close

Вход

Забыли?

вход по аккаунту

?

JP2006173916

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 JP2006173916
PROBLEM TO BE SOLVED: To appropriately suppress a noise signal among respective signals
outputted by respective microphones. SOLUTION: A component calculation unit 41 is a signal
including difference information indicating differences between signals including sound source
signals output by a plurality of microphones, and a sound source signal output by one of the
plurality of microphones. The sound source component S having a correlation with the sound
source signal and the noise component N having a correlation with the sound source signal and
other signals are calculated based on the signal information indicating. Then, based on the ratio
between the sound source component S and the noise component N calculated by the component
calculation unit 41, the signal suppression unit 42 selects the signal of the frequency band in
which the noise component exists from the signals output by the plurality of microphones.
Suppress. [Selected figure] Figure 1
Suppression device, sound collecting device and program
[0001]
The present invention relates to a suppression device, a sound collection device, and a program
for suppressing a signal related to noise contained in signals output by a plurality of
microphones.
[0002]
BACKGROUND ART Conventionally, a suppression device has been provided that suppresses
noise signals included in signals output by a plurality of microphones (see, for example, Patent
04-05-2019
1
Document 1).
For example, in the case where there is difference information (for example, a wind noise signal
generated by a wind striking a microphone) indicating the difference between the signals output
by the first microphone and the second microphone, the suppression device selects the first
microphone and Among the signals output by the second microphone, a signal of a specific
frequency band (for example, a noise signal included in a signal of a frequency band in which a
noise component is present) has been cut. Unexamined-Japanese-Patent No. 5-7292
[0003]
However, when the difference information exists, the suppression device uniquely cuts off a
signal (such as a noise signal) in a specific frequency band among the signals output by the first
and second microphones. Therefore, the source signal close to the noise signal is also cut
uniquely. Therefore, the suppression device may output an unnatural signal obtained by cutting
the sound source signal included in the signal of the specific frequency band. This is because the
larger sound source signal is cut as the sound source signal included in the signal of the specific
frequency band is larger, so a more unnatural signal is output.
[0004]
Specifically, when the sound source is close to the first microphone, the sound wave from the
sound source reaches the second microphone after a predetermined time delay after reaching the
first microphone. In this case, regardless of whether or not the noise signal is included in each
signal output from each microphone, a difference exists in each signal output from each
microphone. Therefore, when the difference exists, or when the difference exceeds a
predetermined threshold, the suppression device uniquely cuts off a signal (such as a noise
signal) in a specific frequency band of each signal output by each microphone, The suppression
device also cuts unambiguously the sound source signal close to the noise signal, and can not
output an appropriate signal.
[0005]
Therefore, the present invention has been made in view of the above points, and provides a
suppression device, a sound collection device, and a program capable of appropriately
suppressing a noise signal of each signal output by each microphone. With the goal.
04-05-2019
2
[0006]
In order to solve the above problems, the present invention includes difference information
indicating differences between signals including sound source signals output by a plurality of
microphones and a sound source signal output by one of the plurality of microphones.
Component calculation means for calculating a sound source component correlated with the
sound source signal and noise components correlated with the sound source signal and the other
signals based on the signal information indicating the signal, and the sound source calculated by
the component calculation means It is characterized by including signal suppression means for
suppressing a signal of a frequency band in which a noise component is present from signals
outputted by a plurality of microphones based on a ratio of a component and a noise component.
[0007]
According to the present invention as described above, the suppression device is a signal of the
frequency band in which the noise component is present from the signals output by the plurality
of microphones based on the ratio between the sound source component and the noise
component as well as the noise component. By suppressing (noise signal + sound source signal
contained in noise component), the suppression apparatus can prevent the sound source signal in
the frequency band in which the noise component is present from being unambiguously cut.
That is, even if the suppression device calculates the noise component, it does not suppress the
signal (noise signal + sound source signal) in the frequency band in which the noise component is
present unless the ratio between the noise component and the sound source component is not a
specific ratio. Therefore, the suppression apparatus can prevent the sound source signal of the
frequency band in which the noise component is present from being uniquely cut, and can output
a more appropriate signal.
