close

Вход

Забыли?

вход по аккаунту

?

JP2007081815

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 JP2007081815
PROBLEM TO BE SOLVED: To provide a speaker device capable of reducing non-linear distortion
generated from a speaker with high accuracy. SOLUTION: A linear signal generation unit 10
simulates a linear signal based on a transfer function of a linear component in a transfer function
from an input of an amplifier 14 to an output of a sensor 16 and outputs the linear signal to an
adder 12. The sensor 16 detects the vibration or sound output of the speaker 15 and outputs a
detection signal to the adder 12. The adder 12 subtracts the linear signal from the detection
signal of the sensor 16, and outputs the subtracted signal as an error signal. The error signal
output from the adder 12 is output to the adder 11 through the feedback control filter 13 and is
subjected to negative feedback on the input acoustic signal in the adder 11. [Selected figure]
Figure 1
Speaker device
[0001]
The present invention relates to a speaker device, and more particularly to a speaker device that
reduces distortion generated from a speaker.
[0002]
Conventionally, in the process of converting an electrical signal to an acoustic output by a
speaker, non-linear distortion occurs due to mechanical limitations of the speaker, so that a
difference occurs in the acoustic output level depending on the frequency of the input electrical
signal.
08-05-2019
1
That is, there is a problem that the frequency characteristic of the sound output level is not flat.
Then, in order to solve this subject, the speaker apparatus which reduces the nonlinear distortion
resulting from the said subject is proposed (for example, refer patent document 1). FIG. 10 is a
block diagram showing a configuration of a conventional speaker device for reducing the nonlinear distortion.
[0003]
The conventional speaker device shown in FIG. 10 includes an impedance equalizer 91, adders
92 and 93, switches 94 and 95, a measurement signal generator 96, an amplifier 97, a speaker
98, a resistor 99, a measurement storage unit 100, and control. A unit 101 is provided. In FIG.
10, first, the impedance of the speaker 98 is measured. At the time of measurement, the switch
94 is switched so that the measurement signal output from the measurement signal generator 96
is input to the amplifier 97. Then, the measurement signal is input to the speaker 98 via the
amplifier 97. At this time, the drive current flowing through the speaker 98 is detected as the
terminal voltage of the resistor 99. Further, the switch 95 is switched so that the signal input to
the speaker 98 is input to the measurement storage unit 100. The measurement storage unit 100
measures the impedance of the speaker 98 based on the detected drive current, and stores the
measured value. The control unit 101 outputs a control signal to the impedance equalizer 91 so
that the impedance equalizer 91 expresses the same characteristic as the impedance stored in
the measurement storage unit 100. Next, at the time of operation of the speaker 98 to which an
acoustic signal is inputted (when the switches 94 and 95 are in the state shown), the impedance
equalizer 91 uses the impedance characteristic set at the time of measurement to input the
acoustic signal. The linear component of the drive current of the speaker 98 is output. Then, in
the adder 93, the drive current of the linear component processed in the impedance equalizer 91
is subtracted from the drive current of the speaker 98 including non-linear distortion, and the
subtracted signal is fed back to the adder 92. Thereby, the non-linear component contained in
the drive current of the speaker 98 is removed, and the non-linear distortion generated from the
speaker 98 is reduced.
[0004]
Further, as shown in FIG. 11, a speaker apparatus has been proposed which corrects the
frequency characteristic of the speaker 98 to a target characteristic including the solution of the
above-mentioned problem (see, for example, Patent Document 2). FIG. 11 is a block diagram
08-05-2019
2
showing a configuration of a conventional speaker device that corrects frequency characteristics
to a target characteristic.
[0005]
The conventional speaker device shown in FIG. 11 includes a target characteristic filter 102, an
equalizer 103, an adder 104, an amplifier 97, a speaker 98, and a microphone 105. In FIG. 11, an
acoustic signal is input to the equalizer 103 and the target characteristic filter 102, respectively.
In the target characteristic filter 102, digital filter coefficients are set so as to realize a
characteristic (hereinafter referred to as target characteristic) as a correction target of the
frequency characteristic of the speaker 98. The adder 104 subtracts the output signal of the
microphone 105 from the signal processed by the target characteristic filter 102, and outputs
the subtracted signal to the equalizer 103 as an error signal. The equalizer 103 self-adjusts the
frequency characteristics of the amplitude and phase of the internal filter so that the root mean
square value of the value of the error signal is minimized. That is, the equalizer 103 is configured
to achieve the so-called minimum mean square error method. The equalizer 103 processes the
input acoustic signal so that the frequency characteristic of the acoustic output outputted from
the speaker 98 becomes the target characteristic set in the target characteristic filter 102.
[0006]
Here, an example of the correction result in the conventional speaker device shown in FIG. 11 is
shown using FIG. 12 and FIG. FIG. 12 is a diagram showing an amplitude frequency characteristic
of the speaker 98 before the correction process. On the other hand, FIG. 13 is a diagram showing
the amplitude frequency characteristic of the speaker 98 after the correction processing. The
amplitude frequency characteristic shown in FIG. 13 indicates the characteristic when the
transfer function of the target characteristic set by the target characteristic filter 102 is “1”.
