close

Вход

Забыли?

вход по аккаунту

?

JPH07281672

код для вставкиСкачать
Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JPH07281672
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
noise suppressor using active noise control.
[0002]
2. Description of the Related Art In recent years, an active noise control method has been
proposed in which environmental noise is output from a speaker using digital signal processing
technology and control noise is silenced at a listening position.
[0003]
Hereinafter, a conventional muffling apparatus will be described with reference to the drawings.
(FIG. 27) shows a block diagram of a conventional noise reduction device, and shows a case
where one noise source is controlled at two points. In FIG. 27, microphones 1a to 1c are first and
second noise detectors, 2a and 2b adaptive filters, 3a and 3b transfer function correction circuits,
and FIR filters 4a and 4b. Is an LMS operator which is a coefficient operator, 6 is an adder, and 8
is a speaker.
[0004]
03-05-2019
1
The operation of the noise reduction device configured as described above will be described
below. First, noise from a certain noise source is detected by the microphone 1a. The detection
signal is input to the adaptive filter 2a and the FIR filter 3a. Then, the noise signal subjected to
signal processing by the adaptive filter 2 a is output to the adder 6. Similarly, the noise from
another noise source is detected by the microphone 1b, the detection signal is input to the
adaptive filter 2b and the FIR filter 3b, and the noise signal processed by the adaptive filter 2b is
output to the adder 6 Be done. The adder 6 adds the outputs of the adaptive filters 2a and 2b and
reproduces them from the speaker 8. Then, in the microphone 1c, the interference sound
between this reproduced sound and the noise from the two noise sources is detected. The LMS
arithmetic units 4a and 4b respectively perform LMS operation (least squares method) so that
the detection signal of the microphone 1c is minimized by the detected sound and the outputs of
the FIR filters 3a and 3b. Update the coefficients. By this, the noise level is attenuated at the point
where the microphone 1 c is disposed. Here, in the FIR filters 3a and 3b, a transfer function C11
from the speaker 8 to the microphone 1c is approximated as a coefficient in advance. This
method is called the Filtered-x algorithm (e.g., B. Widrow and S. Stearns, "Adaptive Signal
Processing" (Prentice-Hall, Englewood Cliffs, NJ, 1985)). Using this, the coefficient update of the
adaptive filter 2 can be expressed by the following equation.
[0005]
In the example (FIG. 27), since there are a plurality of noise sources but one control point,
basically a system of one noise source, one control point and one control speaker is used. It only
increased by the number of noise sources. Therefore, next, the case where the control speaker
and the control point are also plural will be described.
[0006]
Hereinafter, another conventional silencer will be described with reference to the drawings. (FIG.
28) shows a block diagram of a conventional silencer, and shows a case where two noise sources
are controlled at two points. In FIG. 28, microphones 1a to 1d are first and second noise
detectors, FIR filters 2a to 2d are adaptive filters, and 3a to 3h are transfer function correction
circuits, and 4a to 4h. Is an LMS operator which is a coefficient operator, a coefficient adder
which is a coefficient updater, 5a to 5d, 6a and 6b are adders, and 8a and 8b are speakers.
03-05-2019
2
[0007]
The operation of the noise reduction device configured as described above will be described
below. First, noise from a certain noise source is detected by the microphone 1a. The detection
signals are input to the adaptive filters 2a and 2c and the FIR filters 3a, 3b, 3e and 3f. Then, the
noise signal subjected to signal processing by the adaptive filter 2a is output to the adder 6a, and
the noise signal subjected to signal processing by the adaptive filter 2c is output to the adder 6b.
Similarly, the noise from the other noise source is detected by the microphone 1b, the detected
signal is processed by the adaptive filters 2b and 2d, and output to the adders 6a and 6b,
respectively. Therefore, the adder 6a adds the outputs of the adaptive filters 2a and 2b and
reproduces them from the speaker 8a, and the adder 6b adds the outputs of the adaptive filters
2c and 2d and reproduces them from the speaker 8b. Then, in the microphones 1c and 1d, this
reproduced sound interferes with the noise from the two noise sources, and the noise is
attenuated by changing the coefficients of the adaptive filters 2a to 2d.
