close

Вход

Забыли?

вход по аккаунту

?

JP2015204535

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 JP2015204535
[PROBLEMS] To provide a sound emission and collection device capable of suppressing
reverberation by using a sound collection and echo cancellation function for a conference
without reproducing a test voice in advance. A speaker, a filter for processing an emitted sound
signal which is an audio signal supplied to the speaker, a plurality of microphones having
directivity, a plurality of echo cancelers, a first integration unit, a reverberation characteristic
estimation unit, and an arithmetic unit Equipped with A plurality of echo cancelers are provided
corresponding to each of the plurality of microphones, and each cancels the recurrent sound
signal of the sound emitted by the speaker from the sound collection signal of the corresponding
microphone. The first integration unit integrates the adaptive filter coefficient sequence extracted
from the echo canceller. The reverberation time estimation unit estimates the reverberation time
for each frequency band of the space in which the speaker and the microphone are present,
based on the integrated filter coefficient sequence. The arithmetic unit extracts a frequency band
having a long reverberation time, calculates a filter coefficient for suppressing the power of the
frequency band, and sets the filter coefficient. [Selected figure] Figure 1
Sound emission device
[0001]
The present invention relates to a sound emitting and collecting apparatus used for remote audio
conferencing and the like, and more particularly to suppression of reverberation of sound to be
emitted.
[0002]
03-05-2019
1
An audio conference system has been put to practical use, which connects bases with a network
to transmit and receive voice.
The acoustic characteristics of the conference room used for the conference vary, and in some
cases the conference may be conducted in a room with a very long reverberation. If the
reverberation is long, the intelligibility of the sound emitted from the speaker is reduced. In order
to cope with this, there has been proposed an apparatus for suppressing the reverberation of the
sound to be emitted (Patent Document 1).
[0003]
In the device of Patent Document 1, by operating the key Kia corresponding to the participant
Ma, the inverse filter coefficient Ga of the space transfer function Ha from the participant Ma to
the microphone 31 is read from the ROM 41 and is output to the digital filter 34i. The digital
filter 34i performs inverse filter calculation in real time to reverse filter the voice signal of the
participant Ma. That is, in this apparatus, the space transfer functions from the seats of the
participants Ma to Mn to the plurality of microphones 31 are measured in advance, and inverse
filter coefficients Ga to Gn of the transfer functions are stored in the ROM 41 in advance.
[0004]
Unexamined-Japanese-Patent No. 09-247788
[0005]
However, in the above-mentioned conventional apparatus, since it is necessary to measure the
impulse response in advance, it is necessary to reproduce the test signal immediately before the
start of the conference, or to reproduce the test signal in the middle of the conference. It was an
obstacle to the smooth progress of the meeting.
[0006]
It is conceivable to go before the attendees enter the meeting, but it is preferable to do so once
the attendees are present, as the acoustic characteristics of the conference room will change
between the presence and absence of the attendees. .
03-05-2019
2
[0007]
An object of the present invention is to provide a sound emission and collection device capable of
suppressing reverberation by using a sound collection and echo cancellation function for a
conference without reproducing test sound in advance.
[0008]
A sound emission and collection device according to the present invention includes a speaker, a
filter for processing a sound emission signal that is an audio signal supplied to the speaker, a
plurality of directional microphones, a plurality of echo cancelers, and a first integration unit.
And a reverberation characteristic estimation unit and an operation unit.
A plurality of echo cancelers are provided corresponding to each of the plurality of microphones,
and each cancels the recurrent sound signal of the sound emitted by the speaker from the sound
collection signal of the corresponding microphone.
The first integration unit integrates the adaptive filter coefficient sequence extracted from the
echo canceller.
The reverberation time estimation unit estimates the reverberation time for each frequency band
of the space in which the speaker and the microphone are present, based on the integrated filter
coefficient sequence.
The arithmetic unit extracts a frequency band having a long reverberation time, calculates a filter
coefficient for suppressing the power of the frequency band, and sets the filter coefficient.
