close

Вход

Забыли?

вход по аккаунту

?

JP2011254242

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 JP2011254242
The present invention provides a technology for suppressing the sound emitted from the speaker
of another handsfree device. SOLUTION: A speaker 2, four microphones 31, 32, 33, 34 arranged
at intervals of 90 degrees around a circumference of the speaker 2 and four microphones 31, 32,
33, 34 A phase shifter 5 for shifting the phase of the collected sound signal by 90 degrees, and a
control unit 4 for reducing or removing at least one of the sound signals collected by the four
microphones 31, 32, 33, 34 , A plurality of handsfree devices including an adding unit 6 that
adds the phase-shifted and at least one reduced or eliminated sound signal, and the control unit
of each handsfree device The sound signal collected by the microphone facing the speaker of the
free device is reduced or eliminated. [Selected figure] Figure 1
Sound pickup reproduction apparatus, method and program, hands free apparatus
[0001]
The present invention relates to a device for acquiring sounds such as voices and musical tones
in a hands-free manner such as a voice conference, and in particular, to prevent generation of
echoes and howlings due to surround sound from a speaker to a microphone. The present
invention relates to a technology for suppressing sound coming from a direction.
[0002]
The hands-free device described in Patent Document 1 includes one speaker and N microphones
arranged at equal intervals on a circumference centered on the speaker, and the sound is
collected by N microphones. The phase of the sound signal is shifted by 2π / N [rad].
04-05-2019
1
By shifting the phase in this manner, it is possible to pick up sounds emitted from other points
while suppressing the sound emitted from the speaker.
[0003]
Patent No. 3740313
[0004]
However, although the handsfree device described in Patent Document 1 can suppress the sound
emitted from its own speaker, it has a problem that it can not suppress the sound emitted from
other points.
For example, in the case where a plurality of handsfree devices described in Patent Document 1
are arranged in the same space in order to use in a large number of meetings or in a wide space,
suppressing the sound emitted from the speakers of other handsfree devices There was a
problem of being unable to
[0005]
In order to solve the above problems, the phases of the sound signals collected by each of the
four microphones arranged at intervals of 90 degrees on the circumference around the speaker
are shifted by 90 degrees, Among the microphones, the sound signals collected by the
microphones directed to the other speakers than the above-mentioned speakers are reduced or
eliminated, and the phase-shifted and at least one reduced or eliminated sound signals are added.
[0006]
It was emitted by other speakers not only by suppressing the sound emitted by its own speaker,
but also by reducing or eliminating the sound signal collected by the microphone directed to the
other speakers other than its own speaker You can suppress the sound.
[0007]
04-05-2019
2
FIG. 2 is a functional block diagram of an example of the handsfree device according to the first
embodiment.
The figure for demonstrating how to attach a weight.
The figure for demonstrating how to attach a weight. The figure for demonstrating how to attach
a weight. FIG. 3 is a functional block diagram of an example of a sound collection and
reproduction device configured by connecting hands-free devices in series. The functional block
diagram of the example of a phase shifter. The functional block diagram of the example of a
control part. The functional block diagram of the other example of a control part. The functional
block diagram of the other example of a sound collection reproducing apparatus. FIG. 8 is a
functional block diagram of an example of a sound collection and reproduction apparatus
according to a second embodiment. The functional block diagram of the example of a specific
direction suppression gain calculation part. The functional block diagram of the example of a
beam former part. The functional block diagram of the example of a signal amount estimation
part. The figure for demonstrating the directivity characteristic of a beam former part. The
functional block diagram of the example of a gain coefficient calculation part. The figure which
shows the experimental result by a prior art. FIG. 6 is a diagram showing experimental results by
the sound collection reproduction method of the first embodiment. The flowchart which shows
the example of the sound pickup reproduction method. The other functional block diagram of the
example of the handsfree apparatus of 1st embodiment. The figure for demonstrating how to
attach a weight.
[0008]
First Embodiment As shown in FIG. 1, the handsfree device according to the first embodiment
collects a speaker 2 for reproducing sound based on a sound signal sent from the network 1, and
picks up surrounding sound. Four microphones 31, 32, 33, 34, a control unit 4 for reducing or
removing at least one of the collected sound signals, a phase shifter 5 for shifting the phases of
the collected sound signals, and a phase For example, an adder 6 is provided which adds the at
least one reduced or reduced sound signal. The sound collection and reproduction apparatus is
one in which at least two hands free apparatuses are connected in series.
