vol. 30, No. 1, 198
2 years, pages 27-34: "Reference 1") is known. The generalized sidelobe canceller includes a plurality of microphones, a space-pass filter circuit having directivity that allows signals in a predetermined direction to pass by filtering and adding signals from the microphones, and a microphone for each microphone. A space cutoff filter circuit group composed of a plurality of space cutoff filter circuits having a characteristic of cutting off only signals in a predetermined direction by filtering and adding signals, and each of the space cutoff filter circuits. The multi-input canceller circuit removes a component correlated with the output signal of the space cutoff filter circuit from the output of the space pass filter circuit by a plurality of adaptive filters whose outputs are reference signals.

However, there are two space cut-off filters that compose the space cut-off filter circuit group used in "Document 1".
Since it is a fixed filter using two microphone outputs, the range of the direction in which the signal is blocked (blocking direction) is extremely narrow. Therefore, if the arrival direction of the target signal is slightly different from the direction specified in advance, the target signal leaks greatly to the output of the spatial cutoff filter circuit, and a problem occurs that a part of the target signal is also removed in the multi-input canceller circuit. .

In order to solve the problem of the deterioration of the target signal, a method has been proposed in which the spatial cutoff filter circuit group is composed of a plurality of spatial filter circuits using three or more microphone outputs (IEE, transaction). Zon Antennas and Propagation (IEEE, Transactions on Ant
enas and Propagation), No. 4
Volume 0, No. 9, 1992, pp. 1093-1096: "Reference 2"). The point of the method shown in "Reference 2" is to widen the range in the cutoff direction by receiving a large number of spatial sampling points, that is, a large number of microphone outputs, as the respective spatial cutoff filter circuits constituting the spatial cutoff filter circuit group. Using a wide-angle space cutoff filter circuit,
In addition, a leak adaptive filter is used as each adaptive filter forming the multi-input canceller circuit. FIG. 16 is a block diagram showing the configuration of the microphone array proposed in “Document 2”. The conventional technique will be described below with reference to the drawings.

FIG. 16 shows a case where a plurality of microphones having the same characteristics are arranged on a straight line at equal intervals. The case where the position of each sensor is sufficiently far from the plurality of microphones and the wavefront of each signal can be regarded as a plane wave will be described. It is also assumed that the respective signals including both the target signal which is the signal to be picked up and the ambient noise which is the signal to be removed are uncorrelated with each other.

The direction expected as the arrival direction of the target signal, which is obtained by prior knowledge or by a device other than this device, is called a specified direction. Further, a direction in the vicinity of the specified direction that may possibly be the target signal arrival direction is called the vicinity of the specified direction, and a direction other than the specified direction and the vicinity of the specified direction is called a non-specified direction.

First, the operation of FIG. 16 will be described. In the microphone array shown in FIG. 16, the outputs from the _M microphones l _{0 to} l _M-1 are input to the space pass filter circuit 2 and the space cutoff filter circuit 12. The space-pass filter circuit 2 receives signals from the microphones l _{0 to} l _M-1 and performs filtering and addition so as to pass signals in a specified direction. Space-pass filter circuit 2
The output signal of is transmitted to the delay circuit 9. Delay circuit 9
Delays the signal received from the space-pass filter circuit 2,
It is supplied to the multi-input canceller circuit 13. The delay time of the delay circuit 9 must be decided together with the multi-input canceller circuit 13. The determination method will be described together with the description of the multi-input canceller circuit 13. The space cutoff filter circuit group 12 is composed of a plurality of space cutoff filter circuits. Each of the space cutoff filter circuits receives a part of signals from a plurality of output signals of the microphones l _{0 to} l _M-1 , and performs filtering and addition so that signals coming from a specified direction or near the specified direction cancel each other. The output group of the plurality of space cutoff filter circuits is
It is supplied to the multi-input canceller circuit 13. The multi-input canceller circuit 13 operates so as to remove a component having a correlation with the output of the spatial cutoff filter circuit group 12 from the output of the delay circuit 9, and outputs the removal result to the output terminal 11.

The target signal arriving from the specified direction or in the vicinity of the specified direction is included in the output of the delay circuit 9, but not included in the output of the space cutoff filter circuit group 12. On the other hand, the ambient noise signal arriving from the non-specified direction is included in both the output of the delay circuit 9 and the output of the space cutoff filter circuit group 12. Further, by assumption, the signal coming from the specified direction and the signal coming from the non-specified direction are uncorrelated. Therefore, removing the component having a correlation with the output of the spatial cutoff filter circuit group 12 from the output of the delay circuit 9 means that only the ambient noise signal coming from the non-specified direction is removed at the output 11, and the component from the specified direction is removed. This means that the target signal coming from near the specified direction can be obtained.

Next, the components shown in FIG. 16 will be described.

First, the structure of the space-pass filter circuit 2 will be described with reference to FIG. FIG. 17 shows a configuration example of the space pass filter circuit 2. In the space-pass filter circuit 2, the output signals of the microphones l _{0 to} l _M-1 in FIG. 16 are transmitted through the input terminals 201 _{0 to} 201 _M-1 and the corresponding M filter circuits 202 _{0 to} 202 _M-. Supplied to ₁ each. Filter circuit group 202 _{0 to} 202
The output signal group of _M−1 is transmitted to the adder circuit 203. The adder circuit 203 includes a plurality of filter circuits 202 _{0 to} 202 202.
The sum of the signals received from _M-1 is calculated, and the result is transmitted to the output terminal 204 as the output of the space pass filter circuit 2.

The filter circuit 202 _m (m = 1, 2, ...,
M-1) has, for example, an FIR type filter configuration shown in FIG.

Referring to FIG. 18, the filter circuit 202
_{The configuration of m} (m = 1, 2, ..., M-1) will be described. An input signal is supplied to the input terminal 301 via the input terminal 201 _m , and a filter output signal is output to the output terminal 305 and transmitted to the adding circuit 203. The signal input to the input terminal 301 has a delay of 1 sampling period
It is supplied to a delay line with taps made up of a plurality of delay circuits 302 _{0 to} 302 _G-2 (G is an integer constant of 2 or more). Input signal samples supplied to delay circuit 302 ₀ is transferred to the delay circuit adjacent to each clock. The output signal of each delay circuit 302 _g (g = 0, 1, ..., G-2) is supplied to the corresponding multiplication circuit 303 _{g + 1} and the corresponding tap coefficient supplied from the tap coefficient memory circuit 306. Is multiplied. The multiplier circuit 303 _0, an input terminal 30
The signal input to 1 is directly supplied and multiplied by the tap coefficient supplied from the tap coefficient memory circuit 306. The output signals of the multiplication circuits 303 _{0 to} 303 _G-1 are added to the addition circuit 3
04. The addition circuit 304 is a multiplication circuit 303.
The sum of the output signals received from _{0 to} 303 _G-1 is calculated, and the result is output to the output terminal 305. The tap coefficient memory circuit 306 stores tap coefficients such that the entire filter circuits 202 _{0 to} 202 _M-1 have characteristics described later, and the tap coefficients correspond to the multiplication circuits 303 ₀ to 303 ₀ to, respectively.
Supply to 303 _G-1 respectively.

The characteristics of the filter circuits 202 _{0 to} 202 _M-1 are designed so that the output signal of the space-pass filter circuit 2 includes a signal in a specified direction. For example, when the direction orthogonal to the straight line in which the microphones are arranged is taken as the specified direction, the integer constant G is set to 2 in all the filter circuits 202 _{0 to} 202 _M−1 , and the multiplication circuit 30 in FIG.
The tap coefficients to be multiplied in the 3 ₀ may be 1 and set. Other design methods include, for example, (multidimensional digital signal processing,
Prentice Hall, Inc., (Multidimension
nal Digital Signal Procedures
sing, Prentice-Hall, Inc. ) 2
89-315, 1984: "Reference 3" below)
And (IEE, Proceedings of
International Conference on Acoustics Speech and Signal Processing 93, (IEEE, Proceedings of
International Conference
on Acoustics, Speech, and Si
general Processing 93), I-169.
Page 172, 1993: hereinafter "Reference 4").

