close

Вход

Забыли?

вход по аккаунту

?

JPH10111691

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 JPH10111691
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
loudspeaker system which amplifies an input speech signal and outputs it from a speaker.
[0002]
2. Description of the Related Art There is a speaker with an integrated microphone, power
amplifier and speaker as a loudspeaker. Few such devices have anti-howling devices. At present,
howling prevention devices often reduce the level near the howling frequency to prevent
howling, such as graphic equalizers and notch filters, and measure the transfer function between
the speaker and microphone using an adaptive filter, and the coefficients thereof There are few
ways to cancel howling using. Also, in order to operate the adaptive filter, it is necessary to take
out a training signal such as white noise from the speaker and pick it up with a microphone to
obtain a transfer function.
[0003]
FIG. 5 is a block diagram showing an example of a conventional loudspeaker system using an
adaptive filter. In the system shown in FIG. 5, white noise is first generated by inputting the
output of the white noise generation unit 10 to the power amplifier PA, and the adaptive filter
unit 20 is operated. Then, the reverse phase of the impulse response between the speaker SP and
03-05-2019
1
the microphone M is automatically measured to adjust the characteristics of the adaptive filter
APF in the adaptive filter unit 20. By using the adaptive filter APF whose characteristics are
adjusted in this manner, howling is prevented in the block including the space 30 including the
microphone M, the microphone amplifier MA, the adaptive filter unit 20, the power amplifier PA,
the speaker SP, the speaker and the speaker SP. It can be performed. However, since the voice of
the speaker 40 corresponds to disturbance noise, it must be silent during measurement. Then,
when the measurement is completed, the characteristics of the adaptive filter APF are fixed. In
this case, the adaptive filter unit 20 shown in FIG. 5 includes an adder 60 for adding the output
signal RI of the microphone amplifier MA and the inverted signal of the output of the adaptive
filter APF, the output signal EO of the adder 60 and the white noise generator 50. An output
signal DI of the adder 50 for adding the output and an adaptive filter APF having an input error
signal EI (signal EO) as an input signal are included. Further, as an adaptive algorithm of the
adaptive filter APF, the LMS (least squares) method is adopted.
[0004]
By the way, the above-mentioned conventional loudspeakers have the following problems. (1) In
a conventional device such as a microphone, a power amplifier, and a speaker platform
integrated with a speaker, the volume is limited due to the arrangement of the microphone and
the speaker, and howling occurs when the volume is higher than a certain volume. I can not
speak loud. (2) In order to prevent this, graphic equalizers, notch filters, etc. are put through the
system to reduce the level near the howling frequency to prevent howling, but with this method
the sound quality changes, and the level above which howling occurs Has the disadvantage that
the volume can not be raised. (3) On the other hand, in the method using an adaptive filter, as
described above, it is necessary to output a training signal such as white noise from a speaker
and pick it up with a microphone to obtain a transfer function etc. It is inconvenient to (4) Also,
when the speaker's voice is used to operate the adaptive filter without using the training signal,
the level fluctuation is large due to the speech, and the calculation of the adaptive filter is prone
to errors.
[0005]
The present invention has been made under such a background, and for example, in a
microphone, a power amplifier, and a speaker-integrated loudspeaker system, loud speech is
generated at a high volume without causing feedback even if training signals are not necessarily
used. It aims to provide a device that can
[0006]
03-05-2019
2
SUMMARY OF THE INVENTION In order to solve this problem, the invention according to claim 1
measures the transfer function between the microphone and the speaker using an adaptive filter,
and uses an FIR filter that uses this as a coefficient. In a loudspeaker system that prevents
howling, a pair of adjacent directional microphones on the voice side and the speaker side is
used, while the microphone signal on the voice side is amplified and the adaptive filter is
operated with the microphone sound pickup signal on the speaker side A means for measuring
the transfer function between the microphone and the speaker; a means for convoluting the
signal on the voice side with the measured transfer function using the FIR filter and means for
canceling the echo on the voice side with the opposite phase signal thereof; Calculates the level
of a certain length of time of the voice pickup signal in real time, and determines it from the
momentary voice signal level. The voice signal detection signal, when the predetermined large
level and small level, a public address system, characterized by comprising means for controlling
the stop or start of the adaptive filter operation.
[0007]
According to the second aspect of the present invention, in the loudspeaker system according to
the first aspect, an audio signal recorded in the storage means is internally transmitted as
background music, and the adaptive filter is operated to operate between the microphone and
the speaker. It is a loudspeaker characterized in that it comprises means for measuring the
transfer function and performing howling cancellation using this.