[0009]
The sound emission and collection device of the present invention includes a plurality of
directional microphones suitable for a meeting or the like. Each microphone is provided with an
echo canceller for canceling the echo of the speaker sound. The echo canceller includes an
adaptive filter that generates a simulated return sound signal, and has an adaptive filter
coefficient (estimated impulse response) that simulates an impulse response between the speaker
03-05-2019
3
and the microphone. Also, this estimated impulse response is constantly updated based on the
sound emission signal and the collected sound signal of the microphone. Since the microphone is
a directional microphone, this estimated impulse response contains only a large amount of
reverberation components coming from the direction of the directivity of the microphone, and
can not be said to completely represent the reverberation characteristics of the entire conference
room. However, by integrating the parameters of the plurality of directional microphones by the
first integration unit, it is possible to simulate an impulse response including reverberation
components coming from a wide range of directions while being a conference directional
microphone. . Then, a reverberation time is calculated using this integrated parameter (estimated
impulse response), and a filter coefficient that suppresses the reverberation is calculated. As a
result, the reverberation characteristics of the entire conference room can be accurately
reproduced, and effective reverberation can be suppressed. In addition, since the estimated
impulse response can be used as it is for the echo canceller to remove the return sound as it is,
no special calculation amount is required, and it is not necessary to emit the test voice for this
purpose.
[0010]
In the above invention, the plurality of microphones may be arranged in different directions so
that any microphone has sensitivity in all horizontal directions. By arranging the microphones in
this manner, it is possible to make the parameters integrated in the first integration unit into
substantially omnidirectional parameters.
[0011]
In the above invention, a sound pickup including a plurality of microphones and a plurality of
echo cancellers, and a communication unit including a speaker and a filter may be separately
provided. As a result, the degree of freedom of installation is increased, and a plurality of sound
collectors can be provided.
[0012]
In the above invention, a plurality of sound pickups are provided, a plurality of first integrating
units are provided corresponding to the plurality of sound pickups, and a reverberation time
estimation unit integrates the reverberation time of each sound pickup. An integration unit may
03-05-2019
4
be further provided. Then, the calculation unit may calculate the filter coefficient based on the
reverberation time integrated by the second integration unit.
[0013]
As a result, sound collectors can be installed at a plurality of places in a room where the device is
installed, and speech can be collected without leak even in a meeting in which many people
participate. Furthermore, since the reverberation time of each part of the room can be calculated
and integrated in the second integration unit, it is possible to obtain an average reverberation
time without bias.
[0014]
According to the present invention, appropriate reverberation can be suppressed by using a
directional microphone suitable for a conference and utilizing an echo canceller parameter (filter
coefficient of an adaptive filter, etc.).
[0015]
The figure which shows an example of the installation form of the audio conference system
which is embodiment of this invention The figure explaining the form of reflection of the audio |
voice in a conference room The block diagram of the sound collection system of an audio
conference system Diagram showing directivity. Block diagram of echo canceller of sound pickup
unit Block diagram of communication unit of voice conference system Functional block diagram
of parameter estimation unit of communication unit Diagram showing process flow of
dereverberation Signal waveform appearing in parameter estimation unit Diagram illustrating an
example of the gain table of the correction characteristic calculation unit of the communication
unit diagram illustrating another example of the gain table diagram showing an embodiment in
which the frequency characteristic correction function is added to the parameter estimation unit
Figure showing another connection form of the figure Figure showing an example of configuring
a sound pickup device by grouping individual microphones
[0016]
A voice conference system according to an embodiment of the present invention will be
described with reference to the drawings.
03-05-2019
5
FIG. 1 is a diagram showing an example of an installation mode of a voice conference system
according to an embodiment of the present invention.
[0017]
The audio conference system 1 is installed on the conference desk D of the conference room C.
The audio conference system 1 has one communicator 10 and one or more sound collectors 11
(four in this embodiment). The communication device 10 has a speaker 26. The sound pickup 11
includes a plurality of microphone elements 31. The communication device 10 and the sound
collection device 11 are mutually connected by the communication cable 12, and perform digital
communication. The sound pickup device 11 transmits the voice signal picked up by the
microphone element 31 and the filter coefficient of the echo canceller 32 (see FIG. 3) to the
communication unit 11. The communication device 10 is connected to a personal computer 2
which is a host device. The personal computer 2 communicates with another audio conference
system installed at another site via the network 3 such as the Internet, and is input from the
communicator 10 of the audio conference system 1 (the microphone element 31 is collected
While transmitting the sound signal (sounded) to the other audio conference system, the audio
signal received from the other audio conference system is input to the communication device 10.