Next, the structure of the space cutoff filter circuit group 12 in FIG. 16 will be described with reference to FIG.
In the space cutoff filter circuit group 12, “reference 1” uses two microphone outputs per space cutoff filter circuit, but in “document 2”, three or more microphone outputs per space cutoff filter circuit. The microphone output of is used. FIG. 19 shows an example of the configuration of the space cutoff filter circuit shown in "Document 2" when the number of microphone outputs used per space cutoff filter circuit is Q. Q is an integer of 3 or more. The space cutoff filter circuit shown in "Document 1" corresponds to the case where Q is 2.
In the space cutoff filter circuit group shown in FIG. 19, the output signal of the microphone l _m (m = 0, 1, ..., M−1) of FIG. 16 is supplied to the input terminal 901 _m and the output terminal 905.
_The output signal obtained at _m (m = 0, 1, ..., MQ) is
It is supplied to the multi-input canceller circuit 13 of FIG.

Filter circuit 906 _{m, q} (m = 0, 1,
, M-Q, q = 0, 1, ..., Q-1) is input terminal 9
The output signal of the microphone l _{m + q} supplied via 01 _{m + q} is filtered according to the characteristics described later, and the output signal is transmitted to the adding circuit 907 _m . Adder circuit 90
7 _m (m = 0, 1, ..., M−Q) calculates the sum of the output signals of the plurality of filter circuits 906 _{m, q} (q = 0, 1, ..., Q−1), and outputs the result as a space. It is output to the output terminal 905 _m as the output of the cutoff filter circuit. Adder circuit 907 _m (m
= 0, 1, ..., MQ) and a plurality of filter circuits 906
_A space cutoff filter circuit is configured by _{m, q} (q = 0, 1, ..., Q-1).

Filter circuit 906 _{m, q} (m = 0, 1,
, MQ, q = 0, 1, ..., Q-1) is the same as that of the filter circuits 202 _{0 to} 202 _M-1, and can be represented as shown in FIG. However, the integer constant G and the characteristic are as follows: filter circuit 906 _{m, q} (m = 0, 1, ..., MQ, q =
0, 1, ..., Q-1) and filter circuits 202 _{0 to} 202
Different from _M-1 .

Filter circuit 906 _{m, q} (m = 0, 1,
, M−Q, q = 0, 1, ..., Q−1), the output signal of the microphone l _{m + q} of FIG. 16 is supplied to the input terminal 301 via the input terminal 901 _{m + q} , and output. The filter output signal obtained at the terminal 305 is transmitted to the adding circuit 907 in FIG.

Filter circuit 906 _{m, q} (m = 0, 1,
, M−Q, q = 0, 1, ..., Q−1) is designed so that the signal arriving not only in the specified direction but also in the vicinity of the specified direction is blocked at the output terminal 905 _m .
That is, the space cutoff filter circuit group 12 is designed to be a wide-angle space cutoff filter circuit group. The design guideline is described in "Reference 2". The greater the number of microphones, the greater the degree of freedom of the spatial cutoff filter characteristics,
It is possible to widen the range in the blocking direction or increase the attenuation rate of the signal coming from the blocking direction. Thus,
The space cutoff filter circuit group of "Document 2" can cut off the target signal at the output even when the direction of the target signal has an error from the specified direction. As a result, the phenomenon that the target signal is removed at the output terminal of FIG. 16 can be prevented when there is an error between the direction of the target signal and the specified direction.

However, the spatial degree of freedom of the output group of the spatial cutoff filter circuit of "Document 2" shown here is smaller than that of "Document 1". Because the total number of microphones is M
When Q microphone outputs are used for each space cutoff filter circuit group that constitutes the space cutoff filter circuit group 12, the spatial freedom degree of the output of the space cutoff filter circuit group is (M−Q + 1). Yes, “Reference 1” is Q =
This is because “Document 2” corresponds to the case of Q ≧ 3.
In the case of the configuration shown in FIG. 19, the spatial degree of freedom directly appears in the number of output terminals 905 _m of the spatial cutoff filter circuit group 12.

Next, the structure of the multi-input canceller circuit 13 will be described with reference to FIG. In the multi-input canceller circuit 13, each output signal of the space cutoff filter circuit group 12 is supplied as a reference signal to the corresponding leak adaptive filter 7 _{0 to} 7 _MQ, and the output signal of the delay circuit 9 is supplied to the subtraction circuit 10. . Leak adaptive filter 7 ₀
.About.7 _MQ filters the corresponding output signal of the space cutoff filter circuit group 12 received as the reference signal, and transmits the output signal to the adder circuit 8. Adder circuit 8 calculates the sum total of the output signals received from leak adaptive filter groups 7 _{0 to} 7 _MQ , and transmits the result to subtraction circuit 10. Subtraction circuit 10
Subtracts the sum calculated by the adder circuit 8 from the signal supplied from the delay circuit 9 and outputs the result to the leak adaptive filter group 7
₀ to _{7-M- Q} with transmitting a common error signal to all, to the output terminal 11 as an output signal of the whole device. The leak adaptive filter group 7 _{0 to} 7 _MQ is a subtraction circuit 10
The tap coefficient is updated by the error signal received from.

The total number of taps of the leak adaptive filters 7 _{0 to} 7 _MQ is determined by the distance between the most distant microphones. This is for signals coming from any direction,
This is because the phase difference between the microphone outputs can be compensated by the leak adaptive filters 7 _{0 to} 7 _MQ .

The delay time of the delay circuit 9 is from the microphone groups l _{0 to} l _M-1 to the space pass filter circuit 2 and the delay circuit 9.
Through the microphone group l _{0 to} l _M-1 to the space cutoff filter circuit group 12, the leak adaptive filter groups 7 _{0 to} 7 _{M- Q} , and the addition circuit 8 to the subtraction circuit 10. It is set so that the delay of the route to reach is equal. The delay time of the space-pass filter circuit 2 is determined by the filter circuit groups 202 _{0 to} 20 20 constituting the space-pass filter circuit 2.
It is determined by the delay characteristic of 2 _M-1 and the delay time of the adder circuit 203. The delay time of the space cutoff filter circuit group 12 is
Filter circuit group 906 _{m, q} (m = 0, 1, ..., M−1, q = 0, 1, constituting the space cutoff filter circuit group 12
, MQ) and the delay time of the adder circuit 907. The delay time of the leak adaptive filter groups 7 _{0 to} 7 _MQ is determined by the total number of taps.

With this configuration, the leak adaptive filter group 7
_{0 to} 7 _MQ are the error signals, that is, the signals at the output terminal 11, from the spatial cutoff filter circuit group 12 which is the reference signal.
It operates so as to remove a component having a correlation with the output signal group of. The reason for using the leak adaptive filter instead of the normal adaptive filter will be described later.

The configuration of the leak adaptive filters 7 _{0 to} 7 _MQ will be described with reference to FIGS. 20 and 21. FIG.
Shows an example block diagram of an adaptive filter. Leak adaptive filter 7 ₀ to _7-MQ as tap coefficient updating circuit 326 in FIG. 14, an adaptive filter is used to leak with tap coefficient updating circuit shown in FIG. 21. In FIG. 20, an input terminal 321 has a space cutoff filter circuit group 12
Is supplied as a reference signal. Output terminal 32
The filter output signal is output to 5 and is transmitted to the adding circuit 203.