[0008]
Also, in the invention according to claim 3, the transfer function measurement means
corresponds to the amplification gain of the microphone signal on the voice side, and the
amplification gain of the microphone collected signal on the speaker side or the amplification
gain of the input signal of the adaptive filter It is a loudspeaker according to claim 1 or 2,
characterized in that
[0009]
Also, the invention according to claim 4 is characterized in that the measured transfer function is
set to the FIR filter after multiplying by a predetermined magnification corresponding to the
loudness gain of the microphone signal on the voice side. Loud-sounding equipment.
[0010]
Also, in the invention according to claim 5, the means for controlling the stop or start of the
adaptive filter operation uses a memory having a plurality of taps to calculate the sum of squares
of a plurality of past samples of the collected sound signal of the voice. The loudspeaker
03-05-2019
3
according to any one of claims 1 to 4, wherein the level is calculated in real time by determining.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will
be described below with reference to the drawings.
1 and 2 are block diagrams showing the configuration of a loudspeaker 1 according to an
embodiment of the present invention, FIG. 1 shows the internal configuration of the loudspeaker
1, and FIG. 2 shows a schematic external view of the loudspeaker 1 There is.
Here, in the configurations shown in FIGS. 1 and 2, the same components as those shown in FIG.
5 are assigned the same reference numerals.
The present embodiment is an example in which the present invention is applied to an
integrated-type loudspeaker system 1 including a microphone 100, a power amplifier PA, a
speaker SP and the like as shown in FIG.
The loud-speaking device 1 according to the present embodiment uses a microphone 100
including two directional microphones MA and MB integrally formed or disposed close to each
other, and predetermined signals using output signals (A) and (B) of the microphone 100 as input
signals. A signal processing apparatus 300 which performs processing and outputs a signal (C),
and a speaker SP driven by an output signal (C) of the signal processing apparatus 300.
The two directional microphones MA and MB are disposed such that the directivity directions are
directed to the space 30A in the sound output direction of the speaker SP and the space 30B
behind the speaker 200, ie, the space 30B around the speaker 40.
The signal processing apparatus 300 amplifies the output signals (A) and (B) of the microphone
100, the processor 320 that receives the output signal of the microphone amplifier 310, and the
output signals of the processor 320. Are amplified and output from the power amplifier PA.
03-05-2019
4
[0012]
Next, an internal block of the signal processing device 300 will be described with reference to
FIG. The microphone amplifier unit 310 amplifies the output signal of the directional microphone
MA whose directivity direction is directed to the space 30A on the speaker SP side, and
directivity whose directivity direction is directed to the space 30B on the speaker 40 side. A
microphone amplifier MAB amplifies the output signal of the microphone MB. The processor 320
sets the output of the microphone amplifier MAB as a non-inverted input signal and the output
signal of the FIR filter (finite impulse response filter) 410 as an inverted input signal. The level of
the output power of the microphone amplifier MAB is within a predetermined range Level
detection circuit 430 for detecting whether the signal is within the range, gain control G for
adjusting the level of the signal output from the adder 420, and signals recorded in storage
means such as an optical disc such as a compact disc or a magnetic disc or memory. The white
noise · BGM source generator S that generates music source signals for white noise or BGM, the
output of gain control G, the output signal EO of adaptive filter section 440, and the output of
white noise · BGM source generator S Adder 450, adjust the magnitude of the output signal of
microphone amplifier MAA and input to adaptive filter 440 Gain control G 'which is composed of
a gain control G "to enter by adjusting the magnitude of the output signal of the adder 450 to the
adaptive filter unit 440.
[0013]
The adaptive filter unit 440 includes an adaptive filter APF configured using an FIR filter, and an
adder 460 that uses the output of the adaptive filter APF as an inverting input signal and the
output RI of the gain control G ′ as a noninverting input. There is. The adaptive filter unit 440
sets the output signal RI of the gain control G ′ as a target signal and the output of the gain
control G ′ ′ as an input signal DI, so that the output signal from the adaptive filter APF
becomes close to the target signal RI, The filter coefficients of the adaptive filter APF are
adaptively changed such that the power of the signal EI is minimized. For example, the LMS
method can be used as the adaptive algorithm. The adaptive filter unit 440 transfers each
coefficient optimized by the adaptive algorithm to the FIR filter 410 after multiplying it by a
predetermined scaling factor as necessary, thereby updating the coefficient W of each tap of the
FIR filter. On the other hand, the gain controls G, G ′, G ′ ′ are configured to operate in
conjunction with one another. Also, the white noise / BGM source generator S is not normally
operated, and is configured to operate only when a predetermined instruction is input from the
operator.
03-05-2019
5
[0014]
Next, the operation of the present embodiment configured as described above will be described.