The communication device 10 emits from the speaker 26 the audio signal sent from the other
audio conference system.
[0018]
FIG. 2 is a diagram for explaining the form of sound reflection in the conference room C. As
shown in FIG. The sound emitted from the speaker 26 directly reaches the participant M of the
conference and the microphone element 31 and is variously reflected on the wall and the ceiling
of the conference room C to reach the participant M and the microphone element 31.
[0019]
When the sound emitted from the speaker 26, that is, the sound collected by another audio
conference system installed at another location is collected by the microphone element 31 and
03-05-2019
6
transmitted to the other audio conference system, transmission is performed. The voice that has
been played back becomes so-called echo that is reproduced. In order to prevent this echo, an
echo canceller 32 (see FIG. 3) that cancels the sound emitted from the speaker 26 is connected to
the microphone element 31. In addition, in order to improve that the sound emitted from the
speaker 26 becomes a voice embedded by the reverberation reflected by the wall or ceiling of the
conference room C and the intelligibility is lowered, the filter for reverberation suppression is
provided in the communication device 10 24 (see FIG. 6) are provided. The filter coefficients of
the filter 24 are calculated using the filter coefficients of the adaptive filter 35 (see FIG. 5) of the
echo canceller 32.
[0020]
The functions and operations of the echo canceller 32 and the reverberation suppressing filter
24 will be described later with reference to the drawings following FIG. The function units built
in the communication device 10 and the sound collection device 11 described below may be
configured by an electronic circuit, or may be realized by the cooperation of a program such as a
processor such as a computer.
[0021]
FIG. 3 is a block diagram of the sound pickup device 11. FIG. 4 is a diagram showing the
directivity of each of the three microphone elements 31 of the sound collection device 11. FIG. 5
is a block diagram of the echo canceller 32 of the sound collection device 11.
[0022]
The sound pickup device 11 includes three microphone elements 31. As shown in FIGS. 1 and 4,
the sound pickup device 11 has a disk-like planar shape, and three microphone elements 31 are
directed outward (in the normal direction) at intervals of 120 degrees on its circumference. It is
provided radially. Each microphone element 31 is a unidirectional microphone and has a
cardioid-shaped sound collection characteristic centered on the direction in which the
microphone element 31 faces. Each microphone element 31 is provided at an interval of 120
degrees, and its directivity characteristic is arranged as shown in FIG. 4. Therefore, if the sound
collection signals of each microphone element 31 are combined, a signal having almost near
omnidirectional characteristics can get. The directivity of the microphone element 31 is not
03-05-2019
7
limited to that of the cardioid. It may have some directivity at the rear, or may be bi-directional.
[0023]
In FIG. 3, each microphone element 31 is provided with an echo canceller 32. Although the
detailed configuration of the echo canceller 32 will be described with reference to FIG. 5, the
sound emitted from the speaker 26 from the sound signal collected by the microphone element
31 is canceled. A voice signal from which the wraparound voice of the speaker 26 has been
canceled by the echo canceller 32 is input to the voice selection unit 33. The audio selection unit
33 receives an audio signal collected by the three microphone elements 31. Based on the levels
and durations of the three input voice signals, the voice selection unit 33 determines which voice
signal input from which microphone element 31 is dominant, that is, it is estimated to be the
speech voice signal of the speaker And select one speech signal estimated as a speech signal.
That is, in the collection of voice signals in a conference, the optimum one microphone element
31 is selected from the three microphone elements 31 by utilizing the characteristics of the
directional microphone, and speech voice with a good S / N ratio is collected. ing. The selected
voice signal is transmitted to the communicator 10 via the communication interface 34. When a
plurality of sound pickup devices 11 are connected to the communication device 10, the
communication device 10 (microphone mixer 22: see FIG. 6) is configured to transmit the level,
duration time, and the like of the sound signal from the sound signal received from each sound
collection device 11 The degree of correlation is compared to select one more, or a plurality of
audio signals are mixed, and the selected or mixed audio signal is transmitted to the other
system.