The configuration of the adaptive filter will be described with reference to FIG. The reference signal input to the input terminal 321 is a delay circuit 32 that delays by one sampling period.
2 ₀ ~322 _L-2 (L is an integer constant representing the tap total) is supplied to the tapped delay line consisting of. Delay circuit 322
_The input signal sample supplied to ₀ is transferred to the adjacent delay circuit every clock. Each delay circuit 322 _l (l
= 0, 1, ... L-2) is output by the multiplication circuit 323.
_It is supplied to _{l + 1} and is multiplied by the corresponding tap coefficient supplied from the tap coefficient update circuit 326. Multiplication circuit 323
_The signal input to the input terminal 321 is directly supplied to _0, and is multiplied by the tap coefficient supplied from the tap coefficient updating circuit 326. The output signals of the multiplication circuits 323 _{0 to} 323 _L-1 are supplied to the addition circuit 324. Adder circuit 324
Calculates the total sum of the output signals received from the multiplying circuits 323 _{0 to} 323 _L-1 , and outputs the result to the output terminal 325.

The tap coefficient updating circuit with leak used as the tap coefficient updating circuit 326 will be described with reference to FIG. FIG. 21 is a block diagram example of a tap coefficient update circuit with leak when the LMS algorithm is assumed as the tap coefficient update algorithm. Input terminal 327
Is supplied with an error signal from the subtraction circuit 10 of FIG. The signal supplied to the input terminal 321 of FIG. 20 is directly supplied to the input terminal 402 ₀ , and the input terminal 402 _l (l
= 1, 2, ..., L-1) is the delay circuit 322 of FIG.
_{The l-1} output signal is provided. Output terminal 401 _l (l =
The signals obtained at 0, 1, ..., L-1) are transmitted to the multiplication circuit 323 _l in FIG. 20 as tap coefficients.

The error signal supplied to the input terminal 327 is
It is transmitted to the multiplication circuit 407. The multiplication circuit 407 multiplies the error signal by the step size μ and transmits the result to all the multiplication circuits 406 _{0 to} 406 _L−1 having the same number as the total number of taps. The multiplication circuit 406 _l (l = 0, 1, ..., L−1) is
The signal supplied from the multiplication circuit 407 is multiplied by the delayed signal supplied via the terminal 402 _l , and the result is supplied to the addition circuit 405 _l . The adder circuit 405 _l is a multiplier circuit 4
The signal supplied from 09 _l and the signal supplied from the multiplication circuit 406 _l are added and transmitted to the delay circuit 403 _l .
The output signal of the delay circuit 403 _l is transmitted to the terminal 401 _l as a tap coefficient and is also supplied to the multiplication circuit 409 _l . The multiplication circuit 409 _l multiplies the output signal supplied from the delay circuit 403 _l by the leak control coefficient α,
It is transmitted to the adder circuit 405 _l .

In the tap coefficient updating circuit with leak, the leak control coefficient α multiplied by the multiplying circuits 409 _{0 to} 409 _L-1 is 1.
It is a slightly smaller positive constant, and is an adder circuit 405 _l , a delay circuit 403 _l , and a multiplier circuit 409 _l (l = 0, 1, ...,
A feature is that a leak integration circuit is configured by L-1). In FIG. 21, the portion surrounded by the alternate long and short dash line is the leak integration circuit. A normal adaptive filter corresponds to the case where α is 1, and is a normal integrating circuit instead of a leak integrating circuit.

The characteristics of the leak integration circuit will be described using equations. The multiplication circuit 405 _l (l =
The signal supplied from 0, 1, ..., L-1) to the adder circuit 405 _l is x ₁ (k), and the output signal of the delay circuit 403 _l is y.
₁ (k). Output signal y ₁ of delay circuit 403 _l at sample number (k + 1) (after one sampling cycle)
(K + 1) is expressed by equations (1) and (2).

Y ₁ (k + 1) = x ₁ (k) + αy ₁ (k) (1) = x ₁ (k) + y ₁ (k)-(1-α) y ₁ (k) (2) Normal integration In the circuit, since α = 1, the third side on the right side of equation (2)
The term becomes 0. On the other hand, in the leak integration circuit,
The third term on the right side of the equation (2) acts to reduce y ₁ (k + 1), that is, the output signal of the delay circuit 403 _l . By this action, the growth of the tap coefficient is suppressed in the leak adaptive filter as compared with the normal adaptive filter. Also, the strength to suppress the growth of the tap coefficient is
It is proportional to the value of (1-α) and the magnitude of the tap coefficient y ₁ (k).

If a normal adaptive filter is used instead of the leak adaptive filters 7 _{0 to} 7 _MQ , the target signal leaks to the output of the space cutoff filter circuit group 12 due to the influence of the microphone arrangement and the characteristic error. In this case, even if the leak amount is small, the tap coefficient grows endlessly and the target signal is removed. However, in the configuration of FIG. 16, the growth of the tap coefficients is suppressed because it uses leakage adaptive filter group 7 ₀ to _7-MQ, can leak amount to prevent removal of the target signal if slightly.

[0034]

In the conventional example described above, in order to reduce the deterioration of the target signal when there is an error between the arrival direction of the target signal and the specified direction, the space cutoff is performed even in such a case. A spatial cutoff filter circuit having a wide range in the cutoff direction is used so as to sufficiently cut off the target signal at each output of the filter circuit group 12. In this conventional example, in order to reduce the removal of the target signal even when the error between the target signal arrival direction and the specified direction is large, it is necessary to particularly widen the range in the cutoff direction. This means that the number of microphones Q required for each space cut-off filter circuit constituting the space cut-off filter circuit group 12 increases. Therefore, when the total number of microphones is fixed, the spatial freedom for removing ambient noise is reduced, which causes a problem that the ambient noise removing performance is deteriorated. Alternatively, in order to maintain the ambient noise removing performance, it is necessary to increase the total number of microphones, which causes a problem of increasing the hardware scale.

It is an object of the present invention to provide an adaptive array device which realizes signal reception with a small number of sensors so that the target signal deterioration is small even when the error between the target signal arrival direction and the specified direction is large.

[0036]

According to the present invention, a plurality of sensors arranged spatially at different positions and a group of output signals from the plurality of sensors are respectively filtered and added to obtain a signal in advance. A first space-pass filter circuit having directivity that allows a signal coming from a specified direction to pass through; a first delay circuit that delays by receiving an output signal of the first space-pass filter circuit; A plurality of space cut-off filter circuit groups, each of which includes a plurality of space cut-off filter circuits that cuts off only signals arriving from a predetermined direction by filtering and adding a plurality of sensor outputs from the sensor outputs. A first leak adaptation comprising a plurality of leak adaptive filters that receive and filter the output signals of the plurality of spatial cutoff filter circuits in a one-to-one relationship Filter group, an adder circuit that calculates the sum total of output signals of the first leak adaptive filter group, and a sum total calculated by the adder circuit from the output signal of the first delay circuit, and the subtraction result is the received signal. At the same time that the subtraction result is transmitted as a coefficient updating error signal common to all the first leak adaptive filter groups.
In the adaptive array device using the generalized sidelobe canceller, the spatial cutoff filter circuit receives output signal groups from the plurality of sensors, filters the output signal groups, and adds the output signal groups in advance, A second space-pass filter circuit having directivity for passing signals coming from a specified direction; a second delay circuit for receiving and delaying output signals of the plurality of sensors in a one-to-one relationship; A second leak adaptive filter that receives and filters the output signal of the second space-pass filter circuit, and the output signal of the second leak adaptive filter is subtracted from the output signal of the second delay circuit. A second subtraction circuit for transmitting as a coefficient update error signal to the leak adaptive filter which has supplied the signal to be subtracted, Second constituting a blocking filter circuit
The space-pass filter circuit of is shared by all the space cut-off filter circuits.