[0008]
In the above invention, the first signal including the sound source signal output by the first
microphone of the plurality of microphones and the second signal including the sound source
signal output by the second microphone of the plurality of microphones are added An addition
unit may be provided to calculate the signal information as signal information.
04-05-2019
3
In this case, the suppression device does not use the signal output from one microphone as the
signal information, but uses the addition result of each signal output from a plurality of
microphones as the signal information. More appropriate noise components and sound source
components can be calculated.
[0009]
In the above invention, the first signal including the sound source signal output by the first
microphone of the plurality of microphones, and the second signal output by the second
microphone of the plurality of microphones and delayed from the first signal A first difference
calculation means for calculating first difference information indicating a difference from the
second signal, a second signal output from the second microphone, and a signal output from the
first microphone and delayed from the second signal Second difference calculation means for
calculating second difference information indicating a difference with the first signal, and the
first difference information calculated by the first difference calculation means and the second
difference information calculated by the second difference calculation means And the component
calculating means is configured to generate a sound source having a correlation with the sound
source signal based on the average value calculated by the average calculating means and the
signal information. When, may be calculated and a noise component correlated with the
excitation signal and the other signals. In this case, the suppression apparatus calculates the
average value of the first difference information and the second difference information so that
the suppression apparatus converts the directivity characteristics of the plurality of microphones
into the directivity characteristics of non-direction. It is possible to calculate more appropriate
noise components and sound source components based on the average value.
[0010]
According to the present invention, it is possible to appropriately suppress the noise signal of
each signal output by each microphone.
[0011]
First Embodiment A suppression apparatus according to the present embodiment will be
described with reference to the drawings.
04-05-2019
4
FIG. 1 is a view showing a suppression device 1 in the present embodiment. As shown in FIG. 1,
the suppression apparatus 1 includes a directivity control unit 10, LPFs 20a and 20b, root mean
square calculation units 30a and 30b, and a suppression unit 40. The signal output by the
microphone M1 (first microphone) is X1, and the signal output by the microphone M2 (second
microphone) is X2.
[0012]
The directivity control unit 10 calculates, as difference information, a difference between the
signal X1 including the sound source signal output by the microphone M1 and the signal X2
including the sound source signal output by the microphone M2.
[0013]
The LPFs 20a and 20b remove the high frequency band signal among the signals included in the
difference information {(X1−X2) or (X2−X1)} output by the directivity control unit 10.
The LPFs 20a and 20b may not be provided.
[0014]
Here, although the noise signal and the source signal are included in the low frequency band
signal, the signal in the high frequency band always includes more source signals than the noise
signal. Therefore, the magnitude relationship between the noise signal and the sound source
signal largely changes not in the high frequency band signal but in the low frequency band
signal. Therefore, since the magnitude relationship between the noise signal and the sound
source signal is clarified by the LPFs 20a and 20b removing signals in the high frequency band,
the suppressing unit 40 described later can calculate the SNR more accurately.
[0015]
The root mean square calculation units 30a and 30b calculate a time average of squares of the
signals {LPF (X1) -LPF (X2) or LPF (X2) -LPF (X1)} output from the LPFs 20a and 20b. Although
04-05-2019
5
Equations (1) to (14) in the present embodiment are established as that for which the LPF is not
present for the sake of convenience of calculation, the same applies even if the LPF is present.
[0016]
The suppression unit 40 includes a component calculation unit 41 and a signal suppression unit
42. The component calculating unit 41 generates difference information (X1-X2) indicating the
difference between the signals X1 and X2 output by the microphone M1 and the microphone
M2, and the sound source signal output by one microphone (in the present embodiment, the
microphone M2). Component calculation means for calculating a sound source component S
correlated with the sound source signal and a noise component N correlated with the sound
source signal and the other signals based on the signal information (X2) indicating a signal
including . The component calculating unit 41 in the present embodiment calculates the time
average of squared {(X1-X2) ^ 2} of the difference information output from the mean square
calculating unit 30a and the signal output from the mean square calculating unit 30b. The sound
source component S and the noise component N are calculated based on the time average of the
square {X2 ^ 2} of the signal information to be shown. The specific calculation method of the
sound source component S and the noise component N is as follows.