When these amplitude frequency characteristics are compared, it can be understood that the
complex relief is flattened by the correction. As described above, the conventional speaker device
shown in FIG. 11 automatically corrects the frequency characteristic of the sound output
outputted from the speaker 98 to be the target characteristic. Japanese Patent Application LaidOpen No. 62-87000 Patent No. 2530474
[0007]
08-05-2019
3
However, in the conventional speaker device shown in FIG. 10, since the drive current of the
speaker 98 to be detected is weak and the resistor 99 used for detection thermally fluctuates, it
is difficult to detect the drive current having high SN. is there. Further, by detecting the drive
current of the speaker 98, air distortion can be detected in a frequency band in which the
diaphragm vibrates in a pictonic manner, but detection accuracy of air distortion is lowered in
the middle and high frequencies where divided vibration occurs. Further, since the impedance
equalizer 91 is configured by an analog circuit, it is difficult to model the impedance with high
accuracy, and as a result, only the drive current of the non-linear component can be extracted as
an error signal in the adder 93. It will be difficult. For these reasons, in the conventional speaker
device shown in FIG. 10, there is a problem that the nonlinear distortion reduction processing
can not be performed with high accuracy, and a sufficient reduction effect can not be obtained.
Furthermore, in the conventional speaker apparatus shown in FIG. 10, since the error signal is
directly input to the adder 92, there is a problem that only the maximum “−6 dB” reduction
effect can be theoretically obtained.
[0008]
Further, in the conventional speaker device shown in FIG. 11, as a result of correction to the
target characteristic, the speaker 98 is overloaded, particularly when the output level is increased
in the low frequency band. Then, there is a problem that non-linear distortion caused by the
correction is generated due to the overload. In particular, due to this non-linear distortion, the
sound quality deterioration at frequencies near the primary resonance frequency of the speaker
98 becomes remarkable. As described above, the conventional speaker device shown in FIG. 11
has a problem that it is not possible to freely correct the target characteristics.
[0009]
Therefore, a main object of the present invention is to provide a speaker device capable of
reducing non-linear distortion generated from a speaker with high accuracy. Another object of
the present invention is to provide a speaker device capable of correcting frequency
characteristics of a speaker's acoustic output to target characteristics while accurately reducing
non-linear distortion generated from the speaker.
[0010]
A first invention is a speaker device, which comprises a speaker, a sensor for detecting vibration
or sound output of the speaker, an adder for receiving an input sound signal and outputting the
08-05-2019
4
sound to the speaker, and an input sound signal as an input A signal generation unit that
generates the same signal as the linear component included in the detection signal of the sensor
based on the processing coefficient of the error; an error detection unit that detects an error
between the detection signal of the sensor and the generation signal of the signal generation
unit; And a control unit that controls the gain and / or phase of the error signal detected in the
unit and outputs the signal to the adder.
[0011]
In a second aspect based on the first aspect, the output signal of the adder is corrected based on
a predetermined filter coefficient so that the frequency characteristic of the sound output of the
speaker becomes a predetermined frequency characteristic, and the corrected signal is output to
the speaker The image processing apparatus further includes a first correction filter, and a
second correction filter that corrects the input sound signal based on the same filter coefficient
as the first correction filter and outputs the corrected signal to the signal generation unit.
[0012]
In a third aspect based on the second aspect, the filter coefficient is calculated at predetermined
time intervals using the output signal of the adder and the detection signal of the sensor, and the
filter coefficients of the first and second correction filters are calculated And an adaptive
updating unit that updates the filter coefficients of the first and second correction filters so as to
obtain the calculated filter coefficients.
[0013]
A fourth invention is the process according to the second invention, wherein the output signal of
the first correction filter and the detection signal of the sensor are used to represent the linear
component of the transfer function from the output of the first correction filter to the output of
the sensor The system further includes an identification updating unit that calculates the
coefficients at predetermined time intervals and updates the processing coefficients of the signal
generation unit such that the processing coefficients of the signal generation unit become the
calculated processing coefficients.
[0014]
In a fifth aspect based on the first aspect, processing coefficients representing a linear
component of a transfer function from the output of the adder to the output of the sensor using
the output signal of the adder and the detection signal of the sensor at predetermined time
intervals The system further includes an identification update unit that calculates and updates
the processing coefficient of the signal generation unit such that the processing coefficient of the
08-05-2019
5
signal generation unit becomes the calculated processing coefficient.
[0015]
A sixth invention is characterized in that, in the first invention, the signal generation unit is
constituted by a digital filter.
[0016]
According to the first aspect of the invention, when nonlinear distortion occurs from the speaker,
a signal of a nonlinear component caused by the nonlinear distortion is detected in the detection
signal, and the error signal is a signal including only the nonlinear component. Become.
Then, the signal of the error is input to the adder, so that non-linear distortion generated from
the speaker can be reduced with high accuracy.