A feature of the loudspeaker system according to the invention shown in FIG. 1 is that the
adaptive filter APF can be operated even when the speaker 40 is speaking, and it is possible to
constantly update the coefficients. However, also in the present embodiment, white noise is first
generated by operating the white noise / BGM source generator S as in the conventional
loudspeaker shown in FIG. 5, and the adaptive filter APF is operated to operate the speaker, It is
possible to set an impulse response between microphones in the FIR filter 410. The operation in
the case of setting the characteristics of the FIR filter 410 when the speaker 40 is talking will be
described. First, the voice of the speaker 40 is picked up by the directional microphone MB, and
is amplified from the speaker SP via the microphone amplifier MAB, gain control G, power
amplifier PA and the like. The loud sound is picked up by the directional microphone MA. The
output signal of the directional microphone MA is input to the adaptive filter unit 440 via the
microphone amplifier MAA and the gain control G ′. On the other hand, the level detection
circuit 430 receives the output signal LD of the microphone amplifier MAB, detects the level of
its power, and the adaptive filter unit 440 can operate only when it is within a predetermined
range. Control signal D is output. In the adaptive filter unit 440, the adaptive filter APF measures
the in-path response between the speaker SP and the directional microphone MA, and adaptively
changes its own coefficient. The measured coefficients are copied to the FIR filter 410 by
multiplying by a specific magnification. In this state, echo components fed back from the speaker
SP to the directional microphones MA and MB are canceled, and howling can be prevented.
[0015]
The adaptive filter APF performs the coefficient adaptive operation only when the output level of
the microphone amplifier MAB is within a predetermined range in accordance with the control
signal D input from the level detection circuit 430. Therefore, it is possible to suppress the
influence of the level fluctuation of the speech and to reduce the error in the calculation of the
adaptive filter. The coefficient W of the FIR filter 410 can be updated at a predetermined time
interval or at a time interval synchronized with the sampling period of the adaptive filter APF or
the like.
[0016]
In the above operation, when the set gain (sound amplification gain) of the gain control G is set to
03-05-2019
6
set the loop gain of the system mainly including the speaker SP and the directional microphone
MB to 1 or more, the speaker SP and the directional microphone In order for the loop gain of the
system consisting of MA to be maintained at the value before changing the setting gain of gain
control G (or at least to make the loop gain of the latter less than 1), Lower each set gain to
prevent howling in this system. In this case, the coefficient is copied by multiplying the
coefficient by a factor corresponding to the increase of the setting gain of the gain control G and
copying it to the FIR filter 410. By adjusting each setting gain of gain control G 'and G
"corresponding to the loop gain of the directional microphone MB side, the adaptive filter APF is
stably operated even when the loop gain becomes 1 or more. Also in such a case, it becomes
possible to update the tap coefficient W of the FIR filter 410 based on the real-time measured
value.
[0017]
In the system of the present embodiment, the transfer function from the speaker SP to the
directional microphones MA, MB does not necessarily match precisely between the two
microphones. However, since the directional microphones MA and MB are very close to each
other, for example, integrally formed, the impulse response waveform has a magnification ratio
for the direct sound from the speaker SP, the early reflection sound, etc. Although they are
different, they can be approximated as the same (similar) shape. The present invention utilizes
this characteristic. FIG. 3 is a diagram for explaining this characteristic. FIG. 3 shows waveforms
(a) and (b) output from the microphone amplifiers MAA and MAB when an impulse-like signal is
output from the speaker SP in the loudspeaker apparatus having the configuration as shown in
FIG. 2, and the microphone amplifier MAA. It is a figure which shows the waveform (c) which
synthesize | combined the thing which changed the amplitude of the signal output from and
changed it to the reverse phase, and the signal output from microphone amplifier MAB. As can be
seen from the waveform (c), even if it is not possible to cancel the later reflections of a relatively
small level, it is sufficient if the direct sound and the early reflections can be canceled. You can
get it.
[0018]
Next, the internal configuration of the level detection circuit 430 will be described with reference
to FIG. The level detection circuit 430 includes a power calculation circuit 510, an N-tap memory
520 which shifts every sampling period, and a level determination circuit 530. Where X is the
present input (corresponding to signal D in FIG. 1). The calculation result of the power
calculation circuit 510 is obtained by squaring and adding the input X of the past N samples, and
03-05-2019
7
while shifting the memory 520 for each sampling, the input of this time is obtained as P0 = P1 +
X2-PN + 1. It can be obtained by repeatedly storing the most recent power value PN + 1 from the
value obtained by squaring X and the previously calculated power value P1 and storing the result
in the most recent memory 0. The newest power value P0 is the sum of squares of the input of
the past N samples. The level detection circuit 530 compares the currently calculated power P 0
with the previously set small level and large level, and outputs an ON (“1”) signal as a level
magnitude detection flag in this range or less. In other cases, an OFF signal ("0") is output. The
level magnitude detection flag is a control signal D for controlling the stop (“1”), start or
continuation (“0”) of the adaptive filter APF. According to the level detection circuit 430
shown in FIG. 4, the level of a predetermined time length of the voice pickup signal is calculated
in real time, and the operation of the adaptive filter is performed by the signal obtained by
determining the voice signal power every moment. It is possible to control the stop or the
continuation. That is, since it can be calculated in real time, it is possible to usually perform level
detection suitable to control the operation of an adaptive filter that is always operated, and the
level can be calculated by the sum of squares of the past N samples. Since the detection is
performed, it is possible to obtain a value suitable for the LMS method adaptive algorithm of the
adaptive filter.