[0008]
Considering coefficient updating for the adaptive filter 2a, detection signals of the microphones
1c and 1d arranged at two control points are respectively input to the LMS arithmetic units 4a
and 4b, and the detection sound and the outputs of the FIR filters 3a and 3b The LMS operation
(least squares method) is performed so that the detection signals of the microphones 1c and 1d
become minimum, and the coefficients obtained by the coefficient adder 5a are added to update
the coefficients of the adaptive filter 2a. As a result, among the noise detected by the microphone
1a, the noise controlled by the speaker 8a is attenuated at the point where the microphones 1c
and 1d are disposed. Here, the transfer function C11 from the speaker 8a to the microphone 1c
is approximated in advance to the FIR filter 3a, and the transfer function C12 from the speaker
8a to the microphone 1d is approximated to the FIR filter 3b in advance as coefficients. The same
applies to the other adaptive filters 2b to 2d. This method is described in the Multiple Error
Filtered-x LMS algorithm (for example, S. J. Elliott, I. M. Stothers and P. A. Nelson, ("A multiple
error LMS algorithm and its application to the active control of sound and vibration. "IEEE Trans.
Acoust. Speech Signal Process. ASSP-35, pp 1423-1434 (1987))). If this is generally expressed by
a formula, it can be expressed as follows. Now, let m control speakers for one noise and l
microphones at the control point position.
[0009]
Applying this equation to (FIG. 11) and looking at the noise detected by the microphone 1a,
03-05-2019
3
[0010]
【0013】よって
[0011]
As described above, by using this algorithm, it is possible to control a plurality of noises at a
plurality of control points.
[0012]
However, in the configurations shown in FIGS. 27 and 28, for example, among the noises
detected by the microphones 1a and 1b as shown in FIG. 2 and FIG. If there is only a part where
the noise and the signal detected in 1d are strongly correlated with each other, as shown by the
solid line in FIG. It had the problem that it would diverge.
[0013]
Furthermore, when the characteristics of noise detected by the microphones 1c and 1d are
excessive in the low range as shown in FIG. 5, only the frequency band having a large level is
attenuated as shown in FIG. As a result, as shown in FIG. 30, there is also a problem that the low
band to the high band can not be attenuated uniformly.
[0014]
The present invention solves the above-mentioned problems, and its object is to perform the
signal processing according to the characteristics of the noise signal detected by the first noise
detector without subjecting the noise signal including any new delay element to the noise signal.
It is an object of the present invention to provide a noise reduction device capable of performing
adaptive signal processing, thereby obtaining a noise control effect suitable for hearing and
preventing addition of noise.
[0015]
SUMMARY OF THE INVENTION In order to achieve the above object, the silencer according to
the present invention comprises n first noise detectors each detecting noise of n noise sources; N
× m adaptive signal processing units that divide each noise signal detected by the first noise
detector into m speakers and perform adaptive signal processing, and the n × m adaptive signal
processing units Of the outputs subjected to signal processing, m adders for adding the outputs
of the adaptive signal processor so that they are grouped for each speaker to be reproduced, and
m speakers for reproducing the outputs of the adders A second noise detector disposed at the
03-05-2019
4
control point, and outputs of the respective second noise detectors are detected for each noise
signal detected by the first noise detector. Adjust the frequency characteristic according to the
characteristic and adjust the adjusted signal as the coefficient of each of the adaptive signal
processing units It is composed of n × l frequency characteristic adjustment units as update
information.
[0016]
With this configuration, it is possible to perform adaptive signal processing according to the
characteristics of the noise signal without performing signal processing including any new delay
element on the noise signal detected by the first noise detector. A suitable noise control effect
can be obtained and noise addition can be prevented.
[0017]
The first embodiment of the present invention will be described below with reference to the
drawings.
