close

Вход

Забыли?

вход по аккаунту

?

JPWO2015049921

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 JPWO2015049921
Abstract We propose a signal processing device that emphasizes the target signal to emphasize
the target signal and suppress other components without increasing the array size or the number
of sensors. The signal processing device is configured to generate a first array processing signal
by partially emphasizing a predetermined signal with respect to signals received from a plurality
of sensors, and a plurality of sensors other than the first array processing unit. And a
decorrelation unit for removing a signal component correlated with the signal received from the
auxiliary sensor from the first array processing signal to generate a decorrelation signal.
Signal processing device, media device, signal processing method and signal processing program
[0001]
The present invention relates to a technique for emphasizing or suppressing a signal using
directivity formed by a plurality of sensors.
[0002]
In the above technical field, Non-Patent Documents 1 and 2 process a plurality of sensor signals
to generate an enhanced target signal, and suppress the target signal to generate a pseudojamming that relatively enhances the interfering signal. There is a description of enhancing the
target signal and suppressing the jamming signal by generating the signal and subtracting the
component correlated with the spurious jamming signal from the emphasized target signal.
03-05-2019
1
This technique forms directivity using phase differences of signals based on differences in spatial
positions among a plurality of sensors, and emphasizes or suppresses a specific signal based on
the formed directivity. In addition, Non-Patent Documents 3 and 4 describe a configuration in
which the techniques of Non-Patent Documents 1 and 2 are combined in a plurality of frequency
bands from low to high frequencies by using a plurality of arrays having different sensor
intervals. is there.
[0003]
1982
January, I.E.E. Transactions-On Signal Processing, Vol. 30, No. 1, (IEEE TRANSACTIONS ON
ANTENNAS AND PROPAGATIONS, VOL. 30, NO. 1, PP. 27-34 , Jan. 1982) 27Page 34 2001,
“Microphone Arrays”, Chapter 5, Springer, Berlin Heidelberg New York (CH.5, MICROPHONE
ARRAYS, SPRINGER, BERLIN HEIDELBERG NEW YORK, 2001. In May 1985, Journal of Actual
Society of America, Vol. 78, No. 5, (JOURNAL OF ACOUSTICAL SOCIETY OF AMERICA, VOL.5, No.
5, PP.1508-1518, May 1985) 1508-1518 May 1995, I.E. Proceedings of International
Conference on Axics Speech and Signal Processing, Volume V, (IEEE PROCEEDINGS OF
INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNNAL PROCESSING, VOL. V,
PP. 2995- 2998, May 1995) 2995-2998 2008, "Handbook of Speech Processing", 4th. Chapter
5, Springer, Berlin Heidelberg New York (CH.46.5, HANDBOOK OF SPEECH PROCESSING,
SPRINGER, BERLIN HEIDELBERG NEW YORK, 2008. 1984, "Adaptive Filtering, Prediction and
Control", Chapter 8, Prentice Hall, Englewood Cliffs (SEC. 8, ADAPTIVE FILTERING G,
PREDICTION AND CONTROL, PRENTICE-HALL, ENGLEWOOD CLIFFS, 1984 . In January 1992,
I.E.E. Signal Processing Magazine, Vol. 9, No. 1, (IEEE TRANSACTIONS ON ACOUSTICS, SPEECH,
AND SIGNAL PROCESSING, VOL. 9, NO. 1, PP. 14-37, Jan.
1992) 14Page 37, June 1983, I.E.E. Transactions. Axics Speech and Signal Processing, Vol. 31,
No. 6, (IEEE SIGNAL PROCESSING MAGAZINE, VOL. 31,, NO.3, PP. 609-615, Jun. 1983) 609∼
615 ページ
[0004]
However, the techniques described in the above-mentioned non-patent documents 1 and 2 can
not form sufficient directivity for low frequency signal components. This is because when using a
common sensor with a medium frequency and a low frequency where the wavelength is long
compared to the medium frequency, a relatively narrow sensor interval can not generate a
03-05-2019
2
sufficiently large inter-signal phase difference between a plurality of sensors. It is. Further, the
techniques described in Non-Patent Documents 3 and 4 have problems of cost increase due to
the increased number of sensors and increase in array size due to wide sensor intervals
corresponding to low frequencies.
[0005]
For this reason, with the techniques described in these documents, it is possible to use the
directivity of the sensor array to enhance or suppress a wideband signal without increasing the
size of the sensor array or increasing the number of sensors. could not.
[0006]
An object of the present invention is to provide a technique for solving the above-mentioned
problems.
[0007]
In order to achieve the above object, a signal processing device according to the present
invention performs first array processing for partially emphasizing a predetermined signal with
respect to signals received from a plurality of sensors to generate a first array processing signal.
A decorrelation unit for removing a signal component correlated with the signal received from
the auxiliary sensor other than the plurality of sensors from the first array processing signal to
generate a decorrelation signal.
[0008]
In order to achieve the above object, a signal processing method according to the present
invention comprises the steps of: partially emphasizing a predetermined signal with respect to
signals received from a plurality of sensors to generate an array processing signal; And C.
removing the signal component correlated with the signal received from the other auxiliary
sensor from the array processing signal to generate a decorrelation signal.