[0009]
04-05-2019
3
The microphones 31, 32, 33, 34 are arranged at equal intervals, that is, at intervals of 90
degrees, on the circumference centered on the speaker 2. In this example, the microphones 31,
32, 33, 34 are directed in the outer peripheral direction of a circle centered on the speaker 2.
The sound signals picked up by the microphones 31, 32, 33, 34 are sent to the control unit 4
(step S1, FIG. 18).
[0010]
The control unit 4 reduces or removes the sound signal collected by the microphones facing the
other speakers than the speaker 2 among the microphones 31, 32, 33, 34 (step S2). Here, the
other speaker means, for example, a speaker of another handsfree device installed in the same
space. In addition, that the microphone is facing the speaker means that the directivity of the
microphone is the same as that of the other speakers, and the directivity of the microphone is the
directivity of any microphone other than the microphone. It means either that it is closer to the
direction of the other speaker, or that the direction of directivity of the two microphones
including the microphone is closer to the direction of the other speaker than the direction of the
directivity of the other two microphones. .
[0011]
In this example, the control unit 4 is configured by weight adders 41, 42, 43, and 44 disposed
between the microphones 31, 32, 33, 34 and the addition unit 6, respectively. Weights w1, w2,
w3 and w4 are set to the weight adders 41, 42, 43 and 44, respectively. The weights w1, w2, w3
and w4 are set so that the absolute value of the weight of the microphone whose directionality of
directivity is the direction of the other speakers than the speaker 2 is smaller than the absolute
value of the weights of the other microphones. . The weight adders 41, 42, 43 and 44 multiply
the set weights w1, w2, w3 and w4 by the input sound signal.
[0012]
In FIGS. 2 to 4, it is assumed that a region indicated by dots is a region to collect sound to be
emitted, and a region without dots is a region to suppress sound to be emitted. As shown in FIG.
2, when it is desired to suppress the sound emitted from the direction of the microphone 34, the
absolute value of the weight w4 corresponding to the microphone 34 is set to the weights w1,
04-05-2019
4
w2 and w3 corresponding to the other microphones 31, 32 and 33. Make it smaller than the
absolute value. As shown in FIG. 3, when it is desired to suppress the sound emitted from the
direction of the microphones 32, 33 and 34, the absolute values of the weights w2, w3 and w4
respectively corresponding to the microphones 32, Make it smaller than the absolute value of the
corresponding weight w1. As shown in FIG. 4, when it is desired to suppress the sound emitted
from the direction of the microphones 32, 34, the absolute values of the weights w2, w4
respectively corresponding to the microphones 32, 34 are set to the weights w1 corresponding
to the other microphones 31, 33. , W3 smaller than the absolute value. As shown in FIG. 20,
when the other speaker S is in the middle of the direction of directivity of the two microphones
31 and 32 and it is desired to suppress the sound emitted from the direction of the microphones
31 and 32, the microphones 31 and 32 are respectively The absolute values of the
corresponding weights w1 and w2 are made smaller than the absolute values of the weights w3
and w4 corresponding to the other microphones 33 and 34, respectively. For example, w1 = w2
= 0 and w3 = w4 = 1.
[0013]
Further, as illustrated in FIG. 5, it is assumed that the sound collecting and reproducing
apparatus is configured by three hands free devices H1, H2, and H3 connected in series. Each of
the hands free devices H1, H2 and H3 has the same configuration as the hands free device
illustrated in FIG. The speakers of the hands free devices H1, H2 and H3 reproduce sounds based
on the sound signals sent from the network 1, and the sound signals collected by the
microphones of the hands free devices H1, H2 and H3 are After processing by the control unit,
the phase shifter and the addition unit, a signal obtained by mixing the output signals of the
addition units 6 of all the handsfree units is sent to the handsfree unit in another space via the
network 1. In this case, for example, as shown in FIG. 19, in addition to the components of the
handsfree device of FIG. 1, an input unit 112 for inputting the output signal of the adding unit 6
of another handsfree device in the same space; A mixing unit 111 may be provided to mix the
signal from the unit 112 and the output signal of the addition unit 6 of its own device.