[0017]
First, the signal X1 output by the microphone M1 is calculated by the following equation 1, and
the signal X2 output by the microphone M2 is calculated by the following equation 2. C1i
indicates the transfer coefficient of the sound transmitted from the i-th sound source to the
microphone M1. C2i represents the transfer coefficient of the sound transmitted from the ith
sound source to the microphone M2. N1 indicates a noise signal (for example, a signal related to
wind noise, and so on) output by the microphone M1. N2 represents a noise signal output by the
microphone M2. Si represents the i-th sound source signal output by the sound source. Also
assume that the sound source is far enough away.
[0018]
Although Equations 1 to 14 described later are satisfied on the premise that no LPF is provided
to facilitate calculation, the present invention is not limited to this, and even if an LPF is provided,
the same applies as described later. It can be established by the calculation method.
04-05-2019
6
[0019]
X1 = .SIGMA.C1i.times.Si + N1 Formula 1 X2 = .SIGMA.C2i.times.Si + N2 Formula 2 The
directivity control unit 10 calculates X1-X2 as difference information according to the formula 1
and the formula 2 above.
Further, the root mean square calculation unit 30a calculates a time average of squares of
difference information (X1-X2) indicating differences between the signals X1 and X2 outputted
by the microphone M1 and the microphone M2 as A (the following equation 3).
[0020]
<img class = "EMIRef" id = "200304102-00003" /> Further, the root mean square calculation
unit 30b is a signal information indicating a signal including the sound source signal output by
one microphone (in the present embodiment, the microphone M2). The time average of the
square of (X2) is calculated as B (see Formula 4 below).
[0021]
<img class = "EMIRef" id = "200304102-00004" /> According to Equation 3 and Equation 4, the
component calculation unit 41 determines the sound source component (sum of each sound
source signal) S related to the sound source signal, the sound source signal, and others The noise
component (sum of each noise signal) relating to the signal D.sub.N and the ratio SNR of the
sound source component S and the noise component N are calculated (see Equations 5, 6, and 7
below).
[0022]
S = (A-B) / (α-β ') ... Formula 5 N = (αB-β'A) / (α-β') ... Formula 6 α {α / (α-1) B}-{1 / (Α-1)
A} (where β ′ ≒ 1) SNR = S / N = (A−B) / (αB−β′A) ≒ (A−B) / (αB−A). The signal
suppression unit 42 generates noise from the signal output by the microphone M1 and the
microphone M2 based on the ratio SNR of the sound source component S to the noise component
N calculated by the component calculation unit 41 (see the above equation 7). It is a signal
suppression means which suppresses the signal of the frequency band in which a component
exists.
04-05-2019
7
Note that the signal suppression unit 42 in the present embodiment extracts a signal of a
frequency band higher than a frequency band through which the signal passes through by the
LPF from the signals output by the microphone M1 and the microphone M2 based on the SNR.
[0023]
Specifically, when the relationship of SNR <predetermined threshold is established, the signal
suppression unit 42 determines from the signals output by the microphone M1 and the
microphone M2 a signal in the frequency band in which the noise component is present (noise
signal And the source signal (which is a frequency close to that signal).
On the other hand, when the relationship of SNR <predetermined threshold does not hold, the
signal suppression unit 42 determines from the signals output by the microphone M1 and the
microphone M2 a signal in the frequency band in which the noise component is present (noise
signal and its signal Not suppress the sound source signal which is a frequency close to
[0024]
According to the invention in this embodiment, the suppression device 1 has noise components
from signals output by a plurality of microphones based on not only the noise component N but
also the ratio SNR of the noise component N and the sound source component S. By suppressing
the signals in the frequency band (noise signals and sound source signals having frequencies
close to the signals), the suppression apparatus 1 can prevent the sound source signals in a
specific frequency band from being uniquely cut. Specifically, when the relationship of SNR (S /
N) <predetermined threshold value is established, the noise component N is very large compared
to the sound source component S. In this case, since the noise signal is included more than the
source signal in the frequency band where the noise component is present, the suppression
device 1 is a signal of the frequency band where the noise component is present (a noise signal
and a frequency close to the signal By cutting the sound source signal), it is possible to cut the
noise signal in the frequency band in which the noise component exists while preventing the
influence of the cutting of the sound source signal from being increased.