[0009]
In order to achieve the above object, a signal processing program according to the present
invention generates an array processing signal by partially emphasizing a predetermined signal
with respect to signals received from a plurality of sensors; Removing from the array processed
signal a signal component correlated with a signal received from another auxiliary sensor to
generate a decorrelation signal.
03-05-2019
3
[0010]
In order to achieve the above object, a media device according to the present invention receives
from a plurality of sensors disposed on the front surface, an auxiliary sensor disposed at a
position different in acoustic characteristics from the plurality of sensors, and the plurality of
sensors. An array processing unit that generates an array processing signal by partially
emphasizing a predetermined signal with respect to the received signal, and a signal component
correlated with the signal received from the auxiliary sensor from the array processing signal by
decorrelation And a decorrelation unit that generates a signal.
[0011]
According to the present invention, the directivity of the sensor array can be used to enhance or
suppress a broadband signal without increasing the size of the sensor array.
[0012]
It is a block diagram which shows the structure of the signal processing apparatus which
concerns on 1st Embodiment of this invention.
It is a block diagram which shows the structure of the signal processing apparatus which
concerns on 2nd Embodiment of this invention.
It is a block diagram showing composition of array processing section 203 concerning a 2nd
embodiment of the present invention.
It is a block diagram which shows the structure of the correlation removal part which concerns
on 2nd Embodiment of this invention.
It is a block diagram which shows the structure of the correlation removal part which concerns
on 3rd Embodiment of this invention.
It is a block diagram which shows the structure of the signal processing apparatus which
concerns on 4th Embodiment of this invention.
03-05-2019
4
It is a block diagram which shows the structure of the mixing part which concerns on 4th
Embodiment of this invention.
It is a block diagram which shows the structure of the signal processing apparatus which
concerns on 5th Embodiment of this invention. It is a block diagram which shows the hardware
constitutions of the signal processing apparatus which concerns on 5th Embodiment of this
invention. It is a flowchart explaining the flow of a process of the signal processing apparatus
which concerns on 5th Embodiment of this invention. It is a block diagram which shows the
structure of the array processing part 706 which concerns on 5th Embodiment of this invention.
It is a block diagram which shows the structure of the array processing part 708 which concerns
on 5th Embodiment of this invention. It is a block diagram which shows the structure of the
signal processing apparatus which concerns on 6th Embodiment of this invention. It is a block
diagram which shows the structure of the signal processing apparatus which concerns on 7th
Embodiment of this invention. It is a block diagram which shows the structure of the signal
processing apparatus which concerns on 8th Embodiment of this invention. It is a block diagram
which shows the 1st application example of 1st thru | or 8 embodiment of this invention. It is a
block diagram which shows the 2nd application example of 1st thru | or 8th embodiment of this
invention.
[0013]
Hereinafter, embodiments of the present invention will be exemplarily described in detail with
reference to the drawings. However, the component described in the following embodiment is an
illustration to the last, and it is not a thing of the meaning which limits the technical scope of this
invention only to them. Note that "audio signal" in the following description is a direct electrical
change that occurs in accordance with speech or other sounds, and is for transmitting speech or
other sounds, and is not limited to speech.
[0014]
First Embodiment A signal processing apparatus 100 according to a first embodiment of the
present invention will be described with reference to FIG. The signal processing device 100 is a
device that uses signals from a plurality of sensors 101 to enhance or suppress a wide band
signal.
03-05-2019
5
[0015]
As shown in FIG. 1, the signal processing apparatus 100 includes an array processing unit 103
and a decorrelation unit 104.
[0016]
The array processing unit 103 generates an array processing signal 110 by partially emphasizing
a predetermined signal with respect to the signals 105 received from the number of sensors 101.
[0017]
The decorrelation unit 104 eliminates the signal component correlated with the signal 121
received from the auxiliary sensor 102 different from the plurality of sensors 101 from the array
processing signal 110 to generate the decorrelation signal 112.
[0018]
With such a configuration, the signal processing apparatus 100 can effectively enhance or
suppress the wide band signal using the signal from the sensor array.
[0019]
Second Embodiment << Overall Configuration >> A signal processing apparatus 200 as a second
embodiment of the present invention will be described with reference to FIG. 2 to FIG.
The signal processing device 200 of the present embodiment can be applied to signal
enhancement of various media devices such as digital cameras, video recorders, personal
computers, mobile phones, televisions, voice recorders, game machines, vending machines and
the like.
That is, target signals such as voice, music, and environmental sounds can be emphasized with
respect to signals (noise or disturbing signals) superimposed thereon.
03-05-2019
6
However, the present invention is not limited to this, and is applicable to any signal processing
device that requires enhancement of a specific signal included in the input signal.
[0020]
In the present embodiment, as an example of signal processing, a signal emphasizing device for
emphasizing speech superimposed on background noise and interference signals will be
described.
The signal processing apparatus 200 according to the present embodiment is, for example, a
background noise or the like that interferes with a voice command in a form such as a voice
recognition apparatus for controlling the television using a voice command from a position
distant from the television. Properly suppress the signals of (1) based on their directions. Briefly,
for high frequency components, a plurality of sensor signals are processed to form directivity
and to emphasize speech. In addition, for low frequency components, signals other than voice are
suppressed by eliminating components correlated with the signal of the auxiliary sensor, using
the signal of the auxiliary sensor placed at a position where there is a small amount of voice
input as a reference signal. And emphasize the voice.