[0018]
(FIG. 1) shows a block diagram of a silencer according to a first embodiment of the present
invention.
In FIG. 1, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is an LMS operator
which is a coefficient operator, 6 is an adder, 7a and 7b are filter circuits which are frequency
characteristic adjustment units, and 8 is a speaker, as shown by a dotted line, an adaptive filter 2
and The FIR filter 3 and the LMS computing unit 4 constitute an adaptive signal processing unit.
[0019]
The operation of the noise reduction device configured as described above will be described
below.
First, noise from a certain noise source is detected by the microphone 1a.
03-05-2019
5
The detection signal is input to the adaptive filter 2a and the FIR filter 3a.
Then, the noise signal subjected to signal processing by the adaptive filter 2 a is output to the
adder 6.
Similarly, the noise from another noise source is detected by the microphone 1b, the detection
signal is input to the adaptive filter 2b and the FIR filter 3b, and the noise signal processed by the
adaptive filter 2b is output to the adder 6 Be done.
The adder 6 adds the outputs of the adaptive filters 2a and 2b and reproduces them from the
speaker 8.
Then, in the microphone 1c, the interference sound between this reproduced sound and the noise
from the two noise sources is detected.
The output of the microphone 1c is adjusted in frequency characteristics by the filter circuits 7a
and 7b, and the LMS arithmetic units 4a and 4b are output signals of the filter circuits 7a and 7b
by the outputs of the filter circuits 7a and 7b and the outputs of the FIR filters 3a and 3b,
respectively. The LMS operation (least squares method) is performed so as to minimize the
coefficient of the adaptive filters 2a and 2b.
Thus, the noise level is attenuated according to the characteristics of the filter circuits 7a and 7b
at the point where the microphone 1c is disposed. Here, assuming that the transfer function from
the speaker 8 to the microphone 1c is C11, the transfer function of the filter circuit 7a is G1, and
the transfer function of the filter circuit 7b is G2, the FIR filter 3a is C11 · G1 in advance, and the
FIR filter 3b is In the equation, C11 · G2 is approximated as coefficients in advance.
[0020]
Now, the correlation between the noise detected by the microphone 1a and the noise detected by
the microphone 1c is as shown in FIG. 2, and the correlation between the noise detected by the
microphone 1b and the noise detected by the microphone 1c is as shown in FIG. In some cases,
by setting the characteristics of the filter circuit 7a and the characteristics of the filter circuit 7b
03-05-2019
6
as shown in FIGS. 4 (a) and 4 (b), the muffling effect of the microphone 1c is (FIG. 4 (c)). As
shown in the above, a band with strong correlation is stably controlled while suppressing noise
addition.
[0021]
If the noise characteristic detected by the microphone 1c is not flat as shown in FIG. 5, the
characteristics of the filter circuits 7a and 7b are as shown in FIG. By setting the inverse
characteristic of the detected noise, the outputs of the filter circuits 7a and 7b have flat
frequency characteristics as shown in FIG. 7, so the muffling effect in the microphone 1c is a
solid line in FIG. As shown in FIG. 2, the control is uniformly controlled from the low band to the
high band.
[0022]
Furthermore, in order to make the muffling effect suitable for the sense of hearing, the
characteristics of the filter circuits 7a and 7b may be, for example, an A-curve filter, and an
equalizer may be used when finer adjustment is desired.
[0023]
Next, a second embodiment of the present invention will be described with reference to the
drawings.
(FIG. 9) shows a block diagram of a silencer according to a second embodiment of the present
invention.
In FIG. 9, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is an LMS operator
which is a coefficient operator, 6 is an adder, 7a and 7b are filter circuits which are frequency
characteristic adjusting units, 8 is a speaker, and 9a and 9b are frequency characteristic
adjusting units. It is a computing unit.
[0024]
03-05-2019
7
The operation of the noise reduction device configured as described above will be described
below.