Further, in the present invention, instead of the first leak adaptive filter group, the norm constraint adaptive filter group which controls the tap coefficient so that the norm of the tap coefficient of each filter is equal to or less than the positive constant determined for each filter. It is characterized by including.

Further, according to the present invention, instead of the second leak adaptive filter group, a tap coefficient constraint adaptive filter group is provided which limits the variation range of the tap coefficient of the filter to the range defined for each tap. Characterize.

Further, the present invention is characterized in that the first space-pass filter circuit and the second space-pass filter circuit are shared.

The present invention is also characterized in that the first space-pass filter circuit is configured to output a sensor signal from only one of the plurality of sensors.

The present invention is also characterized in that, as the second space-pass filter circuit, a configuration for outputting only one sensor signal of the plurality of sensors is used.

[0042]

According to the present invention, in each of the space cutoff filter circuits constituting the space cutoff filter circuit group, the characteristic is such that the signals coming from the specified direction and the vicinity of the specified direction are passed and the signals coming from the non-specified direction are attenuated. The output signal of the second leak adaptive filter or the tap coefficient constrained adaptive filter, which uses the output signal of the second space-pass filter having the following as a reference signal, is subtracted from the signal obtained by delaying the signal of each sensor, and the subtraction result is The output of the space cutoff filter circuit is obtained, and the coefficient of the second leak adaptive filter or tap coefficient constraint adaptive filter is updated as an error signal. As a result, the components having correlation with the signals coming from the specified direction and the vicinity of the specified direction are sufficiently removed at the output of the spatial cutoff filter circuit, and the signals coming from the vicinity of the specified direction are prevented from being removed at the output of the multi-input canceller circuit. To be done. In the multi-input canceller circuit, the first leak adaptive filter or the norm constrained adaptive filter is used to suppress the tap coefficient from growing excessively, and the output from the vicinity of the specified direction slightly left in the output of the space cutoff filter circuit is suppressed. It is possible to prevent the signal from removing the signal from the vicinity of the specified direction. Further, in the space cutoff filter circuit group according to the present invention, the degree of freedom in space is less reduced.

[0043]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of the present invention. 16 is a block diagram of FIG. 1 and a conventional example.
Are the same except for the space cutoff filter circuit group 12, and therefore the detailed operation will be described below focusing on this difference.

The space cutoff filter circuit group 12 of the present invention is
Directivity that allows signals coming from a direction specified in advance to pass and attenuates signals coming from directions other than the specified direction by receiving and filtering output groups of the microphone groups l _{0 to} l _M-1 spatial-pass filter circuit 3, a plurality of delay for receiving and delaying an output signal of each of the microphone group l ₀ ~l _M-1 circuit 4 ₀ ~4 _M-1 having
And a delay circuit group including a space-pass filter circuit 3
And a leak adaptive filter group composed of a plurality of leak adaptive filters 5 _{0 to} 5 _{M-1 using} the output signal of
A signal for subtracting the output signal of the corresponding leak adaptive filter of the leak adaptive filter groups 5 _{0 to} 5 _M-1 from each output signal of the plurality of delay circuits 4 ₀ to 4 _M-1 and subtracting the subtraction result. Is transmitted to the leak adaptive filter as an error signal for coefficient update and is also transmitted to the multi-input canceller circuit 13 as an output signal of the spatial cutoff filter circuit group 12 from a plurality of subtraction circuits 6 _{0 to} 6 _M-1. And a subtraction circuit group.

First, the entire space cutoff filter circuit group 12 will be described with reference to FIG.

The space-pass filter circuit 3 receives the output groups of the microphone groups l _{0 to} l _M-1 and performs filtering and addition to allow signals coming from a direction specified in advance to pass therethrough and to output signals from directions other than the specified direction. incoming and the output signal obtained by attenuating the signal and supplies it to leak adaptive filter group 5 ₀ ~5 _M-1 as the reference signal. Leak adaptive filter 5
_m (m = 0, 1, ..., M−1) filters the output from the space-pass filter circuit 3 received as the reference signal, and transmits the output signal to the subtraction circuit 6 _m . Delay circuit 4 _m
(M = 0,1, ..., M -1) delays the signal supplied from the microphone l _m, and transmits to the subtraction circuit 6 _m. The subtraction circuit 6 _m (m = 0, 1, ..., M−1) is a delay circuit 4
from the supplied signals from _m, subtracts the output signal of the leak adaptive filter 5 _m, the results as well as transmitted to leak the adaptive filter 5 _m as an error signal, the multi-input canceller circuit 13 as an output of the spatial cutoff filter circuit group 12 Supply. Leak adaptive filter 5 _m (m = 0, 1, ..., M−
In 1), the coefficient is updated by the error signal received from the subtraction circuit 6 _m .

In this configuration, the leak adaptive filter 5
_m (m = 0, 1, ..., M−1) operates so as to remove a component having a correlation with the reference signal, that is, the output signal of the space-pass filter circuit 3, from the error signal, that is, the output signal of the subtraction circuit 6 _m. To do. As a result, in the output of the subtraction circuit 6 _m , that is, the output of the space cutoff filter circuit group 12, the signals coming from the specified direction and the vicinity of the specified direction are removed, and the signals coming from the non-specified direction remain. Therefore, the space pass filter circuit 3, the delay circuit 4 _m , the leak adaptive filter 5 _m , and the subtraction circuit 6 _m can be regarded as a kind of space cutoff filter circuit.

Leak adaptive filter 5 _m (m = 0, 1,
, M-1) and the delay time of the delay circuit 4 _m are set so that signals coming from near the specified direction can be removed. That is, the total number N of taps of the leak adaptive filter 5 _m (m = 0, 1, ..., M−1) is
The number is sufficiently large so that signals coming from the vicinity of the specified direction can be removed. Further, the delay circuit 4 _m (m = 0, 1, ..., M−
1) the delay time of the spatial-pass filter circuit 3 from the microphone group l _m, a delay of the route leading to leakage adaptive filter group 5 via the _m subtraction circuit 6 _m, the subtracting circuit via a delay circuit 4 _m from the microphone group l _m Set so that the delay of the route up to 6 _m is equal.

The reason for using the leak adaptive filter instead of the normal adaptive filter will be described later.

With this space cutoff filter circuit group 12,
Even when the arrival direction of the target signal does not match the specified direction and is near the specified direction, the target signal is removed from the output of the spatial cutoff filter circuit group 12. As a result, even if there is an error between the arrival direction of the target signal and the specified direction, the removal of the target signal at the output terminal 11 of FIG. 1 is prevented.

Next, each component of the space cutoff filter circuit group 12 will be described.

The structure of the space-pass filter circuit 3 is similar to that of the conventional space-pass filter circuit 2,
It can be represented as in FIG. However, the characteristics as the spatial filter need not be the same in the space pass filter circuit 3 and the space pass filter circuit 2. The input relationship of the space pass filter circuit 3 is the same as that of the space pass filter circuit 2. The output relationships are different, and the output obtained at the output terminal 204 of the space pass filter circuit 3 is supplied to the leak adaptive filters 5 _{0 to} 5 _M-1 in FIG. ₁ as a reference signal. The filter circuit 202 _{m of} FIG. 17 (m = 1, 2, ...,
Since the configuration of M-1) is similar to that of the spatial filter circuit 2, the description thereof is omitted.

The space-pass filter circuit 3 is designed so as to pass the target signal coming from the specified direction and the vicinity of the specified direction and attenuate the signal coming from the non-specified direction.
For example, in the description of the conventional example, the space-pass filter circuit 2
You can use the design method introduced in.