[0021]
The overall configuration of the signal processing apparatus 200 in this embodiment is shown in
FIG. 2A. The schematic configuration is the same as that shown in FIG. 1, and includes an array
processing unit 203 and a correlation removal unit 204.
[0022]
The array processing unit 203 performs array processing on the input signals 205 received from
each sensor of the sensor array 201 to form directivity, and outputs the directivity as a first array
processing signal 210 in which the target signal is emphasized.
[0023]
The correlation removal unit 204 uses the input signal 211 received from the auxiliary sensor
03-05-2019
7
202 as a reference signal, eliminates a component correlated with the input signal 211 from the
array processing signal 210, and outputs the result as a correlation removal signal 212.
The input signal 211 is correlated with the target signal by disposing the sensor 202 at a
position in the acoustic space where the target signal is difficult to be input, or by placing an
acoustic shield in the vicinity where the target signal is difficult to be input. Avoid including
ingredients as much as possible. The decorrelation signal 212 is supplied to the output terminal
209 as an output signal.
[0024]
The array processing unit 203 mainly suppresses high frequency components other than the
target signal based on directivity, and the decorrelation unit 204 mainly suppresses low
frequency components other than the target signal based on decorrelation.
[0025]
<< Configuration of Array Processing Unit 203 >> FIG. 2B is a block diagram showing the
configuration of the array processing unit 203. As shown in FIG.
The array processing unit 203 includes M filters 231 and an adder 233. Here, М is a natural
number representing the number of sensors. An input signal 205 from the sensor array 201 is
supplied to the filter 231. The filter 231 filters these input signals 205 and supplies the obtained
output signal 232 to the adder 233. The adder 233 adds all the signals supplied from the filter
231, and outputs the addition result as an array processing signal 210 in which the target signal
is emphasized and the other components are suppressed.
[0026]
The configuration shown in FIG. 2B is known as a Filter-and-Sum Beamformer. Also, when all the
filters 231 are finite impulse response (FIR) filters, only one of the tap coefficients of each filter is
1 and all other coefficients are zero, FIG. It becomes a configuration known as -and-Sum
Beamformer). More precisely, the non-zero tap coefficients are set to rotate (steer) the wavefront
direction of a plane wave coming from a particular direction. The filter sum beamformer and the
delay sum beamformer are disclosed in Non-Patent Document 5. The array processing unit 203
03-05-2019
8
is known as a fixed beamformer for constructing a generalized sidelobe canceller (Griffith's gym
beamformer). Non-patent documents 1 and 2 disclose in detail an exemplary configuration and
operation of the array processing unit 203.
[0027]
<< Configuration of Correlation Removal Unit 204 >> FIG. 3 is a block diagram showing a first
configuration example of the correlation removal unit 204. As shown in FIG. The decorrelation
unit 204 includes an adaptive filter 301 and a subtractor 302. The adaptive filter 301 receives
an input signal 211 which is dominated by components other than the target signal, and supplies
the result of filtering to the subtractor 302 as an output 311. Another input of the subtractor
302 is supplied with an array processed signal 210 in which the target signal is enhanced and
the other components are suppressed. The subtractor 302 subtracts the output 311 of the
adaptive filter 301 from the array processing signal 210 and outputs the difference as a
decorrelation signal 212. The decorrelation signal 212 is fed back to the adaptive filter 301 as an
error. The adaptive filter 301 finds the correlation between the decorrelation signal 212 and the
input signal 211, and sequentially updates the filter coefficient according to the magnitude of the
correlation. As the filter coefficient update algorithm, various algorithms such as LMS algorithm
and NLMS algorithm can be used. The details of the coefficient update algorithm are disclosed in
Non-Patent Document 6 and the like.
[0028]
In the configuration of FIG. 3, the input signal 211 is divided into a plurality of frequency bands
by a filter bank, filtering is performed by independent adaptive filters in each frequency band,
and the input signal 211 is divided into a plurality of frequency bands by a filter bank, It is also
possible to subtract the filtering result from the input signal 210 band-divided by an independent
subtracter in each frequency band, and combine the subtraction results into one band in a filter
bank to form a decorrelation signal 212. At this time, the output of each subtractor is fed back to
each adaptive filter. The adaptive filter 301 calculates the correlation between the fed back
subtractor output and the band-divided input signal 210, and sequentially updates the adaptive
filter coefficient according to the magnitude of the correlation. Such an arrangement is known as
subband filtering. The details of subband filtering are disclosed in Non-Patent Document 7.
[0029]
03-05-2019
9
The array processing unit 203 uses directivity, and the correlation removal unit 204 suppresses
components other than the target signal by performing correlation removal based on the
reference signal 211. In particular, the array processing unit 203 operates effectively for high
frequency components, and the decorrelation unit 204 operates effectively for low frequency
components. Since the decorrelation unit 204 eliminates low frequency components, even if the
sensor array 201 is small, signals in a wide frequency band from low frequencies to high
frequencies can be suppressed. According to the above configuration, the wide band signal can
be suppressed and the target signal can be sufficiently emphasized without an increase in array
size or an increase in the number of sensors.
[0030]
Third Embodiment A signal processing apparatus according to a third embodiment of the present
invention will be described with reference to FIG. The second embodiment differs from the
second embodiment in that a converting unit 441 is added to the correlation removing unit 204
according to the present embodiment. The other configurations and operations are the same as
those of the second embodiment, and therefore, the same symbols are attached to the same
configurations and operations, and the detailed description thereof will be omitted. Here, only
differences in the configuration of the correlation removing unit 204 will be described. .