[0019]
According to this embodiment, since the operation of the adaptive filter can be performed using
the voice of the speaker, it is not necessary to output a training signal such as white noise from
the speaker. In addition, when the level of the speaker's voice is extremely large or small, the
operation of the adaptive filter is stopped, and the coefficient adaptive error in the adaptive filter
is reduced to operate the adaptive filter only at the average level. it can. In addition, howling
prevention using an FIR filter can be performed without using a training signal, and a large
volume of 1 or more of loop gain, which is impossible with the conventional method using a
graphic equalizer or a notch filter, can be output.
[0020]
In the above description, a loudspeaker integrated with a speaker stand is an embodiment of the
present invention, but the present invention is similarly applicable to a loudspeaker in which a
microphone and a speaker are separated.
[0021]
03-05-2019
8
As described above, according to the present invention, a pair of adjacent directional
microphones on the voice side and the speaker side is used, while the microphone signal on the
voice side is amplified, and on the other hand, on the speaker side A microphone side picks up an
adaptive filter, and means for measuring the transfer function between the microphone and the
speaker, and the FIR filter is used to convolute the signal on the voice side with the above
measured transfer function, and the resulting antiphase signal is used for voice Means for
canceling the echo on the side and the level of a certain time length of the voice pickup signal are
calculated in real time, and the voice signal detection signal obtained by determining the voice
signal level every moment is used to obtain predetermined high level and low level. At the time of
level, a means for controlling the stop or start of the adaptive filter operation is provided, so that
the training signal is not generated internally. Lecturer operates the adaptive filter in the voice
speaking, talk while the microphone, it is possible to measure the transfer function between the
speaker.
In addition, level detection can be performed, and the adaptive filter can be operated only at a
certain appropriate volume, and errors can be reduced. Furthermore, since the adaptive filter and
the FIR filter are used to prevent howling, it is possible to make the loop gain 1 or more, and it is
possible to perform loud sound with a sufficient volume. Therefore, it is particularly suitable for
use in a speaker-integrated type loudspeaker apparatus having a limited arrangement
relationship between microphones and speakers.
[0022]
Also, an audio signal recorded on a compact disc or the like is internally transmitted as
background music, and the adaptive filter is operated to measure the transfer function between
the microphone and the speaker, and provide means for performing howling cancellation using
this. Thus, for example, BGM music can be streamed and measurements can be made in advance
in a waiting time when not speaking.
[0023]
Also, in the case of performing a voice amplification with a loop gain of 1 or more by changing
the amplification gain of the microphone sound pickup signal on the speaker side or the
amplification gain of the input signal of the adaptive filter in accordance with the loudness gain
of the microphone signal on the voice side. Also, the characteristics of the FIR filter can be stably
set, and it becomes possible to cancel the feedback more accurately.
[0024]
Brief description of the drawings
03-05-2019
9
[0025]
FIG. 1 is a block diagram showing an internal configuration of a loudspeaker system according to
an embodiment of the present invention.
[0026]
FIG. 2 is a schematic view showing an appearance of the loudspeaker system shown in FIG. 1;
[0027]
FIG. 3 is a waveform diagram used to explain the operation of the loudspeaker system of the
present invention.
[0028]
FIG. 5 is a block diagram showing an example of a configuration of level detection circuit 430
shown in FIG. 1.
[0029]
FIG. 5 is a block diagram showing an example of configuration of a conventional loudspeaker.
[0030]
Explanation of sign
[0031]
APF: adaptive filter, MA, MB: directional microphone, SP: speaker, S: white noise / BGM source
generation circuit, G, G ', G ": gain control, 1: loudspeaker system, 100: microphone, 200: lecture
300, signal processing apparatus 410, FIR filter 430, level detection circuit 440, adaptive filter
unit
03-05-2019
10
Документ
Категория
Без категории
Просмотров
0
Размер файла
21 Кб
Теги
jph10111691
1/--страниц
Пожаловаться на содержимое документа