The leak adaptive filters 5 _{0 to} 5 _{M-1 have} the same configuration as the leak adaptive filters 7 _{0 to} 7 _MQ described in the conventional example.
In FIG. 20 showing the configuration example of, the total number of taps L is given by replacing it with N. Since the operation is the same, detailed description will be omitted.

Leak adaptive filter 5 _m (m = 0, 1,
, M-1), the input terminal 321 of FIG.
The signal from the space-pass filter 3 in FIG. 1 is supplied as a reference signal. The reference signal is filtered, and the output signal is transmitted to the subtraction circuit 6 _m in FIG. 1 via the output terminal 325. The output signal of the subtraction circuit 6 _m is supplied to the input terminal 327.

The characteristics of the output signals of the subtraction circuits 6 _{0 to} 6 _M-1 derived from the characteristics of the leak adaptive filter will be described in detail below. In the leak adaptive filter, the strength of suppressing the growth of the tap coefficient is proportional to the magnitude of the tap coefficient, as described with reference to FIG. 21 for the leak adaptive filters 7 _{0 to} 7 _MQ of the conventional example. Therefore, when the value of the optimum tap coefficient (the tap coefficient that minimizes the error input) is large, it becomes difficult for the tap coefficient to approach the optimum tap coefficient. As a result, the tap coefficient after convergence has a large error with respect to the optimum tap coefficient. This means that the leak adaptive filter 5 _m (m = 0, 1, ..., M−1) has a correlation from the error signal, that is, the output signal of the subtraction circuit 6 _m , to the reference signal, that is, the output signal of the space-pass filter circuit 3. This means that when a certain component is removed, the amount of removal varies depending on the magnitude of the tap coefficient. That is, components that require a large tap coefficient for removal are not removed much, but components that can be removed sufficiently with a small tap coefficient are largely removed. By the way, the amount of the signal arriving from the vicinity of the specified direction contained in the output of the space-pass filter circuit 3 is substantially equal to the amount contained in each microphone output l _{0 to} l _M−1 .
The maximum value of the tap coefficient required for the removal is as small as about 1. Therefore, in the outputs of the subtraction circuits 6 _{0 to} 6 _M-1 , the signals coming from the vicinity of the specified direction are largely removed. On the other hand, the amount of signals included in the output of the space-pass filter circuit 3 for the signals arriving from the non-specified direction is equal to each microphone output l _0.
Since it is smaller than the amount contained in ˜l _M-1 , the maximum value of the tap coefficient required for its removal is much larger than 1. Therefore, in the outputs of the subtraction circuits 6 _{0 to} 6 _M-1 , the removal amount of the signal coming from the non-specified direction is small. For the above reasons,
In the outputs of the subtraction circuits 6 _{0 to} 6 _M-1 , only a few signals arriving from the specified direction and the vicinity of the specified direction remain, but relatively many signals arriving from the non-specified direction remain.

[0058] If it is assumed that uses the normal adaptive filter in place of the leak adaptive filter 5 ₀ ~5 _M-1, since the tap coefficients of the adaptive filter can endlessly growth, at the output of the subtraction circuit 6 _m, defined direction Not only the signal coming from the vicinity but also the signal coming from the non-specified direction will be removed. This makes it impossible for the multi-input canceller circuit 13 to remove a signal coming from an unspecified direction, that is, ambient noise. The present invention avoids this problem by using a leak adaptive filter rather than a normal adaptive filter.

As is clear from the above description, the ambient noise coming from the non-specified direction is largely removed at the output terminal 11 of the multi-input canceller circuit 13. On the other hand, with respect to the target signal, even if there is an error between the arrival direction and the specified direction, the reference signal contains only a very small amount, so that it is hardly removed at the output terminal 11 of the multi-input canceller circuit 13. .

The differences between the embodiment and the conventional example will be summarized based on the above description. In the conventional example of “Document 2” shown in FIG. 16,
In order to prevent the phenomenon that the target signal is removed at the output terminal 11 when there is an error between the specified direction and the arrival direction of the target signal, the space cutoff filter circuit group 12 requires a plurality of microphone outputs. It consisted of a fixed spatial filter. Therefore, the spatial degree of freedom in the output group of the spatial cutoff filter circuit group 12 is smaller than that in "Document 1". This means a reduction in spatial freedom to remove ambient noise. As a result, when the total number of microphones is fixed, the performance of removing ambient noise deteriorates. Alternatively, it is necessary to increase the total number of microphones in order to obtain constant ambient noise elimination performance.

On the other hand, in the present invention shown in FIG.
The space cutoff filter group 12 includes the leak adaptive filters ₅₀ to.
It is composed of a plurality of variable spatial filters using 5 _M-1 , and the spatial degree of spatial freedom is not reduced as compared with the method of "Reference 2". Therefore, it becomes possible to prevent the removal of the target signal at the output terminal 11 with a smaller number of microphones than in the conventional example shown in "Document 2".

FIG. 2 is a block diagram showing a second embodiment of the present invention. The difference between the first embodiment and the second embodiment shown in FIG. 1 is that in the multi-input canceller circuit 13, the leak adaptive filters 7 _{0 to} 7 _M-1 in FIG. ₁ are the norm constraint adaptive filter 14 _{0 in} FIG. ~ 14 _{M- 1} .

Norm constrained adaptive filters 14 _{0 to} 14 _M-1
Configuration of, in FIG. 20 illustrating a configuration example of a leak adaptive filter 7 ₀ to _7-MQ described in the prior art is given by the tap coefficient updating circuit 326 and norm constraint tap update circuit.

[0064] norm constraint tap updating circuit, p-th power norm L _P (p is an integer of 1 or more constant) of the tap coefficient is a positive constant Θ
The tap coefficient is controlled so as not to exceed. However, the tap coefficients to the filter is L, p-th power norm L _P is calculated as follows.

[0065]

[Equation 1]

Where w _l is the l-th tap coefficient,
| W _l | represents the absolute value of w _l . Also

[0067]

[Outside 1]

Represents the pth root. The fact that L _P is limited to Θ or less means that the respective tap coefficients cannot grow to a certain degree or more, and when the target signal leaks to the output group of the spatial cutoff filter circuit as in the first embodiment. However, the effect of suppressing removal of the target signal in the multi-input canceller output 11 is produced.

The norm constraint tap update circuit will be described with reference to FIG. FIG. 3 is a block diagram of a norm-constrained tap update circuit when the LMS algorithm is assumed as the tap coefficient update algorithm. Input terminal 3
An error signal is supplied to 27 from the subtraction circuit 10 of FIG. The signal supplied to the input terminal 321 of FIG. 20 is directly supplied to the input terminal 402 ₀ , and the input terminal 402 _l (l
= 1, 2, ..., L-1) is the delay circuit 322 of FIG.
_{The l-1} output signal is provided. Output terminal 401 _l (l =
The signals obtained at 0, 1, ..., L-1) are transmitted to the multiplication circuit 323 _l in FIG. 20 as tap coefficients.

The error signal supplied to the input terminal 327 is
It is transmitted to the multiplication circuit 407. The multiplication circuit 407 multiplies the error signal by the step size μ and transmits the result to all the multiplication circuits 406 _{0 to} 406 _L−1 having the same number as the total number of taps. The multiplication circuit 406 _l (l = 0, 1, ..., L−1) is
The signal supplied from the multiplication circuit 407 is multiplied by the delayed signal supplied via the terminal 402 _l , and the result is supplied to the addition circuit 405 _l . The adder circuit 405 _l is a multiplier circuit 4
The signal supplied from 24 _l and the signal supplied from the multiplication circuit 406 _l are added, and the result is transmitted to the delay circuit 403 _l and also to the constraint control coefficient generation circuit 428. Delay circuit 403 _l delays the signal received from adder circuit 405 _l by one sampling period, and transmits the output signal to multiplication circuit 424 _l . The multiplication circuit 424 _l receives the signal received from the delay circuit 403 _l and the constraint control coefficient generation circuit 42
The constraint control coefficient β supplied from 8 is multiplied, and the result is supplied to the adder circuit 405 _l and also transmitted to the terminal 401 _l as a tap coefficient. Constraint control coefficient generation circuit 428
Calculates the restraining control coefficient β from a signal group received from the adder circuit 405 ₀ ~405 _L-1, and supplies the 424 ₀ ~424 _L-1.