When only the noise detected by the microphone 1a is generated, the frequency characteristic
calculator 9a determines the frequency characteristic of the noise signal detected by the
microphone 1c by FFT or the like, and uses the determined characteristic as the coefficient of the
filter circuit 7a. .
[0025]
Similarly, when only the noise detected by the microphone 1b is generated, the frequency
characteristic calculator 9b obtains the frequency characteristic of the noise signal detected by
the microphone 1c by FFT or the like, and the determined characteristic is obtained from the
filter circuit 7b. It is a coefficient.
[0026]
Then, when the coefficients of the filter circuits 7a and 7b are obtained, the noise signal from the
microphone 1c is subjected to signal processing by the coefficients and is input to the LMS
arithmetic units 4a and 4b.
[0027]
Now, if the frequency characteristics of the noise detected by the microphone 1a are as shown in
FIG. 10 and the frequency characteristics of the noise detected by the microphones 1b and 1c are
as shown in FIG. 11 and FIG. Since the characteristic of the circuit 7a is as shown in FIG. 13, the
output is the smaller the level of the frequency characteristic of the noise signal detected by the
microphone 1a among the noise detected by the microphone 1c as shown in FIG. Attenuate.
Similarly, the characteristic of the filter circuit 7b is as shown in FIG. 15, so the output is as
shown in FIG.
[0028]
Since the LMS computing units 4a and 4b calculate the coefficients of the adaptive filters 2a and
2b using this signal, the adaptive filters 2a and 2b mainly process only the band related to the
noise detected by the microphones 1a and 1b. , And it is stable, and the convergence is silenced
03-05-2019
8
quickly.
[0029]
Next, a third embodiment of the present invention will be described with reference to the
drawings.
(FIG. 17) shows a block diagram of a silencer according to a third embodiment of the present
invention.
In FIG. 17, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is an LMS operator
which is a coefficient operator, 6 is an adder, 7a and 7b are filter circuits which are frequency
characteristic adjusting units, 8 is a speaker, and 9a and 9b are frequency characteristic
adjusting units. It is a computing unit.
[0030]
The operation of the noise reduction device configured as described above will be described
below. The configuration of the present embodiment is similar to the configuration of the second
embodiment shown in FIG. 9, but the frequency characteristic calculators 9a and 9b receive
signals from the microphones 1a and 1b, respectively. It is the point that is done. The frequency
characteristic calculator 9a obtains the frequency characteristic of the noise signal detected by
the microphone 1a by FFT or the like, and uses the obtained characteristic as a coefficient of the
filter circuit 7a. Similarly, the frequency characteristic calculator 9b obtains the frequency
characteristic of the noise signal detected by the microphone 1b by FFT or the like, and uses the
obtained characteristic as the coefficient of the filter circuit 7b. Then, when the coefficients of the
filter circuits 7a and 7b are obtained, the noise signal from the microphone 1c is subjected to
signal processing by the coefficients and is input to the LMS computing units 4a and 4b.
[0031]
According to such a configuration, the same effect as in the case of (FIG. 9) is obtained, and
further noise is generated one by one as shown in FIG. 9 in order to obtain the characteristics of
the filter circuits 7a and 7b. It is not necessary, and it can be determined even if noise is
03-05-2019
9
generated simultaneously.
[0032]
(FIG. 18) shows a block diagram of a silencer according to a fourth embodiment of the present
invention.
In FIG. 18, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a, 7b is a filter circuit as a frequency characteristic adjustment unit, 8
is a speaker, and 10a, 10b is a frequency characteristic adjustment unit It is
[0033]
The operation of the noise reduction device configured as described above will be described
below. The correlation operation unit 10a obtains the correlation characteristic between the
noise detected by the microphone 1a and the noise detected by the microphone 1c, and uses the
obtained correlation characteristic as a coefficient of the filter circuit 7a. Similarly, the
correlation computing unit 10b obtains the correlation characteristic between the noise detected
by the microphone 1b and the noise detected by the microphone 1c, and uses the obtained
correlation characteristic as a coefficient of the filter circuit 7b. Then, when the coefficients of
the filter circuits 7a and 7b are obtained, the noise signal from the microphone 1c is subjected to
signal processing by the coefficients and is input to the LMS computing units 4a and 4b.