Further, according to the present invention, since the signal of the error does not include the
signal of the linear component, it is possible to reduce the non-linear distortion without changing
the frequency characteristic of the linear component output from the speaker.
[0017]
According to the second aspect of the invention, it is possible to correct the frequency
characteristic of the acoustic output of the speaker to a predetermined frequency characteristic
while reducing non-linear distortion generated from the speaker.
[0018]
According to the third aspect of the invention, even if the transfer function from the output of the
adder to the output of the sensor changes, the frequency characteristic of the acoustic output of
the speaker can be automatically corrected to a predetermined frequency characteristic.
[0019]
According to the fourth aspect, even if the transfer function from the output of the first
08-05-2019
6
correction filter to the output of the sensor changes, the signal generation unit generates the
same signal as the linear component of the detection signal corresponding to the change. can do.
As a result, highly accurate non-linear distortion reduction processing corresponding to the
above change can be performed.
[0020]
According to the fifth aspect, even if the transfer function from the output of the adder to the
output of the sensor changes, the signal generation unit generates the same signal as the linear
component of the detection signal corresponding to the change. it can.
As a result, highly accurate non-linear distortion reduction processing corresponding to the
above change can be performed.
[0021]
According to the sixth aspect of the present invention, the signal generation unit can accurately
generate the same signal as the linear component included in the detection signal.
[0022]
First Embodiment A speaker device 1 according to a first embodiment of the present invention
will be described with reference to FIG.
FIG. 1 is a block diagram showing the configuration of the speaker device 1 according to the first
embodiment.
In FIG. 1, the speaker device 1 includes a linear signal generator 10, adders 11 and 12, a
feedback control filter 13, an amplifier 14, a speaker 15, and a sensor 16.
[0023]
08-05-2019
7
In FIG. 1, an acoustic signal such as music is input to the linear signal generator 10 and the adder
11 respectively.
In the linear signal generation unit 10, a filter coefficient that simulates the transfer function of
the linear component of the transfer function from the input of the amplifier 14 to the output of
the sensor 16 is set.
Then, the linear signal generation unit 10 simulates an acoustic signal (hereinafter, referred to as
a linear signal x (t)) based on a transfer function of the linear component from an input acoustic
signal (hereinafter, referred to as an input acoustic signal). Generate
That is, the linear signal generation unit 10 generates the same signal as the linear component
included in the detection signal y (t) detected by the sensor 16 described later. Here, it is
assumed that the linear signal generation unit 10 is configured by a digital filter. Thereby, the
transfer function of the linear component from the input of the amplifier 14 to the output of the
sensor 16 can be accurately modeled. The generated linear signal x (t) is output to the adder 12.
[0024]
Here, the setting method of the filter coefficient of the linear signal generation unit 10 will be
described. In the linear operation range of the speaker 15, for example, the impulse response
from the input of the amplifier 14 to the output of the sensor 16 is measured, and the response
value is discretized along the time axis as a coefficient of a finite impulse response (FIR) filter
There is a way to set it. However, in the process of converting an electrical signal into an acoustic
output by the speaker 15, generally, the larger the electrical signal level, the larger the non-linear
distortion. That is, the speaker 15 does not operate linearly, and the output of the speaker 15
and the measured value include a linear component and a non-linear component. Therefore,
when measuring the impulse response, the level of the impulse signal is reduced and measured
so that the non-linear distortion component included in the output of the speaker 15 becomes
sufficiently small.
[0025]
08-05-2019
8
The sensor 16 is installed near the diaphragm of the speaker 15 and detects vibration or sound
output of the speaker 15. Examples of the sensor 16 include a microphone, an acceleration
sensor, a speed sensor, or a displacement sensor. For example, when a microphone is used as the
sensor 16, the sound pressure of the speaker 15 is detected. The detection signal y (t) detected
by the sensor 16 is output to the adder 12.
[0026]
The adder 12 subtracts the linear signal x (t) from the detection signal y (t) of the sensor 16, and
outputs the subtracted signal (hereinafter referred to as an error signal e1 (t)) to the feedback
control filter 13. . Thus, in the present invention, the adder 12 serves as an error detector for
detecting an error between the detection signal y (t) and the linear signal x (t).
[0027]
Here, with reference to FIG. 2, an error signal e1 (t) when, for example, a sine wave electric signal
is input will be considered. FIG. 2 is a diagram schematically showing the waveforms of the
detection signal y (t), the linear signal x (t), and the error signal e1 (t) when a sine wave electric
signal is input to the speaker device 1. is there. Now, it is assumed that non-linear distortion
occurs at the output of the speaker 15. At this time, the sensor 16 detects a signal including a
component of the non-linear distortion (hereinafter, referred to as a non-linear component). In
the detection signal y (t) shown in FIG. 2, the portion where the waveform is clipped corresponds
to the non-linear component. As described above, the linear signal generation unit 10 is set with
filter coefficients that simulate the transfer function of the linear component of the transfer
function from the input of the amplifier 14 to the output of the sensor 16. Thus, the linear signal
x (t) generated by the linear signal generation unit 10 is a signal of only the linear component of
the detection signal y (t), and the waveform of the detection signal y (t) is not clipped. It becomes
a waveform. Therefore, since the error signal e1 (t) output from the adder 12 is a signal obtained
by subtracting the detection signal y (t) from the linear signal x (t), the detection signal y (t) as
shown in FIG. The signal contains only the non-linear component of That is, the error signal e1 (t)
is a signal including only the non-linear component in the output of the speaker 15.