FIG. 4 shows a configuration example of the constraint control coefficient generation circuit 428. The p-th power norm calculation circuit 510 receives the output signal of the adder circuit group 405 _l (l = 0, 1, ..., L-1) of FIG. 3 via the input terminal 501 _l and calculates the p-th power norm L _p . To do. The calculation result of the p-th power norm L _p is the division circuit 506.
Is transmitted to The division circuit 506 calculates the limit value Θ of the p-th power norm given to the terminal 507 by the p-th power norm calculation circuit 51.
The division is performed by the p-th power norm L _p received from 0, and the result is transmitted to the minimum value selection circuit 508. Minimum value selection circuit 508
Is the division result received from the division circuit 506 and the terminal 509.
Is compared to the constant l, and the smaller one is output as the constraint control coefficient β to the output terminal 502. Output terminal 5
02 is transmitted to the multiplication circuit groups 424 _{0 to} 424 _L−1 and is multiplied by the signal received from the delay circuit 403 _l . As described above, the constraint control coefficient generation circuit 408 reduces all tap coefficients so that L _p becomes equal to or less than Θ when the p-th power norm L _p exceeds a certain positive constant Θ.

In the L _p norm calculation circuit 510 of FIG. 4, p
The multiplication circuit 503 _l (l = 0, 1, ..., L−1) receives the output signal of the addition circuit 405 _{l in} FIG. 3 via the input terminal 501 _l , calculates the p-th power value thereof, and adds the results. It is transmitted to the circuit 504. The adder circuit 504 is a p-th power circuit group 503 _0-5.
The sum of the p-th power values received from 03 _L-1 is calculated, and the result is transmitted to the p-th root circuit 505. The p-root circuit 505 calculates the p-root of the value received from the adder circuit 504. The output signal of the p-th root circuit 505 is the p-th power norm L _p and is transmitted to the division circuit 506.

FIG. 5 is a block diagram showing a third embodiment of the present invention. The difference between the first embodiment and the third embodiment shown in FIG. 1 lies in that in the space cutoff filter circuit group 12, FIG.
5 is that the leak adaptive filters 5 _{0 to} 5 _M-1 of FIG. 5 are replaced with the tap coefficient constraint adaptive filters 15 ₀ to 15 _M-1 in FIG.

Tap coefficient constraint adaptive filters 15 ₀ to 15
_In the configuration of _M-1, the total number of taps L is set to N, and the tap coefficient updating circuit 326 is set to the tap coefficient constraining tap updating circuit in FIG. 20 _showing the configuration example of the leak adaptive filters 7 _{0 to} 7 _MQ described in the conventional example. Given by substituting.

Tap coefficient constraint filter 15 ₀ to 15 _M-1
Then, the coefficient maximum absolute values Φ _{0 to} Φ _N−1 are set for each tap as the upper limit of the absolute value of the tap coefficient. The coefficient maximum absolute values Φ _{0 to} Φ _N-1 for each tap are the subtraction circuits 6 _{0 to} 6 of FIG.
It is determined in consideration of the removal of the signal in _M-1 . That is, when the signal exists only in a certain direction, the subtraction circuit 6
_The delay circuit 4 supplied to _m (m = 0, 1, ..., M-1)
Consider whether the output signal of _{m and} the output signal of the tap coefficient constraint adaptive filter 15 _m can be equal to each other.

Since the signal arriving from the vicinity of the specified direction should be removed by the subtraction circuit 6 _m , the maximum absolute coefficient value of the tap which plays a main role in the removal is the output signal of the tap coefficient constraint adaptive filter 15 _m. The value is set to be sufficiently large so as to be equal to the output signal of the delay circuit 4 _m . On the other hand, since the signal arriving from the non-specified direction should not be removed, the maximum coefficient absolute value of the tap coefficient, which plays a main role in the removal, is the output signal of the tap coefficient constraint adaptive filter 15 _{m from} the delay circuit 4 _m . Use a small value that does not equal the output signal. For example, the tap coefficient constraint adaptive filter 15
Maximum coefficient absolute value Φ _{0 of m} (m = 0, 1, ..., M−1)
Φ _N-1 is the output signal amplitude of the delay circuit 4 _m with respect to the output signal amplitude of the space-pass filter circuit 3 (= the input signal amplitude of the tap coefficient constraint adaptive filter 15 _m ) when the signal exists only in a certain direction. It can be determined by using the ratio (direction output ratio).

For the nth tap of the tap coefficient constraint adaptive filter 15 _m , if the tap is the center tap (center tap) for removing the signal coming from the specified direction, the direction about the specified direction A value slightly larger than the output ratio is set as the coefficient maximum absolute value Φ _n . If the tap is a main tap (near the center tap) for removing the signal coming from the vicinity of the specified direction, a value slightly larger than the maximum value of the direction output ratio in the vicinity of the specified direction is set as the coefficient maximum absolute value Φ. _{Let n} . For all other taps, the coefficient maximum absolute value Φ increases with increasing distance from the center tap.
_{Make n} smaller.

The maximum absolute coefficient value Φ ₀ set in this way
When the absolute value of the n-th tap coefficient w _n exceeds the maximum coefficient absolute value Φ _n according to Φ _N-1 , the tap coefficient is reduced so that the absolute value of the tap coefficient becomes Φ _n or less.

As described above, by limiting the tap coefficients respectively, in the outputs of the subtraction circuits 6 _{0 to} 6 _M-1 , the signals arriving from the vicinity of the specified direction are largely removed, and the signals from the non-specified direction are removed. The incoming signal is not well removed. In the multi-input canceller circuit 13 supplied with this output as the reference signal, the target signal in the multi-input canceller output 11 can be removed even when there is an error between the target signal arrival direction and the specified direction, as in the first embodiment. The effect of being suppressed occurs.

The tap coefficient constrained tap update circuit will be described in detail with reference to FIG. FIG. 6 is a block diagram example of a tap coefficient constrained tap coefficient update circuit when the LMS algorithm is assumed as the tap coefficient update algorithm. The subtraction circuit 1 of FIG.
An error signal is supplied from 0. The input terminal 402 ₀ has
The signal supplied to the input terminal 321 of FIG. 20 is directly supplied, and the output signal of the delay circuit 322 _n-1 of FIG. 20 is supplied to the input terminal 402 _n (n = 1, 2, ..., N−1). It
The signal obtained at the output terminal 401 _n (n = 0, 1, ..., N−1) is used as a tap coefficient in the multiplication circuit 323 _{n in} FIG.
Is transmitted to

The error signal supplied to the input terminal 327 is
It is transmitted to the multiplication circuit 407. The multiplication circuit 407 multiplies the error signal by the step size μ and transmits the result to all the multiplication circuits 406 _{0 to} 406 _N−1 having the same number as the total number of taps. The multiplication circuit 406 _n (n = 0, 1, ..., N−1) is
The signal supplied from the multiplication circuit 407 is multiplied by the delay signal supplied via the terminal 402 _n , and the result is supplied to the addition circuit 405 _n . Adding circuit 405 _n transmits a signal supplied from the limiter circuit 434 _n, adds the signal supplied from the multiplier circuit 406 _n to the delay circuit 403 _n. The delay circuit 403 _n delays the signal supplied from the adder circuit 405 _n by one sampling period and transmits the output signal to the limiter circuit 434 _n . Limiter circuit 434 _n
Is the signal received from the delay circuit 403 _n, performs limiter below in accordance with the coefficient maximum absolute value [Phi _n supplied from the constraints generation circuit 438. The result of the limiter calculation is supplied to the addition circuit 405 _n and also transmitted to the terminal 401 _n as a tap coefficient.