[0031]
FIG. 4 is a block diagram showing the configuration of the correlation removal unit 204. As
shown in FIG. The decorrelation unit 204 includes a transform unit 441, an adaptive filter 301,
and a subtractor 302. The conversion unit 441 generates the frequency domain signal by
receiving the input signal 211 dominated by components other than the target signal and
decomposing it into a plurality of frequency components. The adaptive filter 301 receives this
frequency domain signal, performs weighted addition of signal values corresponding to a
plurality of frequency components, and supplies the obtained addition result to the subtractor
302 as an output 311. Another input of the subtractor 302 is supplied with an array processing
signal 210 in which the target signal is emphasized. The subtractor 302 subtracts the output 311
of the adaptive filter 301 from the array processing signal 210 and outputs the difference as a
first decorrelation signal 212.
[0032]
03-05-2019
10
The decorrelation signal 212 is fed back to the adaptive filter 301 as an error. The adaptive filter
301 finds the correlation between the decorrelation signal 212 and the output signal 442 of the
converter 441 and sequentially updates the filter coefficient according to the magnitude of the
correlation.
[0033]
By the operation of the decorrelation unit 204 described above, components that are correlated
with components other than the target signal included in the decorrelation signal 212 are
minimized. As a result, the decorrelation signal 212 is a signal in which the target signal is
emphasized and the others are suppressed. The details of the configuration of the conversion
unit 441 shown in FIG. 4 are disclosed in Non-Patent Document 8.
[0034]
Fourth Embodiment A signal processing apparatus 500 as a fourth embodiment of the present
invention will be described with reference to FIG. Compared with the first embodiment, the signal
processing apparatus 500 is different in that a mixing unit 501 is added and a mixed signal of
the decorrelation signal 212 and the array processing signal 210 is supplied to the output
terminal 209 as an output 511. . The other configurations and operations are the same as those
of the first embodiment, and therefore, the same symbols are attached to the same configurations
and operations and the detailed description thereof is omitted, and only the operation of the
mixing unit 501 will be described here.
[0035]
The decorrelation signal 212, which is the output of the decorrelation unit 204, and the array
processing signal 210, which is the output of the array processing unit 203, are both supplied to
the mixing unit 501. The mixing unit 501 mixes the decorrelation signal 212 and the array
processing signal 210 to generate a mixed signal 511, and supplies the mixed signal 511 to the
output terminal 209.
03-05-2019
11
[0036]
FIG. 6 is a block diagram showing the configuration of the mixing unit 501. As shown in FIG. The
mixing unit 501 includes a low pass filter 601, a high pass filter 602, and an adder 603. The low
pass filter 601 receives the decorrelation signal 212, passes only the low frequency component
thereof, and supplies it to the adder 603. The high pass filter 602 receives the array processing
signal 210, passes only the high frequency component thereof, and supplies it to the adder 603.
The adder 603 mixes the low frequency component of the decorrelation signal 212 and the high
frequency component of the array processing signal 210 at a predetermined ratio, and outputs
the mixed signal 511 as a mixed signal.
[0037]
The adder 603 may perform simple addition or weighted addition. The weight for the output of
the low pass filter 601 and the output of the high pass filter 602 can be determined in advance,
or can be determined adaptively sequentially using the result of analyzing the frequency
spectrum of the signal. For example, in the case where a spectrum other than the target signal
has a spectrum biased to the low band, a larger weight is applied to the output of the low pass
filter 601. By such weight setting, the effect of the correlation removal unit 204 becomes
relatively larger than that of the array processing unit 203, and a larger suppression effect can
be expected in the addition signal.
[0038]
Similarly, the output of the low pass filter 601 and the setting of the pass band of the high pass
filter 602 can also be adaptively determined one by one according to the result of analysis of the
frequency spectrum of the signal. For example, when the spectrum other than the target signal
has a spectrum biased to the low band, the pass band of the low pass filter 601 is set wide, and
the pass band of the high pass filter 602 is set narrow. With such a pass band setting, the effect
of the correlation removal unit 204 becomes relatively larger than that of the array processing
unit 203, and a larger suppression effect can be expected in the addition signal.
[0039]
03-05-2019
12
Furthermore, the output of the low pass filter 601 and the pass band of the high pass filter 602
can be set according to the sensor interval. For example, when the sensor interval is narrow, the
pass band of the low pass filter 601 is set wide, and the pass band of the high pass filter 602 is
set narrow. With such a pass band setting, the effect of the correlation removal unit 204
becomes relatively larger than that of the array processing unit 203, and a larger suppression
effect can be expected in the addition signal.
[0040]
The low frequency components of the components other than the target signal are greatly
suppressed in the decorrelation signal 212 for the high frequency components and in the array
processing signal for the high frequency components. The mixed signal 511 obtained by mixing
the low frequency component of the decorrelation signal 212 and the high frequency component
of the array processing signal 210 has the advantages of both, so the suppression effect of
components other than the target signal is higher than any single signal. .