[0025]
04-05-2019
8
On the other hand, when the relationship of SNR (S / N) <predetermined threshold does not hold,
the sound source component S is larger than the noise component N. In this case, the sound
source signal is included more than the noise signal in the frequency band in which the noise
component is present, so all signals in the frequency band in which the noise component is
present (the sound source signal and the frequency close to the signal) When the noise signal is
removed, the influence of the removal of the sound source signal is significant. As a result, the
suppression device 1 does not cut the signal in the frequency band in which the noise component
is present (the noise signal and the sound source signal having a frequency close to that signal),
thereby a large sound source signal included in the frequency band in which the noise
component is present. Can not be cut uniquely.
[0026]
In the present embodiment, B is a time average of squares of signal information (X2) indicating a
signal including the sound source signal output by the microphone M2, but the present invention
is not limited to this, and B is output by the microphone M1. A time average of squares of signal
information (X1) indicating a signal including the sound source signal, or a time average of
squares of signal information (X1 + X2) indicating a signal including the sound source signal
output by the microphones M1 and M2; It is also good.
[0027]
Note that the noise signal in the present embodiment may be a noise signal that is not correlated
between microphones (for example, a noise signal mixed from any one of the microphones).
[0028]
First Modification FIG. 2 is a view showing the suppression device 1 in the present modification.
Unlike the first embodiment shown in FIG. 1, in the present modification shown in FIG.
In addition, each part contained in the suppression apparatus 1 shown in FIG. 2 has the same
function as each part of the same name contained in the suppression apparatus 1 shown in FIG.
[0029]
04-05-2019
9
The addition unit 50 is an addition unit that calculates an addition (X1 + X2) of the signal X1
output by the microphone M1 and the signal X2 output by the microphone M2 as the signal
information. Here, in the first embodiment, B, which is a time average of the square of signal
information, is calculated by Equation 4, but in the present modification, B which is a time
average of the square of signal information is represented by Equation 8 below. It is calculated.
[0030]
<img class = "EMIRef" id = "200304102-000005" /> According to Equation 3 and Equation 8, the
component calculation unit 41 generates a sound source component S related to the sound
source signal, and a noise component N related to the sound source signal and the other signals.
The ratio SNR of the sound source component S and the noise component N is calculated (see
Equation 9, Equation 10, and Equation 11 below).
[0031]
S = (A−B) / (α−β) Formula 9 N = (αB−βA) / (α−β) ≒ α / (α−1) B−1 / (α−1) A
Formula 10 (Where β ≒ 1) SNR = S / N = (A−B) / (αB−βA) ≒ (A−B) / (αB−A) (11)
According to the invention in this modification, The suppression device 1 does not use the signal
output by the microphone M2 as the signal information (X1) but the addition result of the signals
output by the microphone M1 and the microphone M2 as the signal information (X1 + X2). 1 can
calculate more appropriate noise component N and sound source component S based on the
signal information (X1 + X2).
[0032]
In the present embodiment, regardless of whether or not the noise component N is present, the
suppression unit 40 suppresses the signal in the frequency band in which the noise component is
present by establishing the relationship of SNR <predetermined threshold. However, the present
invention is not limited to this, and the suppression unit 40 may suppress the signal in the
frequency band in which the noise component is present only when the sound source component
S or the noise component N is present based on the difference information or the signal
information. .
Further, the suppression unit 40 determines whether the sound source component S or the noise
component N is present based on the difference information or the signal information, and when
the sound source component S or the noise component N is present, the sound source
04-05-2019
10
component By establishing the relationship of S <predetermined threshold or noise component N
<predetermined threshold, the signal in the frequency band in which the noise component is
present may be suppressed.
[0033]
Note that the suppression device 1 may be included in the sound collection device.
[0034]
Second Embodiment FIG. 3 is a view showing a suppression device 1 in the present embodiment.
In the present embodiment shown in FIG. 3, unlike the first embodiment shown in FIG. 1, two
directivity control units 10a and 20b and an average calculation unit 43 are provided.