[0024]
Next, the configuration of the echo canceller 32 will be described with reference to FIG. FIG. 5 is
a block diagram of the echo canceller 32. As shown in FIG. The echo canceller 32 includes an
adaptive filter 35 having a filter coefficient setting unit 35B and a variable filter 35B, and further
includes an adder 37. In general, an adaptive filter is a filter that automatically adapts its own
transfer function (adaptive filter coefficient sequence) in accordance with a predetermined
optimization algorithm.
[0025]
The filter coefficient setting unit 35B estimates the transfer function of the acoustic transfer
system (the acoustic propagation path from the speaker 26 to the microphone element 31) of the
03-05-2019
8
conference room C, and changes the filter coefficient that can be a filter of the estimated transfer
function. Set to
[0026]
An audio signal (sound emission signal) emitted from the speaker 26 is input to the variable filter
35A.
Since the transfer function of the variable filter 35A is a transfer function in which the acoustic
transfer system (the sound propagation path from the speaker 26 to the microphone element 31)
of the conference room C is simulated, the sound emission signal filtered by the variable filter
35A is It is a voice signal (pseudo-regression sound signal) simulating a voice signal (regression
sound signal) emitted from the speaker 26 and propagated to the conference room C and
collected by the microphone element 31. The pseudo-regression sound signal is input to the
adder 37.
[0027]
Further, an audio signal (sound collection signal) collected by the microphone element 31 is
input to the adder 37. The adder 37 subtracts the simulated regression sound signal from the
collected sound signal and outputs it. The collected sound signal includes a voiced speech signal
of the meeting attendee M and a return sound signal which is emitted from the speaker 26 and
goes around. The adder 37 can remove the regression sound from the collected signal, that is,
cancel the echo, by subtracting the simulated regression sound signal from the collected signal.
The sound pickup signal from which the echo has been canceled is input to the voice selection
unit 33, and is also input to the filter coefficient setting unit 35B as a reference signal. A sound
emission signal, which is an audio signal emitted from the speaker 26 as a reference signal, is
also input to the filter coefficient setting unit 35B. The filter coefficient setting unit 35B
continuously updates the filter coefficient based on these reference signals. Further, the updating
of the filter coefficient automatically detects a time interval in which the voice is emitted from the
speaker 26 and the participant M who is in the meeting room C does not speak, and the
reference of the time interval is performed. It is done using a signal.
[0028]
03-05-2019
9
Here, the variable filter 35A is an FIR filter. Therefore, the filter coefficients set in the variable
filter 35A are simulated by estimating the impulse response of the sound propagation path from
the speaker 26 to the microphone element 31 by the filter coefficient setting unit 35B. The filter
coefficient setting unit 35B transmits the filter coefficient to the communication unit 10 via the
communication interface 34 as an estimated impulse response.
[0029]
As described above, one of the voice signals picked up by the three microphone elements 31 is
selected by the voice selection unit 33 and transmitted to the communicator 10, but 3
corresponding to the three microphone elements 31. Three estimated impulse responses are all
transmitted to the communicator 10. As described later, the parameter estimation unit 23 of the
communication device 10 combines these three estimated impulse responses. The three
estimated impulse responses are impulse responses including reverberant components coming
from the direction in which the corresponding microphone element 31 faces as shown in FIG. 4,
but they arrive from all directions by combining the three. It is possible to simulate impulse
responses from all directions of the conference room C collected by an omnidirectional
microphone including a reverberation component.
[0030]
FIG. 6 is a block diagram of the communication device 10. The communication device 10 has a
communication interface 21 for communicating with the personal computer 2, a microphone
mixer 22, a parameter estimation unit 23, a filter 24, an audio circuit 25, a speaker 26 and a
communication interface 27 for communicating with the sound collector 11. . The
communication interface 21 is an interface for performing digital communication with the
personal computer 2, and for example, a USB interface is used. When the USB interface is used,
the personal computer 2 is a host and the communication device 10 is an audio device. A
plurality of communication interfaces 27 are provided, and the individual sound collecting
devices 11 are connected via the cables 12 respectively. The communication interface 27 may
use, for example, a wired LAN interface.