[0028]
08-05-2019
9
The feedback control filter 13 is formed of, for example, a digital filter, and filter coefficients for
adjusting the gain and phase of the input error signal e1 (t) are set. The error signal e1 (t) output
from the adder 12 is output to the adder 11 after its gain and phase are appropriately adjusted
by the feedback control filter 13. Thus, by providing the feedback control filter 13, the gain and
phase of the error signal e1 (t) are adjusted to significantly reduce the non-linear component of
the detection signal y (t) of the sensor 16, and the accuracy is higher. The reduction can be
realized.
[0029]
The adder 11 subtracts the error signal e 1 (t) adjusted by the feedback control filter 13 from the
input sound signal, and outputs the subtracted signal to the amplifier 14. That is, in the adder 11,
the error signal e1 (t) is negatively fed back with respect to the input sound signal. The negative
feedback of the error signal e1 (t) including only the non-linear component in the output of the
speaker 15 can reduce the non-linear component in the output of the speaker 15. The output
signal of the adder 11 from which the non-linear component has been subtracted is
appropriately amplified in the amplifier 14 and input to the speaker 15. As a result, an acoustic
output with reduced non-linear distortion can be obtained in the speaker 15.
[0030]
It should be noted that the stability of the feedback loop composed of such adder 11 to amplifier
14 to speaker 15 to sensor 16 to adder 12 to feedback control filter 13 to adder 11 is a trade-off
between the reduction amount of the non-linear component They are increased or decreased
depending on the gain and phase characteristics of the feedback control filter 13. Therefore, the
gain and phase that satisfy both conditions are appropriately set in the feedback control filter 13.
[0031]
As described above, according to the present embodiment, the linear signal generation unit 10
generates the signal of the linear component of the detection signal y (t), and the error signal e1
(only the non-linear component of the detection signal y (t) By feeding back t), non-linear
distortion output from the speaker 15 can be reduced with high accuracy. Further, the vibration
of the speaker is detected by the sensor 16 and the above-described feedback processing is
08-05-2019
10
performed using the detection signal. As a result, the SN of the detection signal y (t) can be
increased, and air distortion in the middle and high frequencies can be accurately detected.
Further, since the signal of the linear component is not included in the error signal e1 (t) to be
fed back, in the feedback processing, non-linear distortion can be reduced without changing the
frequency characteristic of the linear component output from the speaker 15. it can. As described
above, according to the present embodiment, by providing the feedback loop configured of the
adder 11 to the amplifier 14 to the speaker 15 to the sensor 16 to the adder 12 to the feedback
control filter 13 to the adder 11, The non-linear distortion to be output can be reduced with high
accuracy and more effectively.
[0032]
In the linear signal generation unit 10 described above, it is necessary to identify in advance the
transfer function of the linear component from the input of the amplifier 14 to the output of the
sensor 16 and set a filter coefficient to simulate the transfer function. The However, the transfer
function may change depending on the installation state of the speaker 15 or the temperature
change of the speaker 15. Therefore, a method of setting a filter coefficient corresponding to the
change of the transfer function may be further adopted. In this case, as shown in FIG. 3, the
speaker device 1 further includes a switch 17, a measurement signal generator 18, and a transfer
function calculator 19. FIG. 3 is a block diagram showing a configuration for improving the
identification accuracy of the transfer function. The switch 17 switches whether the signal output
to the amplifier 14 is the output from the adder 11 or the measurement signal from the
measurement signal generator 18. The measurement signal generator 18 is an oscillator that
generates a measurement signal such as, for example, the above-described impulse signal or a
sweep signal of a sine wave. The transfer function calculating unit 19 calculates a transfer
function from the input of the amplifier 14 to the output of the sensor 16 based on the
measurement signal input from the measurement signal generator 18 and the detection signal
input from the sensor 16. However, as described above, in the process of converting an electrical
signal into an acoustic output by the speaker 15, generally, the larger the electrical signal level,
the larger the non-linear distortion. Therefore, in the measurement signal generator 18, the
measurement signal generator 18 calculates the transfer signal of the linear component so that
the non-linear distortion component included in the output of the speaker 15 becomes
sufficiently small, that is, the transfer function of the linear component is calculated. It is
necessary to set the level small.