In addition, each part contained in the suppression apparatus 1 shown in FIG. 3 has the same
function as each part of the same name contained in the suppression apparatus 1 shown in FIG.
[0035]
The directivity control unit 10a is first difference information (X1-Dly (X2)) indicating the
difference between the signal X1 output from the microphone M1 and the signal X2 output from
the microphone M2 and delayed from the signal X1. Is a first difference calculation means for
calculating Dly denotes a time delay operator. Also, Dly (X2) is a delay of X2 for a fixed time.
[0036]
The directivity control unit 10b is second difference information (X2-Dly (X1)) indicating the
difference between the signal X2 output from the microphone M2 and the signal X1 output from
the microphone M1 and delayed from the signal X2. Is a second difference calculation means for
calculating Note that Dly (X1) is obtained by delaying X1 for a predetermined time.
04-05-2019
11
[0037]
The average calculating unit 43 calculates the first difference information (X1-Dly (X2))
calculated by the directivity control unit 10a and the second difference information (X2-Dly (X1))
calculated by the directivity control unit 10b. Means for calculating an average value of The
average calculating unit 43 in the present embodiment uses the time average of the squared
{(X1−Dly (X2)) ^ 2} of the first difference information output from the root mean square
calculating unit 30a and the square average calculating unit 30b. The average value with the
time average of the square of {(X2−Dly (X1)) ^ 2} of the output second difference information is
calculated. The specific calculation method of the sound source component S and the noise
component N is as follows.
[0038]
First, the signal X1 output by the microphone M1 is calculated by the above equation 1, and the
signal X2 output by the microphone M2 is calculated by the above equation 2. C1i represents the
transfer coefficient of the sound transmitted from the i-th sound source to the microphone M1,
as in the first embodiment. C2i represents the transfer coefficient of the sound transmitted from
the ith sound source to the microphone M2. N1 represents a noise signal output by the
microphone M1. N2 represents a noise signal output by the microphone M2. Si represents the ith sound source signal output by the microphone.
[0039]
Then, the root mean square calculation unit 30a calculates the time average of the square of the
first difference information {(X1−Dly (X2)) ^ 2} as A1 (the following formula 12). Similarly, the
root mean square calculation unit 30c calculates the time average of the square of the second
difference information {(X2−Dly (X1)) ^ 2} as A2 (Formula 13 below).
[0040]
<img class = "EMIRef" id = "200304102-000006" /> <img class = "EMIRef" id = "200304102000007" /> Thereafter, the average calculator 43 calculates A1 according to the above equation
12 and the above equation 13. And the average value of A2 is calculated (refer the following
formula 14).
04-05-2019
12
[0041]
<img class = "EMIRef" id = "200304102-00008" /> Further, the root mean square calculation
unit 30b generates signal information (X2) indicating a signal output by one microphone (in the
present embodiment, the microphone M2). The time average of the squares is calculated as B
(the above equation 4).
[0042]
The component calculation unit 41 calculates the sound source component S relating to the
sound source signal, the noise component N relating to the sound source signal and the other
signals, and the ratio SNR of the sound source component S and the noise component N
according to Equation 8 and Equation 14 above. (Refer to the said Formula 5, Formula 6, and
Formula 7).
[0043]
According to the invention in the present embodiment, the suppression device 1 calculates the
average value (see Equation 14), thereby making the directivity characteristic of the plurality of
microphones into the omnidirectional directivity characteristic. The noise component N and the
sound source component S can be calculated more appropriately based on the average value.
[0044]
Further, when the plurality of microphones have directivity, the sound source component S or
the noise component N changes according to the direction in which the sound source signal is
directed to the microphone (hereinafter referred to as the sound source direction). The SNR,
which is the ratio of, also changes according to the sound source direction.
For this reason, when the plurality of microphones have directivity, if the suppression unit 40
appropriately compares the SNR with the predetermined threshold, the suppression unit 40
should not change the predetermined threshold of SNR according to the sound source direction.
It does not.