[0031]
03-05-2019
10
The communicator 10 receives an audio signal (echo-collected signal) and three estimated
impulse responses from the sound collector 11 via the communication interface 27. The received
audio signal is input to the microphone mixer 22. A plurality of audio signals received from
different sound collectors 11 are input to the microphone mixer 22 from the plurality of
communication interfaces 27 respectively. The microphone mixer 22 selects or mixes audio
signals received from the plurality of sound collectors 11 into a monaural audio signal, and
transmits the monaural audio signal to the personal computer 2 via the communication interface
21. The personal computer 2 transmits this voice signal to the voice conference system at
another site via the network 3. The microphone mixer 22 may select the speech signal to be
transmitted to the other system by comparing the level of the speech signal of the
communication device, the duration, or the degree of correlation and transmitting the speech
signal having a good S / N ratio.
[0032]
The personal computer 2 also receives an audio signal from an audio conference system at
another site. The voice signal is input through the communication interface 21 and is input to the
filter 24 as a sound emission signal emitted from the speaker 26, and is also transmitted to each
sound collector 11 through the communication interface 27.
[0033]
The filter 24 performs a filtering process to suppress a drop in the intelligibility of the voice due
to the reverberation in the conference room C. That is, signal processing is performed on the
sound emission signal so as to suppress the level of the frequency band having a long
reverberation time. In particular, since the reverberation of the bass region causes the reduction
of the intelligibility, the degree of suppression is made stronger for the bass region. Such filter
coefficients are determined by the parameter estimation unit 23. A sound emission signal whose
frequency band with a long reverberation time is suppressed by the filter 24 is input to the audio
circuit 25. The audio circuit 25 converts the sound emission signal into an analog audio signal,
amplifies it at a predetermined level, and inputs it to the speaker 26. The speaker 26 emits the
sound emission signal to the conference room C as sound. The emitted sound is heard by the
attendees M of the conference and collected by the microphone 31.
[0034]
The sound emission signal transmitted to the sound collection device 11 via the communication
03-05-2019
11
interface 27 is input as a reference signal to the filter coefficient setting unit 35B of the echo
canceller 32 shown in FIG.
[0035]
FIG. 7 is a block diagram of the parameter estimation unit 23.
Further, FIG. 8 is a diagram showing the procedure of the reverberation suppression process
executed by the voice conference system 1 including the parameter estimation unit 23. Further,
FIG. 9 is a diagram illustrating a signal waveform that appears in the procedure of the
reverberation suppression process.
[0036]
In FIG. 8, the sound collection device 11 is used for sound collection (S 101) by the directional
microphone element 31, echo cancellation processing (S 102), and extraction of the filter
coefficient (estimated impulse response) from the adaptive filter 35 (S 103). Do. The sound
pickup device 11 transmits the filter coefficients of the three echo cancellers 32 provided
corresponding to the three microphone elements 31 to the communication device 10 as an
estimated impulse response.
[0037]
In FIG. 7, the parameter estimation unit 23 includes a filter coefficient integration unit 40 for
each connected sound collection device 11, a reverberation time estimation unit 41 for each
connected sound collection device 11, a reverberation time integration unit 42, and correction. A
characteristic calculation unit 43 and a filter coefficient calculation unit 44 are provided.
[0038]
The communication device 10 receives three estimated impulse responses (filter coefficients)
from each of the sound collection devices 11.
03-05-2019
12
The received estimated impulse response is input to the parameter estimation unit 23. In the
parameter estimation unit 23, the input estimated impulse response is input to the filter
coefficient integration unit 40 provided for each sound collection unit 11. The filter coefficient
integration unit 40 combines the three estimated impulse responses that have been input, with
their time axes aligned. This synthesis may be simply addition synthesis, the weighting of each
estimated impulse response may be changed, or the time shift of each impulse response may be
corrected. By combining the estimated impulse responses in three directions shown in FIG. 4, an
impulse response (ideally 360 degree omnidirectional) including reverberant components
coming from a wider range than the estimated impulse response for one microphone It is
possible to estimate This process is the pre-integration of S104 in FIG. This process is performed
for each of the connected sound collectors 11 (the estimated impulse response is input), and the
impulse response at each sound collector 11 position is estimated.
[0039]
The wide directivity estimated impulse response synthesized by the filter coefficient integration
unit 40 is input to the reverberation time estimation unit 41. The reverberation time estimation
unit 41 performs the following processing. First, the estimated impulse response is passed
through multiple channel band pass filters to divide the band. Although the number of channels
to be divided and the frequency band of each channel are arbitrary, for example, band division
such as dividing 315 Hz to 8000 Hz into 15 channels may be performed. By this processing, the
impulse response of the signal component of each frequency band (channel) is estimated. This
process is the process of S105 of FIG. This process is also performed for each sound collector 11.