[0033]
08-05-2019
11
The switch 17 outputs an output signal from the adder 11 during normal use of the speaker
device for audio reproduction or the like. When the installation environment of the speaker 15
changes, the input of the switch 17 is used as the measurement signal of the measurement signal
generator 18, and the transfer function calculation unit 19 calculates the transfer function from
the input of the amplifier 14 to the output of the sensor 16. Do. Information on the transfer
function is output to the linear signal generator 10 and set as a filter coefficient of the linear
signal generator 10. In the case of reproducing a signal such as audio, the input of the switch 17
is switched to the output of the adder 11 again. By the above operation, the transfer function of
the linear component with high accuracy can be set in the linear signal generation unit 10
without being affected by the installation environment of the speaker 15.
[0034]
Second Embodiment A speaker device 2 according to a second embodiment of the present
invention will be described with reference to FIG. FIG. 4 is a block diagram showing the
configuration of the speaker device 2 according to the second embodiment. The speaker device 2
according to the present embodiment is a speaker device that simultaneously performs the
nonlinear distortion reduction processing described in the first embodiment and the correction
processing of the frequency characteristic of the linear component output from the speaker. In
order to realize this correction processing, the speaker device 2 further includes adaptive filters
20a and 20b, a coefficient updating unit 21, a reference signal generating unit 22, a target
characteristic filter 23, and an adder 24 in addition to those of the first embodiment. Prepare. In
FIG. 4, the linear signal generation unit 10, the adders 11 and 12, the feedback control filter 13,
the amplifier 14, the speaker 15, and the sensor 16 have the same functions as those of the first
embodiment described above. The same reference numerals are given and the description is
omitted.
[0035]
First, the correction process of the frequency characteristic of the linear component output from
the speaker will be described. An acoustic signal such as music passes through the adder 11 and
is input to the adaptive filter 20a, the reference signal generator 22, and the target characteristic
filter 23, respectively.
[0036]
08-05-2019
12
In the reference signal generation unit 22, a filter coefficient that simulates a transfer function
from the input of the amplifier 14 to the output of the sensor 16 is set. The filter coefficient may
simulate a transfer function of a linear component from the input of the amplifier 14 to the
output of the sensor 16 or may be a filter coefficient simulating a transfer function of all
components (linear component and non-linear component) It may be Then, the reference signal
generation unit 22 processes the signal input from the adder 11 based on the set filter
coefficient, and outputs the processed signal to the coefficient update unit 21. The signal
processed by the reference signal generation unit 22 is a signal based on the transfer function of
the speaker 15, and plays a role as a reference signal of a filtered X-LMS algorithm in the
coefficient update unit 21 described later.
[0037]
In the target characteristic filter 23, a filter coefficient that simulates a target characteristic to be
realized at the output of the speaker 15 is set. For example, as shown in FIG. 5, in the case where
the target characteristic is such that the frequency characteristic becomes flat at a frequency
higher than the primary resonance frequency f0 of the speaker 15, an IIR filter for realizing a
high pass filter is used as the target characteristic filter 23 It should be designed as FIG. 5 is a
diagram showing an example of the target characteristic. The target characteristic filter 23
processes the signal input from the adder 11 based on the set filter coefficient, and outputs the
processed signal to the adder 24. The adder 24 subtracts the detection signal y (t) of the sensor
16 from the signal output from the target characteristic filter 23 and outputs the subtracted
signal (hereinafter referred to as an error signal e2 (t)) to the coefficient updating unit 21.
Output.
[0038]
The coefficient update unit 21 calculates filter coefficients at predetermined time intervals based
on the outputs of the reference signal generation unit 22 and the adder 24 so that the error
signal e2 (t) becomes smaller. The coefficient update unit 21 calculates filter coefficients at
predetermined time intervals using a known filtered X-LMS algorithm. Then, the coefficient
updating unit 21 updates the filter coefficients of the adaptive filters 20a and 20b to the
calculated filter coefficients. The time interval to be updated is, for example, the same interval as
the predetermined time interval for which the filter coefficient is calculated.
08-05-2019
13
[0039]
Hereinafter, the process in the coefficient update unit 21 will be specifically described. The
coefficient update unit 21 inputs the signal generated by the reference signal generation unit 22
as a reference signal of the filtered X-LMS algorithm. This is because the signal processed by the
reference signal generation unit 22 is a signal based on the transfer function of the speaker 15.
The coefficient updating unit 21 calculates filter coefficients using a filtered X-LMS algorithm so
that the error signal e2 (t) output from the adder 24 is minimized.
[0040]
Here, the above-mentioned filtered X-LMS algorithm is a known algorithm disclosed in, for
example, “Sound system and digital processing” (Oga, J. et al., The Institute of Electronics,
Information and Communication Engineers, published in March, 1995), Specifically, it is
expressed by the following equation (1). w (k + 1) = w (k) + 2μ * e2 (k) * x (k) (1) In equation (1),
when k represents sampling times indicating predetermined time intervals on a time series Is the
number of each sampling time. Further, w (k) is a filter coefficient vector of the adaptive filters
20a and 20b at the sampling time k, μ is a parameter that defines the amount of adaptive
update, e2 (k) is an error signal at the sampling time k input from the adder 24, x (k) is a
reference signal vector at sampling time k input from the reference signal generation unit 22.