[0041]
As described above, in the present embodiment, the mixed signal 511 of the decorrelation signal
212 and the array processing signal 210 is supplied as an output to the output terminal 209
instead of the decorrelation signal 212. As a result, it is possible to obtain a high quality signal in
which components other than the target signal are further suppressed than in the second
embodiment. That is, the wide band signal can be suppressed and the target signal can be
sufficiently emphasized without an increase in array size or an increase in the number of sensors.
[0042]
Fifth Embodiment A signal processing apparatus 700 as a fifth embodiment of the present
invention will be described with reference to FIG. 7A. As compared with the fourth embodiment,
the signal processing apparatus 700 differs in that an array processing unit 706 and an array
processing unit 708 are added. The other configuration and operation are the same as in the first
embodiment, and therefore the same reference numerals are given to the same configuration and
operation and the detailed description thereof is omitted. Here, only the operations of the array
processing unit 706 and the array processing unit 708 Explain.
03-05-2019
13
[0043]
The array processing unit 706 cancels the signal component correlated with the decorrelation
signal 212 from each of the input signals 205 received from each sensor of the sensor array 201
using the decorrelation signal 212 as a reference signal, and the target signal is suppressed. It
outputs as an array processing signal 707. Since the array processing unit 706 cancels the signal
component correlated with the decorrelation signal 212 in which the target signal is emphasized,
the array processing signal 707 becomes a signal in which the target signal is suppressed and
the others are emphasized.
[0044]
The array processing unit 708 erases a signal component having a correlation with the array
processing signal 707 included in the array processing signal 210 using the array processing
signal 707 as a reference signal, and an array processing signal 713 in which components other
than the target signal are suppressed. Output as That is, the output signal 713 of the array
processing unit 708 is a signal in which the target signal is emphasized and the other signals are
suppressed.
[0045]
<< Configuration of Array Processing Unit 706 >> FIG. 8 is a block diagram showing the
configuration of the array processing unit 706. As shown in FIG. The array processing unit 706
includes M adaptive filters 801 and M subtractors 804. The decorrelation signal 212 is supplied
to the adaptive filter 801 from the decorrelation unit 204. The adaptive filter 801 filters these
signals and supplies the filtering result to the subtractor 804. The subtractor 804 is also supplied
with an input signal 205 received from each sensor of the sensor array 201. The subtractor 804
subtracts the output of the adaptive filter 801 from the input signal 205, and outputs the
resulting difference as an array processing signal 707. The array processing signal 707 is fed
back to the adaptive filter 801 as an error. The adaptive filter 801 finds the correlation between
the array processing signal 707 and the decorrelation signal 212, and sequentially updates the
filter coefficient according to the magnitude of the correlation.
03-05-2019
14
[0046]
By the operation described above, the component correlated with the target signal included in
the input signal 205 is minimized. As a result, the array processing signal 707 is a signal in
which the target signal is suppressed and the other signals are emphasized.
[0047]
The array processing unit 706 is known as a blocking matrix for constructing a generalized
sidelobe canceller (Griffith's gym beamformer). Examples of the configuration and operation of
the array processing unit 706 are disclosed in detail in Non-Patent Documents 1 and 2.
[0048]
<< Configuration of Array Processing Unit 708 >> FIG. 9 is a block diagram showing the
configuration of the array processing unit 708. As shown in FIG. The array processing unit 708
includes M adaptive filters 901, delay elements 902, and a subtractor 903. The adaptive filter
901 is supplied with an array processing signal 707. The adaptive filter 901 filters these signals
and supplies a filtering result 911 to the subtractor 903. The delay element 902 delays the array
processing signal 210 and supplies the delayed array processing signal to the subtractor 903.
The subtractor 903 subtracts all of the output signal 911 of the adaptive filter 901 from the
delayed array processing signal, and outputs the obtained result as an array processing signal
713.
[0049]
The array processing signal 713 is fed back to all adaptive filters 901 as an error. The adaptive
filter 901 finds the correlation between the array processing signal 713 and the array processing
signal 707, and sequentially updates the filter coefficient according to the magnitude of the
correlation.
[0050]
03-05-2019
15
The array processing unit 708 is known as a multi-input canceller for constructing a generalized
sidelobe canceller (Griffith's gym beamformer). Non-patent documents 1 and 2 disclose in detail
the configuration and operation of the array processing unit 708, respectively.
[0051]
As described above, according to the present embodiment, array processing is performed by
using the decorrelation signal 212 as a reference signal to erase signal components correlated
with the decorrelation signal 212, thereby suppressing the target signal and emphasizing the
other signals. Signal 707 is generated. Further, the array processing signal 707 is used as a
reference signal to cancel signal components having a correlation with the array processing
signal 707 included in the array processing signal 210, thereby emphasizing the target signal
and suppressing other signals. Output as 713. For this reason, it is possible to obtain a high
quality signal in which components other than the target signal are further suppressed as
compared with the first embodiment. That is, the wide band signal can be suppressed and the
target signal can be sufficiently emphasized without an increase in array size or an increase in
the number of sensors.
[0052]
<< Relationship with Generalized Side Lobe Canceller >> As disclosed in Non-Patent Documents 1
and 2, the array processing unit 203, the array processing unit 706, and the array processing
unit 708 each have a fixed beam former, a blocking matrix, and multiple inputs. It is known as a
canceller and constitutes a generalized sidelobe canceller (Griffith's Jim beamformer). The
configuration examples and operations of the array processing unit 203, the array processing
unit 706, and the array processing unit 708 are disclosed in detail in Non-Patent Documents 1
and 2.