[0034]
According to such a configuration, as described in (FIG. 2) and (FIG. 3), the correlation
characteristic between the noise detected by the microphone 1a and the noise detected by the
microphone 1c, the noise detected by the microphone 1b, and the microphone 1c Even if the
correlation characteristic of the noise detected in the above is different, the adaptive filters 2a
and 2b perform signal processing only on the band where the correlation is strong, so that the
noise can be muffled stably while suppressing the addition of noise.
[0035]
03-05-2019
10
FIG. 19 shows a block diagram of a silencer according to a fifth embodiment of the present
invention.
In FIG. 19, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
3a and 3b transfer function correction circuits, and FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a, 7b is a filter circuit as a frequency characteristic adjusting unit, 8 is
a speaker, and 11a, 11b is a frequency characteristic adjusting unit as a coherence operation 12a
and 12b are Hilbert transformers which are frequency characteristic adjustment units.
[0036]
The operation of the noise reduction device configured as described above will be described
below. The coherence calculator 11a obtains the coherence characteristic of the noise detected
by the microphone 1a and the noise detected by the microphone 1c, Hilbert transforms the
determined coherence characteristic by the Hilbert transformer 12a, and uses the output as a
coefficient of the filter circuit 7a. Similarly, the coherence computing unit 11b determines the
coherence characteristic of the noise detected by the microphone 1b and the noise detected by
the microphone 1c, Hilbert transforms the determined coherence characteristic by the Hilbert
transformer 12b, and the output thereof is a coefficient of the filter circuit 7b. I assume. Then,
when the coefficients of the filter circuits 7a and 7b are obtained, the noise signal from the
microphone 1c is subjected to signal processing by the coefficients and input to the LMS
computing units 4a and 4b.
[0037]
According to such a configuration, as in the case of (FIG. 18), the correlation characteristic
between the noise detected by the microphone 1a and the noise detected by the microphone 1c,
the noise detected by the microphone 1b, and the noise detected by the microphone 1c Even if
the correlation characteristics are different, the adaptive filters 2a and 2b perform signal
processing only on the strongly correlated bands, so that noise can be muted stably while
suppressing noise addition.
[0038]
(FIG. 20) shows a block diagram of a silencer according to a sixth embodiment of the present
03-05-2019
11
invention.
In FIG. 20, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a to 7d is a filter circuit of a frequency characteristic adjustment unit,
8 is a speaker, and 9a and 9b are frequency characteristic adjustment units. 10a and 10b are
correlation operation units which are frequency characteristic adjustment units.
[0039]
The configuration of the present embodiment is a combination of the embodiments shown in FIG.
9 and FIG. 18 and adaptive filters 2a and 2b are bands having large levels of noise signals
detected by microphones 1a and 1b, respectively. Moreover, since only the strongly correlated
band is signal-processed, it is possible to mute more stable and fast convergence speed.
[0040]
(FIG. 21) shows a block diagram of a muffler according to a seventh embodiment of the present
invention.
In FIG. 21, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
3a and 3b transfer function correction circuits, and FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a to 7d is a filter circuit of a frequency characteristic adjustment unit,
8 is a speaker, and 9a and 9b are frequency characteristic adjustment units. Reference numerals
11a and 11b denote coherence operation units which are frequency characteristic adjustment
units, and 12a and 12b denote Hilbert transformers which are frequency characteristic
adjustment units.
[0041]
The configuration of the present embodiment is a configuration in which the embodiments of
(FIG. 9) and (FIG. 19) are added, and as the effect thereof, each effect can be obtained
simultaneously.
[0042]
03-05-2019
12
(FIG. 22) shows a block diagram of a muffler according to an eighth embodiment of the present
invention.