[0040]
The reverberation time estimation unit 41 obtains the reverberation time of the signal of each
frequency band based on the estimated impulse response of each frequency band. The
reverberation time generally refers to the time until the signal level attenuates to -60 dB (one
millionth of a million). There are various calculation and estimation methods, but the
reverberation time may be determined by the Schroedar method here. The Schroeder method
determines the impulse response by Schroeder integration or backward cumulative addition to
obtain a Schroeder curve (reversing attenuation curve) as illustrated in FIG. 9A, and determines
the time until this curve reaches −60 dB. Just do it. For simplicity, a predetermined section not
including the direct sound or error component of the Schroder curve may be taken out, the slope
of the section may be taken as the slope of this curve, and the time to attenuate from 0 dB to
03-05-2019
13
−60 dB may be estimated by this slope. . This process corresponds to S106 in FIG. 8 and is
performed for each frequency band of each sound collector 11, and the reverberation time for
each frequency band at the position of each sound collector 11 is estimated.
[0041]
Then, the reverberation time for each frequency band at the position of each sound collector 11
estimated by the plurality of reverberation time estimation units 41 is input to the reverberation
time integration unit 42. The reverberation time integration unit 42 synthesizes reverberation
time at the position of each sound collector 11 for each frequency band. This process is the poststage integration process of S107 of FIG. 8 and is performed for each frequency band.
[0042]
The synthesis of this post-integration process is performed by averaging the reverberation time
of each sound collector 11 for each frequency band, but reverberation time values (outliers)
extremely separated from the average value are excluded from the average. You may Also,
regarding the sound collector 11 with many outliers, the reverberation time of the sound
collector 11 is assumed to be the entire surface in all frequency bands, since it may be installed
in a place where biased characteristics such as the corner of a room are likely to appear. It may
be excluded from the post integration process. The processing of this exclusion may be
performed at the discretion of the communicator 10, and a clerk installing the system in the
conference room C manually operates the sound collector 11 or the communicator 10 to select a
specific sound collector 11, for example, The sound collector 11 or the like installed at the corner
of the room may be set to be excluded from the target of the post-integration. In this case, for the
sound collection device 11 excluded from the integration target, the process after S103 becomes
unnecessary, and the process is reduced.
[0043]
By plotting the reverberation time for each frequency band determined by the post-stage
integration process on the frequency axis, for example, the averaged reverberation
characteristics of the entire conference room C as shown in FIG. 9B can be determined. Based on
this reverberation characteristic, it is possible to determine which frequency band the
reverberation time is long.
03-05-2019
14
[0044]
The reverberation characteristic obtained by the reverberation time integration unit 42 is input
to the correction characteristic calculation unit 43. The correction characteristic calculation unit
43, based on the input reverberation characteristic, suppresses the frequency band where the
reverberation time is long so that the sound emitted from the speaker 26 is not obscured by the
reverberation sound of the sound. Determine the correction characteristics of The correction
characteristic is determined by setting a threshold of reverberation time for each frequency band,
extracting a frequency band in which the reverberation time exceeds the threshold, and
suppressing the power of this frequency band, or known to suppress reverberation. It is possible
to select the method of determining the amount of suppression of the power for each frequency
band by using the filtering method of and the gain table for each frequency band. The gain table
for each frequency band as shown in FIG. 10 can be used. In the gain table, the vertical axis
represents gain (dB) and the horizontal axis represents reverberation time RT (seconds), and the
gain value for each frequency band is indicated by a line segment having a slope. The line
segments of f1 to fn correspond to the frequency band divided by the above-mentioned band
pass filter, and f1 is on the bass side and fn is on the treble side. For example, if the reverberation
time is 1.0 second in the band f3, the gain is set to -30 dB. In this gain table, line segments of the
bass region are set to have a steeper slope. At f4 on the high band side, the gain is around -24 dB
if the reverberation time is 1.0 second. As described above, when the reverberation time of the
low tone range is long, a correction characteristic is determined to suppress the low tone range
more strongly than when the reverberation time of the high tone range is long.