[0053]
The array processing unit 203 uses the signals obtained by the respective sensors of the sensor
array 201 to emphasize the target signal. However, in the output signal of the array processing
unit 203, components other than the target signal are not sufficiently suppressed particularly in
the low band. This is because, as described above, the sensor interval in the sensor array 201 is
03-05-2019
16
not wide enough for the wavelength of the low band, and the directivity formed by the array
processing unit 203 is insufficient particularly in the low band.
[0054]
In the generalized side lobe canceller, the array processing unit 706 emphasizes the components
other than the target signal using the output of the array processing unit 203 in which the target
signal is emphasized (others are suppressed) as a reference signal (target signal Generates a
second array processing signal. Since components other than the target signal are not sufficiently
suppressed at the output of the array processing unit 203, components other than the target
signal are sufficient at the output of the array processing unit 706 operating with the output
signal of the array processing unit 203 as a reference signal. (The target signal is not sufficiently
suppressed). Therefore, the array processing unit 708 which erases the signal component
correlated with the output of the array processing unit 706 from the output of the array
processing unit 203 can not sufficiently erase components other than the target signal from the
output of the array processing unit 203, Components other than the target signal, particularly
low frequency components, remain in the output signal 213.
[0055]
In the present embodiment shown in FIG. 7A, the array processing unit 706 suppresses the
target signal using the output signal 212 of the correlation removal unit 204 as a reference
signal instead of the sensor processing signal 210 of the array processing unit 203. The
correlation removal unit 204 uses the input signal 211 as a reference signal to erase signals
correlated with the input signal 211, that is, signal components other than the target signal, and
does not perform array processing. Thus, components other than the target signal contained in
the decorrelation signal 212 are minimized, independently of the relationship between the sensor
spacing of the array and the frequency of the signal being processed.
[0056]
Thus, since the decorrelation signal 212 in which components other than the target signal are
sufficiently suppressed as compared to the array processing signal 210 is used as a reference
signal, the array processing unit 706 performs array processing in which the target signal is
sufficiently suppressed. Signal 707 can be generated.
03-05-2019
17
[0057]
Therefore, the array processing unit 708 which erases the signal component correlated with the
output of the array processing unit 706 from the output of the array processing unit 203 can
sufficiently erase the components other than the target signal in the output of the array
processing unit 203. As a result, no components other than the target signal, particularly low
frequency components, remain in the output signal 213.
[0058]
FIG. 7B is a diagram for describing a hardware configuration when the signal processing device
700 according to the present embodiment is realized using software.
[0059]
The signal processing device 700 includes a processor 710, a read only memory (ROM) 720, a
random access memory (RAM) 740, a storage 750, an input / output interface 760, an operation
unit 761, an input unit 762, and an output unit 763.
The processor 710 is a central processing unit and controls the entire signal processing
apparatus 700 by executing various programs.
[0060]
The ROM 720 stores various parameters and the like in addition to the boot program that the
processor 710 should execute first.
The RAM 740 stores an area for storing an input signal 205, an auxiliary signal 211, an array
processing signal 210, a correlation removal signal 212, an array processing signal 707, an array
processing signal 713 (output signal), etc. Have.
[0061]
The storage 750 also stores a signal processing program 751.
03-05-2019
18
The signal processing program 751 includes an array processing module 751a, a decorrelation
module 751b, an array processing module 751c, and an array processing module 751d. The
processor 710 executes each module included in the signal processing program 751 to realize
the functions of the array processing unit 203, the correlation removal unit 204, the array
processing unit 706, and the array processing unit 708 in FIG. 7A.
[0062]
An array processing signal 713, which is an output related to the signal processing program 751
executed by the processor 710, is output from the output unit 763 via the input / output
interface 760. As a result, for example, it is possible to suppress noise, interference signal, and
the like included in the input signal 205 input from the input unit 762 and to emphasize a target
signal such as voice.
[0063]
FIG. 7C is a flowchart for explaining the flow of processing of emphasizing a target signal such as
speech mixed with noise or interference by the signal processing program 751. In step S771, the
plurality of input signals 205 from the sensor 201 are supplied to the array processing unit 203.
In step S773, the array processing unit 203 executes an enhancement process of voice, that is, a
target signal as a first array process, and generates an array process signal 210.
[0064]
Next, in step S775, the auxiliary signal 211 is input from the sensor 202 and the processing to
be supplied to the correlation removal unit 204 is executed. In step S 777, the decorrelation unit
204 cancels the component correlated with the auxiliary signal 211 included in the array
processing signal 210 using the auxiliary signal 211 as a reference signal, and generates a
decorrelation signal 212. In step S 779, as a second array process, the array processing unit 706
erases the voice included in the input signal 205, that is, the target signal component, using the
decorrelation signal 212 as a reference signal, and generates an array process signal 707. . In
step S 781, with reference to the array processing signal 707 which is the enhanced interference
signal, the interference signal component included in the array processing signal 210 is
03-05-2019
19
eliminated to generate an audio, ie, an array processing signal 713 in which the target signal is
enhanced. .
[0065]
Finally, in step S783, the array processing signal 713 is output as a target signal, that is, a signal
in which speech is emphasized and the others are suppressed.
[0066]
FIG. 7C shows a flowchart for explaining the flow of processing when the signal processing
device 700 according to the present embodiment is realized by software.