According to the present invention, by calculating the averaged difference information between
the first difference information and the second difference information, the sound source signal
and the microphone signal in the state in which each microphone does not depend on the sound
04-05-2019
13
source direction (non-directionalized state) are calculated. Since this is equivalent to the noise
signal being collected, the suppression unit 40 does not have to change the predetermined
threshold of the SNR according to the direction of the sound source, and can more accurately
compare the SNR with the predetermined threshold, A source signal larger than the noise signal
included in the signal of the frequency band in which the noise component is present (the noise
signal and the source signal having a frequency close to the signal) can not be cut uniquely.
[0045]
In the present embodiment, the directivity control units are described as two of the directivity
control unit 10a and the directivity control unit 10b, but the invention is not limited thereto, and
may be three or more.
Thereby, in the present embodiment, the average value of A1 and A2 is calculated by using A1
and A2, but in addition to A1 and A2, A3, ..., An calculated by the same method are used. By
being used, their average value can be calculated.
[0046]
Further, although the LPF 20 b and the root mean square calculation unit 30 b are provided in
the present embodiment, the present invention is not limited thereto, and the LPF 20 b and the
root mean square calculation unit 30 b may not be provided.
[0047]
In the present embodiment, Dly (X1) and Dly (X2) may be X or may be different from each other.
Also, Dly (X1) and Dly (X2) may differ depending on the suppressor 1.
[0048]
Second Modified Example FIG. 4 is a view showing the suppressing device 1 in the present
modified example.
04-05-2019
14
Unlike the second embodiment shown in FIG. 3, in the present modification shown in FIG. 4, the
adding unit 50 is provided in the front stage of the LPF 20 b. In addition, each part contained in
the suppression apparatus 1 shown in FIG. 4 has the same function as each part of the same
name contained in the suppression apparatus 1 shown in FIGS. 1 and 2.
[0049]
The addition unit 50 is an addition unit that calculates an addition (X1 + X2) of the signal X1
output by the microphone M1 and the signal X2 output by the microphone M2 as the signal
information. Since this addition unit 50 has the same function as the addition unit 50 in the first
modification, the detailed description of the addition unit 50 will be omitted.
[0050]
According to the invention in this modification, the suppression device 1 does not use the signal
output by the microphone M2 as the signal information (X1), but adds the result of addition of
each signal output by the microphone M1 and the microphone M2 to the signal information In
order to set (X1 + X2), the suppression apparatus 1 can calculate more appropriate noise
component N and sound source component S based on the signal information (X1 + X2).
[0051]
In the present modification, the directivity control units are described as two directivity control
units 10a and 10b. However, the directivity control units may be three or more without being
limited thereto.
[0052]
Although the suppression device 1 and the plurality of microphones (the microphone M1 and the
microphone M2) are separately provided in each of the above-described embodiments and the
respective modifications, the present invention is not limited to this. Microphone may be
provided.
[0053]
04-05-2019
15
(Program) The contents described in the suppression apparatus 1 can also be realized by
executing a dedicated program for using a predetermined program language in a generalpurpose computer such as a personal computer.
The program may be recorded on a recording medium.
Examples of the recording medium include a hard disk, a flexible disk, a compact disk, an IC chip,
and a cassette tape.
[0054]
As mentioned above, although an example of the present invention was explained, only a specific
example was illustrated, and it does not limit the present invention in particular, and the concrete
composition of each means, etc. can be designed and changed suitably.
Also, the actions and effects described in the embodiment and the modifications merely list the
most preferable actions and effects arising from the present invention, and the actions and
effects according to the present invention are described in the embodiments and the
modifications. It is not limited to things.
[0055]
It is a figure which shows the suppression apparatus in 1st Embodiment. It is a figure which
shows the suppression apparatus in a 1st modification. It is a figure which shows the suppression
apparatus in 2nd Embodiment. It is a figure which shows the suppression apparatus in a 2nd
modification.
Explanation of sign
[0056]
DESCRIPTION OF SYMBOLS 1 ... suppression apparatus, 10 ... directivity control part, 20 ... LPF,
30 ... root mean square calculation part, 40 ... suppression part, 41 ... component calculation part,
04-05-2019
16
42 ... signal suppression part, 43 ... average calculation part, 50 ... addition Department
04-05-2019
17
Документ
Категория
Без категории
Просмотров
0
Размер файла
27 Кб
Теги
jp2006173916
1/--страниц
Пожаловаться на содержимое документа