In FIG. 22, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a and 7d are filter circuits as frequency characteristic adjusting units,
8 is a speaker, and 9a and 9b are frequency characteristic adjusting units. 10a and 10b are
correlation operation units which are frequency characteristic adjustment units.
[0043]
The configuration of the present embodiment is a configuration in which the embodiments of
(FIG. 17) and (FIG. 18) are added, and as the effect, each effect can be obtained simultaneously.
[0044]
FIG. 23 shows a block diagram of a silencer according to a ninth embodiment of the present
invention.
In FIG. 23, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a and 7d are filter circuits as frequency characteristic adjusting units,
8 is a speaker, and 9a and 9b are frequency characteristic adjusting units. Reference numerals
11a and 11b denote coherence operation units which are frequency characteristic adjustment
units, and 12a and 12b denote Hilbert transformers which are frequency characteristic
adjustment units.
[0045]
The configuration of the present embodiment is a configuration in which the embodiments of
(FIG. 17) and (FIG. 19) are added, and as the effect, each effect can be obtained simultaneously.
[0046]
03-05-2019
13
(FIG. 24) shows a block diagram of a muffler according to a tenth embodiment of the present
invention.
In FIG. 24, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a, 7d is a filter circuit as a frequency characteristic adjusting unit, 8 is
a speaker, and 10a, 10b is a frequency characteristic adjusting unit It is
[0047]
Here, the adaptive filters 2a and 2b are microphones by making the characteristics of the filter
circuits 7a and 7b the opposite characteristics in order to flatten the noise detected by the
microphone 1c as described in the example of (FIG. 1) Of the noise detected in 1c, noise can be
muted uniformly in the strongly correlated band.
[0048]
Furthermore, in order to make the muffling effect suitable for the sense of hearing, the
characteristics of the filter circuits 7a and 7b may be, for example, an A-curve filter, and an
equalizer may be used when finer adjustment is desired.
[0049]
(FIG. 25) shows a block diagram of a muffler according to the eleventh embodiment of the
present invention.
In FIG. 25, microphones 1a to 1c are first and second noise detectors, 2a and 2b adaptive filters,
and 3a and 3b transfer function correction circuits FIR filters 4a and 4b. Is a coefficient
calculator, 6 is an adder, 7a to 7d is a filter circuit for a frequency characteristic adjustment unit,
8 is a speaker, and 11a and 11b are frequency characteristic adjustment units for coherence
operation 12a and 12b are Hilbert transformers which are frequency characteristic adjustment
units.
[0050]
In this configuration, by making the characteristics of the filter circuits 7a and 7b the inverse
characteristics of the noise detected by the microphone 1c, or by using an A curve filter or an
03-05-2019
14
equalizer, the same effect as in the case of (FIG. 24) can be obtained. it can.
[0051]
A twelfth embodiment of the present invention will now be described with reference to the
drawings.
(FIG. 26) shows a block diagram of a muffler according to a twelfth embodiment of the present
invention.
In FIG. 26, microphones 1a to 1d are first and second noise detectors, FIR filters 2a to 2d are
adaptive filters, 3a to 3h are transfer function correction circuits, and 4a to 4h. Is an LMS
operator that is a coefficient operator, 5a to 5d is a coefficient adder that is a coefficient updater,
6a to 6b is an adder, and 7a to 7d is a frequency characteristic adjustment unit, a filter circuit, 8a
8b are speakers.
[0052]
The operation of the noise reduction device configured as described above will be described
below. First, noise from a certain noise source is detected by the microphone 1a. The detection
signal is input to adaptive filters 2a and 2c and FIR filters 3a, 3b, 3e and 3f. Then, the noise
signal subjected to signal processing by the adaptive filter 2a is output to the adder 6a, and the
noise signal subjected to signal processing by the adaptive filter 2c is output to the adder 6b.