[0014]
Hereinafter, the weight of the weight adder 4m corresponding to the microphone 3m (m = 1, 2,
3, 4) of the handsfree device H3 is denoted as wm. In this case, the absolute value of the weight
w4 of the weight adder 44 corresponding to the microphone 34 of the handsfree device H3
facing the speaker 2 of the handsfree device H2 is greater than the absolute values of the other
weights w1, w2 and w3. Set small. For example, weights w1 = w2 = w3 = 1 and w4 = 0 are set. In
04-05-2019
5
this way, the sound from the area indicated by the dot in the room R is collected, and the sound
from the area not indicated by the dot is reduced or eliminated.
[0015]
The phase shifter 5 will be described using FIGS. 1 and 6. The phase shifter 5 shifts the phase of
the sound signal collected by the microphones 31, 32, 33, 34 by 90 degrees (step S3). For
example, as shown in FIG. 6, the phase shifter 5 is a so-called 90-degree hybrid circuit, which is a
so-called 90-degree hybrid circuit that shifts one input signal by 90 degrees with respect to the
other signal, It is comprised from the phase inverter 53 and 54 which inverts a phase. The sound
signal whose phase is shifted is sent to the adding unit 6. Assuming that the phases of the four
sound signals input to the phase shifter 5 are α degrees, a sound signal of phase α degrees and
a sound signal of phase α + 90 degrees are output from the phase shifters 51 and 52,
respectively. The phase inverter 53 inverts the phase of the sound signal of phase α degrees
output from the phase shifter 52 and outputs a sound signal of phase (α + 180) degrees. The
phase inverter 54 inverts the phase of the sound signal of phase α + 90 degrees output from the
phase shifter 52 and outputs a sound signal of phase (α + 270) degrees. As a result of addition
by the adding unit 6, the sound signal of the phase α and the sound signal of the phase (α +
180) have an antiphase relation and cancel each other, and the sound signal of the phase α + 90
degrees and the sound signal of the phase (α + 270) degrees are inverted. It becomes the
relation of the phase and cancels each other.
[0016]
The addition unit 6 adds the sound signals whose phases are shifted by the phase shifter 5 after
the control unit 4 reduces or removes at least one sound signal (step S4). The summed sound
signal is sent via the network 1 to the other handsfree device.
[0017]
Thus, while shifting the phase of the sound signal collected by the microphones 31, 32, 33, 34
and reducing the sound emitted from the speaker 2 of one's own, it is directed to another
speaker other than the one of its own. By reducing or eliminating the sound signal picked up by a
certain microphone, sound signals emitted from other speakers can also be suppressed.
[0018]
04-05-2019
6
Second Embodiment The hands-free device of the second embodiment is obtained by adding the
configuration described in JP-A-2009-44588 to the hands-free device of the first embodiment.
Specifically, in addition to the configuration of the handsfree device according to the first
embodiment, the handsfree device according to the second embodiment includes the specific
direction suppression gain calculator 7 and the frequency domain converter 8 as exemplified in
FIG. , Multiplication units 9-1, 9-2,..., 9-.OMEGA., Inverse frequency domain conversion unit 10
are further included. These additional configurations further suppress sound signals emitted by
other speakers. The following description will be made focusing on parts different from the first
embodiment. The same parts will not be repeatedly described.
[0019]
The frequency domain transform unit 8 shown in FIG. 10 transforms the signal output from the
addition unit 6 into a frequency domain signal by Fourier transform (step S5, FIG. 18). Assuming
that the entire frequency band is divided into Ω frequency bands 1, 2, ..., x, ..., Ω, all frequency
domain signals included in the frequency band x are input to the multiplier 9-x. That is, all
frequency domain signals included in frequency band 1 are input to multiplier 9-1, all frequency
domain signals included in frequency band 2 are input to multiplier 9-2, ..., frequency band Ω All
included frequency domain signals are input to the multiplier 9 -Ω.
[0020]
The specific direction suppression gain calculation unit 7 calculates gain coefficients
corresponding to the respective frequency domain signals for suppressing the sound signal from
the specific direction using the sound signals collected by the microphones 31, 32, 33, and 34
(see FIG. Step S6). The calculated Ω gain coefficients are sent to the multipliers 9-1, 9-2, ..., 9-Ω,
respectively.