[0045]
The gain value has a lower limit so that suppression exceeding a predetermined value (−30 dB
in the gain table shown in the drawing) is not performed. In addition, when the predetermined
reverberation time (1.0 seconds in f3) is exceeded in each frequency band, the lower limit value
of the gain is applied. Further, as shown in FIG. 11, the gain table may be set such that
convergence points at which a plurality of line segments converge are shifted in the positive
direction by a fixed reverberation time. In the case of this figure, the gain is 0 when the
reverberation time is 1.0 second or less. This process is S108 of FIG. The determined correction
characteristic is input to the filter coefficient calculation unit 44.
[0046]
03-05-2019
15
The filter coefficient calculation unit 44 determines the filter characteristic so that the filter 24
has the correction characteristic calculated by the correction characteristic calculation unit 43.
The filter 24 is composed of an FIR filter and an IIR filter. The filter coefficients are calculated
according to the configuration of the filter 24 by operations such as discrete time inverse Fourier
transform or parametric peak filter. This process is S109 of FIG. The calculated filter coefficient
is set to the filter 24 (S110). By filtering the sound emission signal with this filter 24, the
reverberation of the sound emitted from the speaker 26 is suppressed, and the voice of the
participant M becomes clear.
[0047]
In the above embodiment, the reverberation characteristic of the conference room C is estimated
using the filter coefficient of the echo canceller 32, and by suppressing the long frequency band
of the reverberation, the intelligibility of the emitted voice is not reduced. Furthermore, the filter
unit of the echo canceller 32 is used to estimate the frequency characteristic of the conference
room C, and the frequency characteristic of the emitted signal is set to the frequency of the
conference room C so that the emitted sound is heard with flat characteristics. It may be
corrected to cancel the characteristic. As a result, it is possible to prevent not only the
reverberation but also the lowering of the intelligibility due to the frequency characteristics of
the conference room C.
[0048]
FIG. 12 shows a modification of the parameter estimation unit. In addition to the correction
characteristic for suppressing reverberation, the parameter estimation unit 23 ′ in this figure
determines the correction characteristic for correcting the frequency characteristic and sets it in
the filter 24. Parts in FIG. 12 identical to those in FIG. 7 are assigned the same reference
numerals and explanation thereof is omitted. In addition to the configuration of the parameter
estimation unit 23 of FIG. 7, the parameter estimation unit 23 ′ of this figure further includes a
frequency characteristic estimation unit 45 for each sound collection unit 11 and a frequency
characteristic integration unit 46.
[0049]
03-05-2019
16
The estimated impulse response of wide directivity (non-directivity) for each sound collector 11
output from the filter coefficient integration unit 40 is input to the reverberation time estimation
unit 41 and is also input to the frequency characteristic estimation unit 45 . The frequency
characteristic estimation unit 45 Fourier-transforms the input impulse response to calculate the
frequency characteristic at the position of the sound pickup device 11. This frequency
characteristic is input to the frequency specification integration unit 46. The frequency
characteristic integrating unit 46 synthesizes the frequency characteristics of the sound
collectors 11 input from the frequency characteristic estimating units 45 and calculates an
average value of the entire frequency characteristics in the conference room C. The calculation of
the average value may be simply arithmetic averaging, or each frequency characteristic may be
normalized and then the average may be obtained.
[0050]
The frequency characteristic of the conference room C obtained by the frequency characteristic
integration unit 46 is input to the correction characteristic calculation unit 43 '. The correction
characteristic calculation unit 43 'is a characteristic that suppresses a long frequency band of
reverberation time, and the emitted sound cancels the frequency characteristic influenced by the
conference room C and passes through a flat transfer characteristic. Calculate correction
characteristics that reach the listener. Further, this calculation method may calculate not the
correction characteristics that reach the listener through flat transfer characteristics, but the
correction characteristics that reach the listener through any ideal transfer characteristics set in
advance. The calculated correction characteristic is input to the filter coefficient calculation unit
44. The filter coefficient calculation unit 44 determines the filter characteristic so that the filter
24 has the correction characteristic calculated by the correction characteristic calculation unit
43. The calculated filter coefficient is set to the filter 24. By filtering the sound emission signal
with this filter 24, the sound emitted from the speaker 26 has characteristics such as passing
through a flat transfer characteristic and the reverberation thereof is suppressed, and the voice
with high clarity for the participant M Become.