From the above equation (1), it can be seen that the filter coefficient vector w is calculated at
each sampling time.
[0041]
The adaptive filter 20 a processes the signal output from the adder 11 based on the filter
coefficient updated by the coefficient update unit 21. Further, the filter coefficient updated by
the coefficient updating unit 21 is a coefficient when the error signal e2 (t) output from the
adder 24 becomes minimum as described above. Therefore, the transfer function from the input
of the adaptive filter 20a to the output of the sensor 16 is corrected to the transfer function of
the target characteristic (for example, see FIG. 5) set by the target characteristic filter 23.
[0042]
08-05-2019
14
Thus, in the present invention, the coefficient updating unit 21, the reference signal generating
unit 22, the target characteristic filter 23, and the adder 24 have a role as an adaptive updating
unit that adaptively updates the filter coefficients of the adaptive filters 20a and 20b. To Then,
even if the transfer function from the output of the adder 11 to the output of the sensor 16
changes, the adaptive updating unit adaptively updates the filter coefficients of the adaptive
filters 20a and 20b, and the acoustic output characteristic of the speaker 15 Can be
automatically corrected to the target characteristic set in the target characteristic filter 23.
[0043]
Next, non-linear distortion reduction processing will be described. In FIG. 4, an acoustic signal
such as music is input to the adaptive filter 20 b and the adder 11 respectively. The adaptive
filter 20b is the same filter as the adaptive filter 20a, and the same filter coefficients as the
adaptive filter 20a are set in the adaptive filter 20b by the coefficient updating unit 21. The
signal processed by the adaptive filter 20 b is output to the linear signal generator 10.
Hereinafter, the non-linear distortion reduction processing by the feedback loop constituted by
the adder 11 to the amplifier 14 to the speaker 15 to the sensor 16 to the adder 12 to the
feedback control filter 13 to the adder 11 is the same as the first embodiment described above.
Therefore, the explanation is omitted.
[0044]
Here, the role of the adaptive filter 20b will be described. The linear signal x (t) generated in the
linear signal generator 10 needs to be the same as the linear component of the detection signal y
(t) detected by the sensor 16 as described above. Further, in the present embodiment, the
adaptive filter 20 a is provided at the front stage of the amplifier 14. Therefore, in order to take
into consideration the variation of the transfer function by the adaptive filter 20a, the adaptive
filter 20b having the same filter coefficient as that of the adaptive filter 20a is provided at the
front stage of the linear signal generation unit 10. As a result, the linear signal x (t) and the linear
component of the detection signal e1 (t) can be made identical, and highly accurate reduction of
nonlinear distortion can be achieved.
[0045]
Moreover, with reference to FIG. 6 and FIG. 7, the processing effect of the speaker apparatus 2
08-05-2019
15
which concerns on this embodiment is demonstrated. FIG. 6 shows a linear component, a secondorder nonlinear component, and the like output from the speaker 15 when the diaphragm
acceleration of the speaker 15 is not fed back to the adder 11 through the detection signal of the
sensor 16 (when the reduction processing of nonlinear distortion is not performed) It is a figure
which shows the calculation result of and the 3rd-order nonlinear component. FIG. 7 shows a
linear component, a second-order nonlinear component, and a second-order nonlinear
component output from the speaker 15 when the diaphragm acceleration of the speaker 15 is
fed back to the adder 11 through the detection signal of the sensor 16 (when nonlinear
distortion reduction processing is performed) It is a figure which shows the calculation result of
and the 3rd-order nonlinear component. 6 and 7, the frequency range of calculation is 20 to 200
Hz, and the transfer function of the feedback control filter 13 is 1. As is apparent from FIGS. 6
and 7, the frequency characteristic of the linear component does not change even if the abovedescribed nonlinear distortion reduction processing is performed. This is because the error signal
e1 (t) is a signal of only the non-linear component of the detection signal y (t) of the sensor 16,
and does not include the linear component of the detection signal y (t). Further, as is clear from
FIG. 7, it can be confirmed that the second-order and third-order nonlinear components are
significantly reduced by performing the above-mentioned nonlinear distortion reduction
processing. That is, by the above-described reduction processing of non-linear distortion by the
feedback loop, it is possible to reduce non-linear distortion with high accuracy without impairing
the correction effect of the frequency characteristic of the linear component by the adaptive filter
20a.
[0046]
As described above, according to the speaker device 2 according to the present embodiment, it is
possible to correct the frequency characteristic of the linear component output from the speaker
to a desired characteristic while reducing non-linear distortion with high accuracy.
[0047]
In the linear signal generation unit 10 and the reference signal generation unit 22 shown in FIG.
4, a transfer function from the input of the amplifier 14 to the output of the sensor 16 is
identified in advance, and a filter coefficient that simulates the transfer function is It was
necessary to set each.