[0021]
The specific direction suppression gain calculation unit 7 includes, as illustrated in FIG. 11, beam
formers 12-1, 12-2,..., 12-Q, and frequency domain converters 13-1, 13-2,. -Q, band division units
19-1, 19-2, ..., 19-Q, signal amount estimation units 14-1, 14-2, ..., 14-Ω, gain coefficient
04-05-2019
7
calculation units 16-1, 16-2, ... ..., includes 16-Ω.
[0022]
FIG. 12 shows the configuration of a beam former 12-x, which is one of the beam formers 12-1,
12-2,.
Similar processing is performed in all beamformers. The input signals xm (n) (m = 1, 2,..., M) are
input to the filter processing units FC1 to FCM. The filter processing units FC1 to FCM output a
signal x'qm (n) obtained by substituting a filter coefficient Wqm (n) given in advance (the
determination method will be described later) in the convolution operation shown in equation
(1). . In the present invention, M = 4 because the number of microphones is four.
[0023]
[0024]
Output signals of the filter processing units FC1 to FCM are input to the addition unit ADD.
The adder ADD adds the input signals as shown in equation (2) to obtain the output signal yq (n)
(q = 1,..., Q) of the beam former.
[0025]
[0026]
Here, the filter coefficient W qm (n) is a Q characteristic region Θ Q given in advance the
directivity characteristics D q (ω, θ) of each of the beam formers 12-1, 12-2, ..., 12-Q. It is
designed to emphasize and receive the sound emitted and to suppress the sound emitted in other
directions.
04-05-2019
8
Different directional regions Θ1 to ΘQ are given in advance to the beam formers 12-1, 12-2, ...,
12-Q. The output signals y1 (n), y2 (n),..., YQ (n) of the beam formers 12-1, 12-2,..., 12-Q are
frequency domain transformers 13-1, 13-2, respectively. ,..., 13-Q.
[0027]
The frequency domain conversion units 13-1, 13-2, ..., 13-Q shown in FIG. 11 decompose the
input signal into frames with a short time length (for example, about 256 samples in the case of a
sampling frequency of 16000 Hz), Discrete Fourier transform is performed in each frame, and
the obtained frequency components are output as output signals Y1 (ω, l), Y2 (ω, l), ..., YQ (ω, l).
The frequency domain transformed output signals Y1 (ω, l), Y2 (ω, l),..., YQ (ω, l) are sent to the
band division units 19-1, 19-2,. It is input.
[0028]
Assuming that the entire frequency band is divided into Ω frequency bands 1, 2, ..., x, ..., Ω, the
band division units 19-1, 19-2, ..., 19-Q respectively output signal Y 1 ( Divide ω, l), Y 2 (ω, l),...,
YQ (ω, l) into frequency bands. Thus, for example, all the output signals Y1 (ω, l), Y2 (ω, l),..., YQ
(ω, l) included in the frequency band 1 are, in other words, all Y1 (ω∈ frequency band 1). ω, l),
Y 2 (ω, l),..., YQ (ω, l) are input to the signal amount estimation unit 14-1. Similarly, output
signals Y1 (ω, l), Y2 (ω, l),..., YQ (ω, l) included in frequency band x are all Y1 (ω , L), Y2 (ω,
l),..., YQ (ω, l) are input to the signal amount estimation unit 14-x.
[0029]
FIG. 13 shows the configuration of the signal amount estimating unit 14-x, which is one of the
signal amount estimating units 14-1, 14-2,. Similar processing is performed in all signal amount
estimation units. The frequency components Y1 (ω, l), Y2 (ω, l),..., YQ (ω, l) input to the signal
amount estimation unit 14-x are power operation units PW-1, PW-2,. , PW-Q, and respective
signal power values | Y1 (ω, l) | <2>, | Y2 (ω, l) | <2>, ..., | YQ (ω, l) | <2> Are output to the area
aggregation unit 14A. The area aggregation unit 14A obtains the average of the power values of
the signals emitted from the set S of areas desired to be picked up and the power average of the
signals emitted from the set N of the areas desired to be suppressed. Output the vector Yx (ω, l).