However, the second to fourth embodiments and the sixth to eighth embodiments can be
similarly realized by appropriately omitting and adding the differences in the respective block
diagrams.
[0067]
With the above configuration, according to the present embodiment, it is possible to obtain an
output signal of higher quality than the generalized side lobe canceller. Therefore, it is possible to
suppress the broadband signal and sufficiently emphasize the target signal without increasing
the array size or the number of sensors.
[0068]
Sixth Embodiment Next, a signal processing apparatus according to a sixth embodiment of the
present invention will be described with reference to FIG. FIG. 10 is a block diagram showing the
configuration of a signal processing apparatus 1000 according to this embodiment.
[0069]
03-05-2019
20
The signal processing apparatus 1000 according to the present embodiment differs from the
array processing signal 210 which is the output of the array processing unit 203 in comparison
with the fifth embodiment, with the correlation removal signal 212 which is the output of the
correlation removal unit 204, It differs in that it is supplied to the array processing unit 708. The
other configurations and operations are similar to those of the third embodiment, and therefore
the same configurations and operations are denoted by the same reference numerals and the
detailed description thereof will be omitted.
[0070]
As described above, according to the present embodiment, the signal supplied to the array
processing unit 708 is a decorrelation signal 212 that has a higher interference signal
suppression effect than the array processing signal 210, that is, a high target signal enhancement
effect. Because of this, it is possible to obtain a signal in which the target signal is further
emphasized at the output of the array processing unit 708. For this reason, a high quality output
signal can be obtained as compared with the fifth embodiment. That is, the wide band signal can
be suppressed and the target signal can be sufficiently emphasized without an increase in array
size or an increase in the number of sensors.
[0071]
Seventh Embodiment Next, a signal processing apparatus according to a seventh embodiment of
the present invention will be described with reference to FIG. FIG. 11 is a block diagram showing
the configuration of a signal processing apparatus 1100 according to this embodiment.
[0072]
The signal processing apparatus 1100 according to this embodiment is different from the fourth
embodiment in that an array processing unit 706 and an array processing unit 708 are added,
and the other configuration and operation are the same as those in the second embodiment. It is.
[0073]
Moreover, compared with the said 5th Embodiment, it differs in the point to which the mixing
part 501 is added, and the other structure and operation | movement are the same as that of 2nd
Embodiment.
03-05-2019
21
[0074]
Therefore, except for the array processing unit 706, the array processing unit 708, and the
mixing unit 501, the same configuration and operation are denoted by the same reference
numerals, and the detailed description thereof is omitted.
[0075]
Further, the configuration and operation of the array processing unit 706 and the array
processing unit 708 have already been described in the fifth embodiment and the mixing unit
501 in the fourth embodiment, and thus the detailed description thereof will be omitted.
[0076]
As described above, according to the present embodiment, the addition of the array processing
unit 706 and the array processing unit 708 makes it possible to obtain an output of higher
quality than the fourth embodiment, and the addition of the mixing unit 501 leads to the fifth
embodiment. You can also get high quality output.
That is, since higher quality output can be obtained than in either of the fourth embodiment and
the fifth embodiment, the wide band signal is suppressed and the target signal is sufficiently
emphasized without increasing the array size or the number of sensors. can do.
[0077]
Eighth Embodiment Next, a signal processing apparatus according to an eighth embodiment of
the present invention will be described with reference to FIG.
FIG. 12 is a block diagram showing a functional configuration of the signal processing device
1200 according to the present embodiment.
[0078]
03-05-2019
22
The signal processing apparatus 1200 according to the present embodiment differs from the
array processing signal 210 which is an output of the array processing unit 203 in the signal
processing apparatus 1200 according to the seventh embodiment, in that the correlation removal
signal 212 which is an output of the correlation removal unit 204 It differs in that it is supplied
to the array processing unit 708.
The other configurations and operations are the same as those of the fifth embodiment, and
therefore, the same configurations and operations are denoted by the same reference numerals
and the detailed description thereof will be omitted.
[0079]
As described above, according to the present embodiment, the signal supplied to the array
processing unit 708 is a decorrelation signal 212 that has a higher interference signal
suppression effect than the array processing signal 210, that is, a high target signal enhancement
effect. As a result, at the output of the array processing unit 708, a signal in which the target
signal is further emphasized can be obtained.
Therefore, it is possible to obtain an output signal of high quality as compared with the seventh
embodiment. That is, the wide band signal can be suppressed and the target signal can be
sufficiently emphasized without an increase in array size or an increase in the number of sensors.
[0080]
[Ninth Embodiment] As an application example of the present invention, it is conceivable to
perform video chat and remote communication via a network using a tablet PC placed on a desk.
A top view of such an application from above is shown in FIG.
[0081]
A sensor array 201 including four sensors implemented by microphones is disposed on the
upper surface of the tablet PC 1301, and a sensor 202 is disposed on the lower surface of the
03-05-2019
23
back surface. The sensor 202 may be disposed on the upper surface or the side surface. By
processing the acoustic signals acquired by these microphones in any of the first to eighth
embodiments, the voice of the user 1302 seated on the sofa is emphasized, and the voice of the
person 1303 behind that and the user's front back Music signals generated from certain left and
right speakers 1304 can be suppressed. For this reason, only the voice of the user is obtained as
an output, and by using this output for speech and speech recognition, a comfortable speech and
a high speech recognition rate can be realized.