Limiter circuit 434 _n (n = 0, 1, ...,
FIG. 7 shows an example of input / output characteristics of N-1). Delay circuit 403
_When the absolute value of the input signal received from _{n is} smaller than the maximum coefficient absolute value Φ _n , the same signal as the input signal is output. When the absolute value of the input signal is larger than the maximum coefficient absolute value Φ _{n, the} sign is the input signal. It outputs the same signal whose absolute value is Φ _n .

FIG. 8 is a block diagram showing a fourth embodiment of the present invention. The difference between the third embodiment and the fourth embodiment is that
In the multi-input canceller circuit 13, the leak adaptive filters 7 _{0 to} 7 _M-1 in FIG. 5 are replaced with the norm constraint adaptive filters 14 _{0 to} 14 _M-1 in FIG. This difference is the same as the difference between Example 1 and Example 2,
The effect is the same.

FIG. 9 is a block diagram showing the fifth embodiment of the present invention. The difference between the first embodiment and the fifth embodiment shown in FIG. 1 is that the output of the space pass filter circuit 2 is shared as the output of the space pass filter circuit 3 in the space cutoff filter circuit group 12 in FIG. That is. However, since the characteristics of the space-pass filter circuit 2 are shared with the space-pass filter circuit 3, not only the signal coming from the vicinity of the specified direction is passed, but also the signal coming from the non-specified direction is attenuated. In the fifth embodiment, the space-pass filter is shared, so that the space-pass filter circuit 3 can be omitted, and the calculation amount and the hardware scale can be reduced as compared with the first embodiment. Example 2 is based on the same principle.
With respect to 4 to 4, a configuration in which the space pass filter circuit 3 and the space pass filter circuit 2 are shared can be considered.

FIG. 10 is a block diagram showing a sixth embodiment of the present invention. The difference between the third embodiment and the sixth embodiment shown in FIG. 5 is that the space-pass filter circuit 2 in the third embodiment is replaced with a configuration for directly outputting the output signal of one microphone. It is that you are. In the third embodiment, the output signal of the space-pass filter circuit 2 is supplied as the input signal of the delay circuit 9, whereas in the sixth embodiment, the microphone l _{C is} input as the input signal of the delay circuit 9. The signal output from is directly supplied.
In the sixth embodiment, the space-pass filter circuit 2 in the third embodiment is not necessary, so that the amount of calculation and the hardware scale can be reduced.

As the microphone l _C for supplying the input signal of the delay circuit 9, any _one of the microphones l _{0 to} l _M-1 can be used. For example, when M is an odd number, the microphone l _{((M-1) / 2)} can be used as the microphone l _C.

FIG. 11 is a block diagram showing the seventh embodiment of the present invention. The difference between the fourth embodiment and the seventh embodiment shown in FIG. 8 is that the space-pass filter circuit 2 in the fourth embodiment is replaced with a configuration for directly outputting the output signal of one microphone. It is that you are. In the fourth embodiment, the output signal of the space-pass filter circuit 2 is supplied as the input signal of the delay circuit 9, whereas in the seventh embodiment, the microphone l _{C is} input as the input signal of the delay circuit 9. The signal output from is directly supplied.
This difference is the same as the difference between Example 3 and Example 6, and the effect is also the same.

On the basis of the same principle, a configuration in which the space-pass filter circuit 2 is replaced with a configuration in which the output signal of one microphone is directly output can be considered in the first or second embodiment.

FIG. 12 is a block diagram showing the eighth embodiment of the present invention. The difference between the third embodiment and the eighth embodiment shown in FIG. 5 is that the space-pass filter circuit 3 in the third embodiment is replaced with a configuration for directly outputting the output signal of one microphone. It is that you are. In the third embodiment, the tap coefficients constraint adaptive filter group l5 ₀ ~l5
_The space-pass filter circuit 3 is used as an input signal common to _M-1.
In the eighth embodiment, the signal output from the microphone l _C is directly supplied. In the eighth embodiment, the space-pass filter circuit 3 in the third embodiment is unnecessary, so that
This has the effect of reducing the amount of calculation and the scale of hardware.

The tap coefficient constraint adaptive filter group l5
₀ As ~l5 _M-1 Microphone l _C supplies a common input signal, it is possible to use any of the microphone from the microphone l ₀ ~l _M-1. For example, when M is an odd number, the microphone l _C is the microphone l
_{((M-1) / 2)} can be used.

FIG. 13 is a block diagram showing the ninth embodiment of the present invention. The difference between the fourth embodiment and the ninth embodiment shown in FIG. 8 is that the space-pass filter circuit 3 in the fourth embodiment is replaced with a configuration for directly outputting the output signal of one microphone. It is that you are. In the fourth embodiment, the tap coefficients constraint adaptive filter group l5 ₀ ~l5
_The space-pass filter circuit 3 is used as an input signal common to _M-1.
In contrast to this, in the ninth embodiment, the signal output from the microphone l _C is directly supplied. This difference is the same as the difference between Example 3 and Example 8, and the effect is also the same.

Based on the same principle, it is possible to consider a configuration in which the space-pass filter circuit 3 is replaced with a configuration in which the output signal of one microphone is directly output in the first or second embodiment.

FIG. 14 is a block diagram showing the tenth embodiment of the present invention. Sixth embodiment and tenth embodiment shown in FIG.
The difference from the embodiment is that the space-pass filter circuit 3 in the sixth embodiment is replaced with a configuration for directly outputting the output signal of one microphone. In the sixth embodiment, the tap coefficients constraint adaptive filter group l5 ₀
The output signal of the space-pass filter circuit 3 is supplied as a common input signal to ~ 15M _-1 while the signal output from the microphone l _C is directly supplied in the tenth embodiment. There is. In the tenth embodiment,
Since the space-pass filter circuit 3 in the sixth embodiment is not necessary, there is an effect that the calculation amount and the hardware scale are further reduced.

The tap coefficient constraint adaptive filter group l5
₀ as a microphone for supplying a common input signal to ~l5 _M-1, it can be any microphone from the microphone l ₀ ~l _M-1. In Figure 14, a microphone for supplying a common input signal to the tap coefficients constraint adaptive filter group l5 ₀ ~l5 _M-1, the microphone provides an input signal of the delay circuit 9, by using a common microphone l _C However, different microphones can be used.

FIG. 15 is a block diagram showing the twelfth embodiment of the present invention. Seventh embodiment and eleventh embodiment shown in FIG.
The difference from the embodiment is that the space-pass filter circuit 3 in the seventh embodiment is replaced with a configuration for directly outputting the output signal of one microphone. This difference is the same as the difference between the sixth embodiment and the tenth embodiment, and the effect is also the same.

According to the same principle, in the first or second embodiment, the space-pass filter circuit 2 is replaced with a configuration which directly outputs the output signal of one microphone, and the space-pass filter circuit 3 is replaced by one microphone. A configuration in which the output signal is directly output can be considered.

The embodiments of the present invention have been described above in the case where the microphones are arranged at equal intervals in a straight line and the specified direction is orthogonal to the microphone arranged line. However, even when the microphones are arranged or the specified directions are different, The characteristics of the filter circuits 202 _{0 to} 202 _M-1 in the space-pass filter circuit 2 are designed so as to pass signals coming from the specified direction, and the space cut-off filter circuit group 12 is designed so as to cut off signals coming from the specified direction. By designing, the same configuration can be applied in principle.