04-05-2019
9
[0030]
[0031]
Here, NS indicates the number of areas included in the set S, and NN indicates the number of
areas included in the set N.
Also, Nx is the number of frequency bins included in the frequency band x. In addition, all
direction areas (1 to Q) are determined in advance to belong to the set S or the set N. For
example, when Q = 4, the set S and the set N may be determined as S = {1, 2} and N = {3, 4}.
[0032]
The aggregated power vector Yx (ω, l) is input to the multiplication unit 14B. The power
estimation matrix Tx <−1> (ω), which is the other input of the multiplier 14B, is an output
signal of the inverse matrix calculator 14C.
[0033]
The aggregation gain matrix Tx (ω) defined by the equation (4) is input to the inverse matrix
operation unit 14C, and the inverse matrix Tx <−1> (ω) is output.
[0034]
[0035]
Each element of the aggregation gain matrix Tx (ω) is a parameter obtained from an average
value of directivity characteristics for each direction area of each beam former as shown in FIG.
14, and, for example, as shown in equation (5) Use the mean value for the direction of the
property.
[0036]
04-05-2019
10
[0037]
αpq is an average value of the directivity characteristic for the q-th direction area of the beam
former 12-p.
The directivity characteristic can be obtained from the filter coefficient Wm (n) using, for
example, the technique described in reference 1.
[0038]
[Reference 1] Oga Juro, Yamazaki Yoshio and Kanada Toyo co-authors, Acoustic system and
digital processing, The Institute of Electronics, Information and Communication Engineers P.
181-P. 186 7.1.2.
The multiplication unit 14B shown in FIG. 13 estimates the multiplication of the input aggregated
power vector Yx (ω, l) and the power estimation matrix Tx <−1> (ω) as the estimated signal
power vector as shown in equation (6). Output Xxest (ω, l).
[0039]
Xxest (.omega., L) = Tx <-1> (. Omega.) Yx (.omega., L) (6) FIG. 15 shows one of the gain
coefficient calculators 16-1, 16-2,. It shows the configuration of a certain gain coefficient
calculation unit 16-x.
Similar processing is performed in all gain coefficient calculation units.
The estimated signal power vector Xxest (ω, l) input from the signal amount estimation unit 14
is input to the vector element extraction unit 16A. The estimated signal power vector Xxest (ω, l)
is a first component of the sound collection domain signal estimated power | S (ω, l) | <2> of the
input aggregated power vector as shown in equation (7), A suppression region signal estimated
power | N (ω, l) | <2> of the input aggregation power vector is set as a second component.
04-05-2019
11
[0040]
Xxest (ω, l) = [| S (ω, l) | <2>, | N (ω, l) | <2>] <T> (7) The vector element extraction unit 16A
estimates the sound collection area signal Each of power | S (ω, l) | <2> and suppression region
signal estimated power | N (ω, l) | <2> is output, and these are input to the SN ratio estimation
unit 16B. The SN ratio estimation unit 16B calculates the gain coefficient Rx (ω, l) for
emphasizing the signal in the desired direction region using equation (8), and outputs the result
to the multiplication unit 9-x.
[0041]
[0042]
Here, α is a parameter for adjusting the emphasis of the signal in the desired direction area by
the gain coefficient Rx (ω, l), and may be, for example, α = 1/2.
[0043]
The multipliers 9-1, 9-2, ..., 9-? Multiply each frequency domain signal by the gain coefficient
corresponding to each frequency domain signal (step S7).
The gain coefficient corresponding to the frequency domain signal included in the frequency
band x is Rx (ω, l).
That is, the multiplying unit 9-x multiplies each frequency domain signal included in the
frequency band x by the gain coefficient Rx (ω, l). The multiplication result is sent to the inverse
frequency domain transform unit 10. The same processing is performed in the other multipliers.
[0044]
The inverse frequency domain transform unit 10 performs inverse discrete Fourier transform to
output the signal y (n) restored to the time signal (step S8).
04-05-2019
12
[0045]
Thus, other signal enhancement suppressors and methods other than the first embodiment may
be implemented in parallel.
Thereby, it is possible to more effectively suppress the sound emitted by the other speakers. If
the signal is to be strongly suppressed by either the signal suppression means described in the
first embodiment or the signal suppression means added in the second embodiment, the output
sound may be distorted. By combining different signal suppression means, distortion of this
output sound can be prevented.