However, the transfer function may change depending on the installation state of the speaker 15
or the temperature change of the speaker 15. Therefore, as in the first embodiment described
above, the identification accuracy of the transfer function may be improved by the configuration
08-05-2019
16
as shown in FIG. FIG. 8 is a block diagram showing a configuration for improving the
identification accuracy of the transfer function in the speaker device 2. In this case, as shown in
FIG. 8, the speaker device 2 further includes a switch 17, a measurement signal generator 18,
and a transfer function calculator 19. The switch 17 switches whether the signal output to the
amplifier 14 is the output from the adaptive filter 20 a or the measurement signal from the
measurement signal generator 18. The measurement signal generator 18 is an oscillator that
generates a measurement signal such as, for example, the above-described impulse signal or a
sweep signal of a sine wave. The transfer function calculating unit 19 calculates a transfer
function from the input of the amplifier 14 to the output of the sensor 16 based on the
measurement signal input from the measurement signal generator 18 and the detection signal
input from the sensor 16. However, as in the first embodiment described above, in the process of
converting an electrical signal into an acoustic output by the speaker 15, generally, the larger the
electrical signal level, the larger the non-linear distortion. Therefore, in the measurement signal
generator 18, the level of the measurement signal is set small so that the non-linear distortion
component included in the output of the speaker 15 becomes sufficiently small.
[0048]
During normal use of the speaker device for audio reproduction or the like, the switch 17 outputs
an output signal from the adaptive filter 20a. When the installation environment of the speaker
15 changes, the input of the switch 17 is used as the measurement signal of the measurement
signal generator 18, and the transfer function calculation unit 19 calculates the transfer function
from the input of the amplifier 14 to the output of the sensor 16. Do. Information on the transfer
function is output to the linear signal generation unit 10 and the reference signal generation unit
22, and is set as filter coefficients of the linear signal generation unit 10 and the reference signal
generation unit 22, respectively. In the case of reproducing a signal such as audio, the input of
the switch 17 is switched to the output of the adder 11 again. By the above operation, the
transfer function of the linear component with high accuracy can be set in the linear signal
generation unit 10 and the reference signal generation unit 22 without being affected by the
installation environment of the speaker 15.
[0049]
Moreover, in the speaker apparatus 2 shown in FIG. 4, although the correction | amendment of
the frequency characteristic of the linear component output from the speaker 15 is performed
using the adaptive filter 20a, it is not limited to this. For example, the filter coefficient may be a
fixed filter, and may be configured by an analog circuit. Thereby, the coefficient updating unit 21
08-05-2019
17
for updating the filter coefficient of the adaptive filter 20a, the reference signal generating unit
22, the target characteristic filter 23, and the adder 24 can be omitted. In this case, in order to
make the linear signal generated by the linear signal generation unit 10 equal to the linear
component of the detection signal detected by the sensor 16, the adaptive filter 20b is also fixed
with the same filter coefficient as the adaptive filter 20a. Need to be a filter. Further, by fixing the
filter coefficients, the adders 11 and 12 can also be configured by analog circuits.
[0050]
Third Embodiment A loudspeaker device 3 according to a third embodiment of the present
invention will be described with reference to FIG. FIG. 9 is a block diagram showing the
configuration of the speaker device 3 according to the third embodiment. The speaker device 3
according to the present embodiment is a speaker device that realizes processing based on a
transfer function that is always identified with high accuracy in the above-described speaker
devices 1 and 2. In FIG. 9, the apparatus which implement | achieved the process based on the
transfer function always identified with high precision in the speaker apparatus 2 mentioned
above as the speaker apparatus 3 is shown. In FIG. 9, the speaker device 3 newly includes an
adaptive identification unit 30, a coefficient updating unit 31, and an adder 32 with respect to
the speaker device 2. Hereinafter, the adaptive identification unit 30, the coefficient update unit
31, and the adder 32 will be mainly described.
[0051]
In FIG. 9, an input acoustic signal such as a music signal processed by the adaptive filter 20 a is
input to the amplifier 14 and is also input to the adaptive identification unit 30 and the
coefficient update unit 31. The signal processed by the adaptive identification unit 30 is input to
the adder 32. The adder 32 subtracts the detection signal y (t) of the sensor 16 from the signal
processed by the adaptive identification unit 30, and the signal (hereinafter referred to as an
error signal e3 (t)) obtained by the subtraction is a coefficient update unit 31. Output to
[0052]
The coefficient updating unit 31 uses the signal input from the adaptive filter 20a as a reference
signal to minimize the error signal e3 (t) based on a known adaptive algorithm (for example, the
above-described filtered X-LMS algorithm, etc.) Filter coefficients are calculated at predetermined
08-05-2019
18
time intervals. Then, the coefficient updating unit 31 updates the calculated filter coefficient as a
filter coefficient of the adaptive identification unit 30 at predetermined time intervals. That is,
when the error signal e3 (t) input to the coefficient updating unit 31 becomes sufficiently small,
the filter coefficient of the adaptive identification unit 30 simulates the transfer function of the
linear component from the input of the amplifier 14 to the output of the sensor 16 It will be
expressed as The filter coefficients of the linear signal generator 10 and the reference signal
generator 22 are updated to be the filter coefficients identified by the adaptive identification unit
30.