[0051]
In the above embodiment, although the communicator 10 performs the pre-stage integration, the
sound collector 11 may perform this. In FIG. 8, it is preferable that the processing of S101 to
S103 be performed by the sound collection device 11. Moreover, it is preferable that the
communicator 10 perform the process of S107 and subsequent steps. The processing of S104 to
03-05-2019
17
S106 during that time may be performed by either the sound collector 11 or the communication
device 10.
[0052]
In the above embodiment, the audio conference system has been described in which the sound
collection device 11 including the microphone element 31 is connected to the communication
device 10 including the speaker 26. However, a plurality of microphone elements 31 and the
speaker 26 are integrated. The present invention is also applicable to an audio conference
apparatus (only communicator 10) provided in the above.
[0053]
Further, the form of connection between the communication device 10 and the sound collection
device 11 is not limited to wired connection.
For example, wireless connection such as a wireless LAN or a short distance wireless
communication standard may be used.
[0054]
Further, the shape of the sound collection device 11 and the number of the microphone elements
31 are not limited to those shown in FIGS. 1 and 4. For example, two or four microphone
elements 31 may be provided at equal intervals at the peripheral edge of the disk-like housing. In
this case, the angle of each microphone element 31 is 180 degrees or 90 degrees. Also, the
intervals (angles) may not be equal. The microphone element 31 may be provided in a single
direction in a direction to be directed to the meeting participant M.
[0055]
In the case where a plurality of sound pickups 11 are connected to the communication device 10,
as shown in FIG. This connection configuration can save the entire cable length. By using the
communication interfaces 21 and 34 as the LAN interface, it is possible to correspond to both the
star connection of FIG. 1 and the daisy chain connection of FIG.
03-05-2019
18
[0056]
Further, as shown in FIG. 14, a plurality of microphones 51, 52, 53, 54 incorporating one
microphone element 31 are combined (grouped), and this one group 60 is made to function as
one sound collecting device 11. May be In this case, the clerk may set grouping information of
the table microphone 51 in the communication device 10 in advance, and a signal distribution
unit is provided at the front end of the communication device 10, and the communication device
10 performs the grouping by itself. You may do so. In this case, the signal distribution unit
groups table microphones collecting similar signals as the same group based on the time position
of the adaptive filter of the echo canceller, the degree of correlation of the collected audio
signals, etc. You may do so.
[0057]
FIG. 14A shows an example in which a plurality of table microphones (stand microphones) 51 are
combined to form a group 60. Further, FIG. 14B shows an example in which a plurality of hand
microphones 52 are grouped 60. The hand microphone 52 may be wired or wireless. In this case,
since a speaker having a hand microphone moves, a plurality of hand microphones 52 existing at
a certain distance may be set as one group 60, and adaptive filter coefficient sequences may be
added for each group 60. When there are a plurality of groups 60 of the hand microphone 52,
the reverberation time of each group may be calculated to obtain the above-mentioned
reverberation characteristics. The presence of the plurality of hand microphones 52 at a
predetermined distance is determined by detecting the position difference by calculating the
sound collection delay difference of the two hand microphones 52 and by mutually detecting the
strengths of the radio waves emitted by the hand microphones 52. be able to.
[0058]
Moreover, the sound collector 11 may not be placed on the conference desk D. That is, as shown
in FIG. 14C, a hanging microphone 53 suspended from a ceiling, a wall microphone 54 installed
on a wall, or the like may be used. Of course, the sound collection device 11, the table
microphone 51, the hand microphone 52, the suspension microphone 53 and the wall
microphone 54 of FIG. 1 may be mixed.
03-05-2019
19
[0059]
In addition, the audio conference system 1 of the present embodiment can naturally be used
other than the conference. Also, the place to be used is not limited to the conference room.
[0060]
C Conference Room D Conference Desk M Conference Participant 1 Audio Conference System 2
Personal Computer 3 Network 10 Communication Device 11 Sound Collector 26 Speaker 31
Microphone Element 51 Table Microphone 52 Hand Microphone 53 Suspension Microphone 54
Wall Microphone 60 Group
03-05-2019
20
Документ
Категория
Без категории
Просмотров
0
Размер файла
34 Кб
Теги
jp2015204535
1/--страниц
Пожаловаться на содержимое документа