[0046]
[Modifications, etc.] In the above embodiment, the control unit 4 is composed of the weight
adders 41, 42, 43 and 44, but when the weight takes either 0 or 1, it is shown in FIG. As
illustrated, switches 45, 46, 47, 48 may be used instead of the weight adders 41, 42, 43, 44. In
this case, the switch corresponding to the microphone directed to another speaker other than the
speaker 2 is turned off. When the switch is on, the sound signal is output to the addition unit 6,
but when the switch is off, the sound signal is not output to the addition unit 6, so the sound
signal is removed.
[0047]
Further, as illustrated in FIG. 8, the control unit 4 may be configured by combining a weight
adder and a switch. By using a switch instead of the weight adder, the operation cost required for
weight multiplication can be reduced.
[0048]
In the above embodiment, the control unit 4 is positioned at the front stage of the phase shifter
5, but as shown in FIG. 9, the control unit 4 may be positioned at the rear stage of the phase
shifter 5. That is, step S3 of FIG. 1 may be performed before step S2. When the phase shifter 5 is
04-05-2019
13
composed of the phase shifters 51 and 52 and the phase inverters 53 and 54 as illustrated in
FIG. 6, the phase shifters 51 and 52 and the phase inverters 53 and 54 The controller 4 may be
positioned between the two.
[0049]
The sound pickup reproduction apparatus can be realized by a computer. In this case, the
processing contents of portions other than the speaker 2 and the microphones 31, 32, 33, 34
included in this device are described by a program. And each part in this apparatus is implement
| achieved on a computer by running this program by computer.
[0050]
The program describing the processing content can be recorded in a computer readable
recording medium. Further, in this embodiment, these devices are configured by executing a
predetermined program on a computer, but at least a part of the processing contents may be
realized as hardware.
[0051]
The present invention is not limited to the above-described embodiment, and various
modifications can be made without departing from the spirit of the present invention.
[0052]
[Experimental Example] The directivity in the case where each of the sound collection
reproduction method according to the related art and the first embodiment is applied to the
hands-free device H1 of the sound collection reproduction apparatus illustrated in FIG.
FIG. 16 shows the directivity of the hands-free device H1 when using the prior art. FIG. 17 shows
the weights of the microphones 31, 32, 33, 34 of the handsfree device H1 as w1 = 1 / √2, w2 =
0, w3 = 1 / √2, w4 = 1, respectively. The directivity of the handsfree device H1 when using the
sound reproduction method is shown. In FIG. 16 and FIG. 17, the intensity of the sensitivity is
expressed by light and shade of color, and the lighter the color is, the higher the sensitivity is. In
04-05-2019
14
the direction of the horizontal axis, the direction of the microphone 34 of the handsfree device
H1 is 0 degrees, and the angle increases in the counterclockwise direction.
[0053]
Comparing FIG. 16 and FIG. 17, in the prior art, in the sensitivity of the microphone array of the
handsfree device H1, a dead angle is formed in a direction different from the direction in which
the handsfree device H2 is (for example, 90 degrees or 270 degrees). The desired directional
characteristics are not realized. On the other hand, in the method according to the first
embodiment, it can be seen that a dead angle is formed in the direction (180-degree direction) in
which the handsfree device H2 is present, and desired characteristics are realized.
[0054]
On the other hand, when the sensitivity of the microphone array of the handsfree device H1 to
the speaker of the handsfree device H1 itself was also measured under the same conditions as
the above experiment, they were -350 dB (prior art) and -22 dB (present invention) respectively. .
Compared to the prior art, in the sound collection and reproduction method according to the first
embodiment, the sensitivity to the speaker of its own is higher, and although the echo
suppression performance is degraded, the suppression effect of -22 or more is obtained and
emitted from the speaker of the own device. It is possible to confirm that not only the noise
generated is suppressed, but also the sounds emitted from other devices connected in series are
suppressed, and the generation of echo and howling is prevented.
04-05-2019
15
Документ
Категория
Без категории
Просмотров
0
Размер файла
26 Кб
Теги
jp2011254242
1/--страниц
Пожаловаться на содержимое документа