Similarly, the noise from the other noise source is detected by the microphone 1b, the detected
signal is processed by the adaptive filters 2b and 2d, and output to the adders 6a and 6b,
respectively. The adder 6a adds the outputs of the adaptive filters 2a to 2b and reproduces them
from the speaker 8a, and the adder 6b adds the outputs of the adaptive filters 2c to 2d and
reproduces them from the speaker 8b. Then, in the microphones 1c and 1d, the interference
sound between this reproduced sound and the noise from the two noise sources is detected. The
output of the microphone 1c is adjusted in frequency characteristics by the filter circuits 7a and
7c, and the output of the microphone 1d is similarly adjusted in frequency characteristics by the
filter circuits 7b and 7d. The LMS arithmetic units 4a to 4h perform LMS operation (least squares
method) so that the output signals of the filter circuits 7a to 7d are minimized by the outputs of
the filter circuits 7a to 7d and the outputs of the FIR filters 3a to 3h, respectively. The coefficient
adders 5a to 5d add the outputs of the LMS arithmetic units 4a to 4h, and update the coefficients
03-05-2019
15
of the adaptive filters 2a to 2d. As a result, the noise level is attenuated according to the
characteristics of the filter circuits 7a to 7d at the points where the microphones 1c and 1d are
arranged. Here, the transfer function from the speaker 8a to the microphone 1c is C11, the
transfer function from the speaker 8a to the microphone 1d is C12, the transfer function from
the speaker 8b to the microphone 1c is C21, and the transfer function from the speaker 8b to the
microphone 1d is C22. Assuming that the transfer function of the filter circuit 7a is G1, the
transfer function of the filter circuit 7b is G2, the transfer function of the filter circuit 7c is G3,
and the transfer function of the filter circuit 7d is G4, the FIR filter 3a has C11 · G1 in advance,
The FIR filter 3b is previously C12 · G2, the FIR filter 3c is C11 · G3 in advance, the FIR filter 3d
is C12 · G4 in advance, the FIR filter 3e is C21 · G1 in advance, and the FIR filter 3f is in advance
C22 · G2 is added to the FIR filter 3g in advance to C21 · G3 , Pre-C22 · G4 is approximated as
coefficients respectively to the FIR filter 3h.
[0053]
Now, as can be seen by comparing (FIG. 26) with (FIG. 1), the two inputs in each LMS arithmetic
unit 4a to 4h, that is, the outputs of the FIR filters 3a to 3h and the outputs of the filter circuits
7a to 7d Since the characteristics of the circuits 7a to 7d are included, the adaptive filters 2a to
2d change in accordance with these characteristics. By this, as in the case of (FIG. 1), noise can be
obtained by obtaining a uniform muffling effect from the low band to the high band, obtaining a
muffling effect according to the audibility effect, or controlling only a high correlation band.
Addition can be suppressed.
[0054]
The first to twelfth examples have been described above. Although the example shows two noise
sources, two control speakers, and two control points, when the number of each increases, the
configuration of FIG. 1 (FIG. 9) As the configuration is changed, the configuration may be
expanded as appropriate.
[0055]
Further, by making the filter circuits 7a to 7d into the configuration of the frequency
characteristic adjusting section in (FIG. 9) and (FIG. 17) to (FIG. 25), even from the second
embodiment, even when the noise sources increase. The effects of the eleventh embodiment can
be obtained.
03-05-2019
16
[0056]
As described above, according to the noise suppressor of the present invention, the frequency
characteristics of the coefficient update information of each adaptive signal processing unit that
processes the noise signal are adjusted according to the characteristics of each noise source. By
using the adjustment unit, each adaptive signal processing unit mainly controls only the strongly
correlated band even when the strongly correlated band is different in each noise signal, noise
addition is suppressed and noise can be muted stably.
[0057]
Also, by setting the characteristics of the frequency characteristic adjusting section in
consideration of the aural feeling of an A-curve filter or the like, it is possible to muffle the
offensive high frequency band as well as the low frequency band even for excessive low
frequency band noise.
[0058]
Further, by adjusting the frequency characteristic of the output of the second noise detector,
which is the coefficient update information of the adaptive signal processing unit, a delay
element is not added to the adaptive signal processing unit. It is possible to realize a silencer
having an excellent effect such as shortening the distance between the noise detector and the
second noise detector, or making it possible to increase the amount of attenuation.