[0053]
As described above, in the present invention, the adaptive identification unit 30, the coefficient
update unit 31, and the adder 32 adaptively identify and update the filter coefficients of the
linear signal generation unit 10 and the reference signal generation unit 22. Play a role as Then,
even if the transfer function from the input of the amplifier 14 to the output of the sensor 16
changes, the identification update unit adaptively updates the filter coefficients of the linear
signal generator 10 and the reference signal generator 22. Thereby, the linear signal generation
unit 10 and the reference signal generation unit 22 can always generate a signal with high
accuracy.
[0054]
In the linear signal generation unit 10, as described above, it is necessary to set a filter coefficient
that simulates the transfer function of the linear component. However, depending on the
magnitude of the input acoustic signal, the detection signal y (t) of the sensor 16 may include a
non-linear component. At this time, the error signal e3 (t) also contains a non-linear component.
Then, in the adaptive identification unit 30, a transfer function based on the error signal e3 (t)
including the non-linear component is identified. As a result, the filter coefficient identified in the
adaptive identification unit 30 simulates the transfer function of a linear component having a
slight level difference ΔH than the linear component of the actual transfer function. As a result,
when the filter coefficient is set in the linear signal generation unit 10, the linear signal
generated by the linear signal generation unit 10 is a signal whose level is higher by the level
difference ΔH with respect to the linear component of the detection signal of the sensor 16. It
becomes. The error signal e1 (t) includes a linear component corresponding to the level
difference ΔH. As a result, the gain of the signal of the feedback loop is increased, and the
frequency characteristic of the linear component output from the speaker 15 corrected by the
adaptive filter 20a is changed.
08-05-2019
19
[0055]
However, the level difference ΔH is a small level compared to the output level from the speaker
15. Along with that, the change of the above-mentioned frequency characteristic becomes slight
and does not become a big problem in hearing. Therefore, even if the filter coefficient indicating
the transfer function identified by the adaptive identification unit 30 is substituted as the filter
coefficient of the linear signal generation unit 10, the speaker device 3 according to the present
embodiment exerts a certain reduction effect of non-linear distortion. can do.
[0056]
In the above, the apparatus for realizing the processing based on the transfer function constantly
identified with high accuracy in the speaker apparatus 2 has been described as the speaker
apparatus 3, but the apparatus may be applied to the speaker apparatus 1.
[0057]
As described above, according to the speaker device 3 according to the present embodiment, by
providing the adaptive identification unit 30, the coefficient updating unit 31, and the adder 32,
processing is always performed based on the transfer function identified with high accuracy. be
able to.
That is, it is possible to obtain a highly accurate reduction effect of non-linear distortion and a
correction effect of the frequency characteristics of the speaker 15 without being affected by a
change in the installation environment of the speaker 15.
[0058]
The speaker device according to the present invention is also applied to a speaker device, an
audio device, and the like that can reduce nonlinear distortion generated from the speaker with
high accuracy.
[0059]
A block diagram showing the configuration of the speaker device 1 according to the first
08-05-2019
20
embodiment A detection signal y (t), a linear signal x (t), and an error signal e1 (t) when a sine
wave electrical signal is input to the speaker device 1 The block diagram showing the
configuration for improving the identification accuracy of the figure transfer function
schematically showing each waveform of) The block diagram showing the configuration of the
speaker device 2 according to the second embodiment The figure showing an example of the
target characteristic of the speaker 15 Diagram showing calculation results of each component
outputted from the speaker 15 when the diaphragm acceleration is not fed back to the adder 11
through the detection signal of the sensor 16. The diaphragm acceleration of the speaker 15 is
fed back to the adder 11 through the detection signal of the sensor 16 Shows the calculation
result of each component outputted from the speaker 15 in the case of the case where the
identification accuracy of the transfer function in the speaker device 2 is improved The block
diagram showing the configuration The block diagram showing the configuration of the speaker
device 3 according to the third embodiment The block diagram showing the configuration of the
conventional speaker device reducing nonlinear distortion The block showing the configuration
of the conventional speaker device correcting the frequency characteristics Diagram showing the
amplitude frequency characteristic of the speaker 98 before the figure correction processing
Diagram showing the amplitude frequency characteristic of the speaker 98 after the figure
correction processing
Explanation of sign
[0060]
Reference Signs List 1 to 3 speaker device 10 linear signal generation unit 11, 12, 24, 32 adder
13 feedback control filter 14 amplifier 15 speaker 16 sensor 17 switch 18 measurement signal
generator 19 transfer function calculation unit 20 adaptive filter 21 coefficient update unit 22
Reference signal generation unit 23 Target characteristic filter 30 Adaptive identification unit 31
Coefficient update unit
08-05-2019
21
Документ
Категория
Без категории
Просмотров
0
Размер файла
36 Кб
Теги
jp2007081815
1/--страниц
Пожаловаться на содержимое документа