[0082]
Further, as shown in FIG. 14, it is also conceivable to perform video chat and remote
communication via a network by using a television receiver 1401 placed at a position away from
the user. FIG. 14 shows a top view of such an application from above.
[0083]
A sensor array 201 including four sensors realized by microphones is disposed on the upper
surface of the television receiver 1401, and a sensor 202 is disposed on the lower surface of the
back surface. The sensor 202 may be disposed on the upper surface or the side surface. By
processing the acoustic signal acquired by these microphones in any of the first to eighth
embodiments, the voice of the user 1402 seated on the sofa is emphasized, and the voice of the
person 1404 diagonally in front of the television receiver 1401 And the music signal generated
from the left and right speakers 1403 on the television receiver side can be suppressed.
Therefore, only the voice of the user 1402 is obtained as an output, and by using this output for
a call or voice recognition, a comfortable call or a high voice recognition rate can be realized. In
particular, by controlling the television receiver 1401 with this voice recognition function, the
user 1402 can change the channel and volume of the television receiver 1401 using voice.
[0084]
[Other Embodiments] Although the present invention has been described with reference to the
embodiments, the present invention is not limited to the above embodiments. The configurations
and details of the present invention can be modified in various ways that can be understood by
those skilled in the art within the scope of the present invention. Also included within the scope
of the present invention are systems or devices that combine the different features included in
03-05-2019
24
each embodiment.
[0085]
Furthermore, the present invention may be applied to a system configured of a plurality of
devices or to a single device. Furthermore, the present invention is also applicable to the case
where a signal processing program for realizing the functions of the embodiments is supplied to
a system or apparatus directly or remotely. Therefore, in order to realize the functions of the
present invention on a computer, a program installed on the computer, a medium storing the
program, and a WWW (World Wide Web) server for downloading the program are also included
in the scope of the present invention. .
[0086]
[Other Representations of the Embodiment] Some or all of the above-described embodiments
may be described as in the following appendices, but are not limited thereto. (Supplementary
Note 1) A first array processing unit that generates a first array processing signal by partially
emphasizing a predetermined signal with respect to signals received from a plurality of sensors,
and an auxiliary other than the plurality of sensors A signal processing apparatus, comprising: a
decorrelation unit configured to delete a signal component correlated with a signal received from
a sensor from the first array processing signal to generate a decorrelation signal. (Supplementary
note 2) The signal processing device according to Supplementary note 1, further comprising: a
mixing unit that mixes the correlation removal signal with the first array processing signal to
generate a mixed signal. (Supplementary Note 3) The mixing unit includes a low pass filter that
passes low pass components of the decorrelation signal, a high pass filter that passes high pass
components of the first array processing signal, and the low pass. The signal processing
apparatus according to claim 2, further comprising: an adder for adding an output of the pass
filter and an output of the high pass filter. (Supplementary Note 4) A second array processing
unit that attenuates the predetermined signal based on the signals received from the plurality of
sensors and the correlation removal signal to generate a second array processing signal, and the
second array The signal processing apparatus according to any one of appendices 1 to 3, further
comprising: a third array processing unit configured to cancel a signal component having a
correlation with a processing signal from the first array processing signal. (Supplementary Note
5) A second array processing unit that attenuates the predetermined signal to generate a second
array processing signal based on the signals received from the plurality of sensors and the
correlation removal signal, and the second array The signal processing apparatus according to
any one of appendices 1 to 3, further comprising: a third array processing unit configured to
03-05-2019
25
cancel a signal component having a correlation with a processing signal from the decorrelation
signal. (Supplementary Note 6) The correlation removal unit includes an adaptive filter that
processes a signal received from the auxiliary sensor, and a subtractor that subtracts the output
of the adaptive filter from the first array processing signal to generate a correlation removal
signal. The signal processing apparatus according to any one of appendices 1 to 5, further
comprising: updating the coefficient of the adaptive filter using the signal received from the
auxiliary sensor and the output of the subtractor. (Supplementary Note 7) A step of partially
emphasizing a predetermined signal with respect to signals received from a plurality of sensors
to generate an array processing signal, and correlation with a signal received from an auxiliary
sensor other than the plurality of sensors Removing certain signal components from the array
processing signal to generate a decorrelation signal.
(Supplementary Note 8) A step of partially emphasizing a predetermined signal with respect to
signals received from a plurality of sensors to generate an array processing signal, and
correlation with a signal received from an auxiliary sensor other than the plurality of sensors And
C. removing the signal component from the array processing signal to generate a decorrelation
signal. (Supplementary Note 9) A plurality of sensors disposed on the front surface, an auxiliary
sensor disposed at a position different in acoustic characteristics from the plurality of sensors,
and a predetermined signal partially for signals received from the plurality of sensors And an
array processing unit for generating an array processing signal, and a decorrelation unit for
removing a signal component correlated with the signal received from the auxiliary sensor from
the array processing signal to generate a decorrelation signal. A media device characterized by
[0087]
This application claims priority based on Japanese Patent Application No. 2013-209731 filed on
October 4, 2013, the entire disclosure of which is incorporated herein.
03-05-2019
26
Документ
Категория
Без категории
Просмотров
0
Размер файла
41 Кб
Теги
jpwo2015049921
1/--страниц
Пожаловаться на содержимое документа