Further, even when the microphone characteristics and the amplifier circuits of the microphones are different from each other by changing the characteristics of the space pass filter circuit 2 and the space cutoff filter circuit group 12, the same configuration can be used in principle. .

Even when the position of the signal source is not sufficiently far from the microphone array and the wavefront of the signal is not a plane wave, the microphone arrangement is changed or the space pass filter circuit 2 or the space cutoff filter circuit group 12 is appropriately changed. By design, the same configuration can be used in principle.

Further, in the above description, only the FIR type filter is used as the filter circuit such as the adaptive filter, but other filter circuits such as the IIR type and the lattice type can be realized.

Although the microphone array has been described above as an example, the present invention can be applied to an antenna array or the like based on the same principle. Further, with respect to the tap coefficient updating algorithm, various algorithms other than the LMS algorithm used as an example can be applied.

[0102]

As described above, according to the present invention,
Even when there is an error between the arrival direction of the target signal and the specified direction, signal reception with less deterioration of the target signal can be performed with a small number of sensors.

[Brief description of drawings]

FIG. 1 is a first embodiment of the present invention.

FIG. 2 is a second embodiment of the present invention.

3 is a block diagram example of a norm constraint tap coefficient updating circuit in the norm constraint adaptive filter of FIG.

4 is an example of a block diagram of a constraint control coefficient generation circuit in FIG.

FIG. 5 is a third embodiment of the present invention.

6 is a block diagram of a tap coefficient constraining tap coefficient updating circuit in the tap coefficient constraining adaptive filter of FIG. 5;

FIG. 7 is a graph showing an input / output characteristic example of the limiter circuit in FIG.

FIG. 8 is a fourth embodiment of the present invention.

FIG. 9 is a fifth embodiment of the present invention.

FIG. 10 is a sixth embodiment of the present invention.

FIG. 11 is a seventh embodiment of the present invention.

FIG. 12 is an eighth embodiment of the present invention.

FIG. 13 is a ninth embodiment of the present invention.

FIG. 14 is a tenth embodiment of the present invention.

FIG. 15 is an eleventh embodiment of the present invention.

FIG. 16 is a block diagram showing a configuration of a conventional example.

FIG. 17 is a block diagram example of a space-pass filter circuit in FIG.

18 is an example block diagram of a filter circuit in FIG.

19 is a block diagram showing a configuration example of a spatial cutoff filter in FIG.

20 is an example block diagram of a leak adaptive filter in FIG.

FIG. 21 is a block diagram example of a tap coefficient updating circuit with leak.

[EXPLANATION OF SYMBOLS] _{1 0} ~1 _M-1 microphone 2,3 spatial pass filter circuit 4 ₀ ~4 _M-1, 9 delay circuits _{5 0 ~5 M-1, 7} 0 ~7 M-1 leak adaptive filter 6 _{0 to} 6 _M-1 , 10 subtraction circuit 8 addition circuit 11 output terminal 12 spatial cutoff filter circuit group 13 multi-input canceller circuit 14 _{0 to} 14 _M-1 norm constraint adaptive filter 15 ₀ to 15 _M-1 tap coefficient constraint adaptive filter 201 _{0 to} 201 _M-1 microphone signal input terminal 202 _{0 to} 202 _M-1 filter circuit 203 addition circuit 204 spatial filter output terminal 301, 321 input terminal 302 _{0 to} 302 _G-2 , 322 _{0 to} 322 _L-2 delay circuit _{303 0 ~303 G-1, 323} 0 ~323 L-1 multiplier circuits 304, 324 adder circuits 305, 325 output terminals 306 tap coefficient memory circuit 326 tap coefficient updating circuit 327 error signal Input terminal 401 ₀ ~401 _L-1 delay signal input terminals 402 ₀ ~402 _L-1 tap coefficient output terminal 403 ₀ ~403 _L-1 delay circuits 405 ₀ ~405 _L-1 addition circuit 406 ₀ ~406 _L-1, 409 _{0 to} 409 _L-1 multiplication circuit 407, 424 _{0 to} 424 _L-1 multiplication circuit 434 _{0 to} 434 _N-1 limiter circuit 428 Constraint control coefficient generation circuit 438 Constraint value generation circuit 501 _{0 to} 501 _{L-1 From} addition circuit Signal input terminal 502 constraint control coefficient output terminal 503 _{0 to} 503 _L-1 p power circuit 504 addition circuit 505 p power root circuit 506 division circuit 507 norm constraint value input terminal 508 minimum value selection circuit 509 constant input terminal 510 p power norm calculation circuit 901 ₀ ~901 _M-1 microphone signal input terminal 905 ₀ ~905 _MQ spatial cutoff filter circuit output terminal _{906 m, Q (m = 0,1} , ..., M- , Q = 0,1,
..., Q-1) Filter circuit 907 _{0 to} 907 _MQ addition circuit

Claims (6)

[Claims] 1. A plurality of sensors arranged at spatially different positions and a group of output signals from the plurality of sensors which are respectively filtered and added to obtain a signal arriving from a predetermined direction. A first space-pass filter circuit having directivity to allow passage, a first delay circuit that receives and delays an output signal of the first space-pass filter circuit, and a plurality of new outputs from the outputs of the plurality of sensors And a plurality of space cutoff filter circuits configured to cut off only signals arriving from a direction defined in advance by selecting sensor outputs of the plurality of space cutoff filter circuits, A first leak adaptive filter group consisting of a plurality of leak adaptive filters for receiving the output signals of 1 to 1 and filtering them, An adder circuit for calculating the sum of output signals of the leak adaptive filter group, and a sum calculated by the adder circuit is subtracted from the output signal of the first delay circuit, and a subtraction result is output as a reception signal, and at the same time, the subtraction result A first subtraction circuit for transmitting as a coefficient update error signal common to all the first leak adaptive filter groups, and an adaptive array device by a generalized sidelobe canceller, wherein the spatial cutoff filter circuit is A second space-pass filter circuit having directivity that allows signals coming from a predetermined direction to pass by receiving output signals from the plurality of sensors and filtering and adding the output signal groups; Second delay circuit that receives the output signal of the sensor of 1: 1 and delays it, and the output signal of the second space-pass filter circuit A second leak adaptive filter for receiving and filtering, and a leak adaptive filter for subtracting the output signal of the second leak adaptive filter from the output signal of the second delay circuit and supplying a signal for subtracting the subtraction result. A second subtraction circuit for transmitting as an error signal for coefficient updating to the second space-pass filter circuit, which constitutes the space cut-off filter circuit, is shared by all of the space cut-off filter circuits, An adaptive array device characterized by the above. 2. A norm constrained adaptive filter group that controls the tap coefficient so that the norm of the tap coefficient of each filter is equal to or less than a positive constant defined for each filter instead of the first leak adaptive filter group. The adaptive array device according to claim 1, wherein: 3. A tap coefficient constrained adaptive filter group in which the variation range of the tap coefficient of the filter is limited to a range defined for each tap instead of the second leak adaptive filter group. The adaptive array device according to claim 1 or 2. 4. The adaptive array device according to claim 1, 2 or 3, wherein the first space-pass filter circuit and the second space-pass filter circuit are shared. 5. The first space-pass filter circuit,
4. The adaptive array device according to claim 1, wherein a configuration that outputs only one sensor signal among the plurality of sensors is used. 6. The second space-pass filter circuit,
The adaptive array device according to claim 1, 2, 3, or 5, wherein a configuration that outputs only one sensor signal among the plurality of sensors is used.