[0059]
Brief description of the drawings
[0060]
1 is a block diagram of a first embodiment according to the noise suppressor of the present
invention
[0061]
Fig. 2 Correlation characteristics between noise in the microphone 1a and noise in the
microphone 1c
[0062]
Fig. 3 Correlation characteristics between noise in the microphone 1b and noise in the
03-05-2019
17
microphone 1c
[0063]
4 shows the frequency characteristics of the filter circuits 7a and 7b and the noise control effect
of the microphone 1c.
[0064]
Fig. 5 Frequency characteristics of noise in the microphone 1c
[0065]
Fig. 6 Frequency characteristics of the filter circuits 7a and 7b
[0066]
Fig. 7 Frequency characteristics of the output signals of the filter circuits 7a and 7b
[0067]
Figure 8 shows the noise control effect of the microphone 1c
[0068]
9 is a block diagram of a second embodiment according to the noise suppressor of the present
invention
[0069]
Fig. 10 Frequency characteristics of noise in the microphone 1a
[0070]
Fig. 11 Frequency characteristics of noise in the microphone 1b
[0071]
Figure 12: Frequency characteristics of noise in the microphone 1c
03-05-2019
18
[0072]
Fig. 13 Frequency characteristics of the filter circuit 7a
[0073]
Fig. 14 Frequency characteristics of the output signal of the filter circuit 7a
[0074]
Fig.15 Frequency characteristics of filter circuit 7b
[0075]
Figure 16: Frequency characteristics of the output signal of the filter circuit 7b
[0076]
17 is a block diagram of a third embodiment according to the noise suppressor of the present
invention
[0077]
18 is a block diagram of a fourth embodiment according to the noise suppressor of the present
invention
[0078]
19 is a block diagram of the fifth embodiment according to the noise suppressor of the present
invention
[0079]
<Figure 20> The block diagram of the 6th example which relates to the silencer of this invention
[0080]
21 is a block diagram of a seventh embodiment according to the noise suppressor of the present
invention
[0081]
03-05-2019
19
<Figure 22> The block diagram of the 8th example which relates to the silencer of this invention
[0082]
23 is a block diagram of a ninth embodiment according to the noise suppressor of the present
invention.
[0083]
24 block diagram of the tenth embodiment according to the noise suppressor of the present
invention
[0084]
25 is a block diagram of an eleventh embodiment according to the noise suppressor of the
present invention
[0085]
26 is a block diagram of a twelfth embodiment according to the noise suppressor of the present
invention.
[0086]
Fig. 27 is a block diagram showing a conventional muffling apparatus
[0087]
Fig. 28 A block diagram showing a conventional noise reduction device
[0088]
Fig. 29 shows the noise control effect of the microphone 1c by the conventional noise suppressor
[0089]
Figure 30 A diagram showing the desired noise control effect in the microphone 1c
[0090]
03-05-2019
20
Fig. 31 illustrates the noise control effect of the microphone 1c by the conventional noise
suppressor
[0091]
Explanation of sign
[0092]
1a, 1b, 1c, 1d Microphones 2a, 2b, 2c, 2d Adaptive filters 3a, 3b, 3c, 3d, 3e, 3f, 3g, 3h FIR filters
4a, 4b, 4c, 4d, 4e, 4g, 4h LMS Arithmetic units 5a, 5b, 5c, 5d Coefficient adders 6, 6a, 6b Adders
7a, 7b, 7c, 7d Filter circuits 8, 8a, 8b Speakers 9a, 9b Frequency characteristic calculators 10a,
10b Correlation calculators 11a, 11b Coherence operator 12a, 12b Hilbert transformer
03-05-2019
21
Документ
Категория
Без категории
Просмотров
0
Размер файла
32 Кб
Теги
jph07281672
1/--страниц
Пожаловаться на содержимое документа