JPH06269084

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 JPH06269084
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
device for reducing wind noise contained in an audio signal collected by a microphone in an
environment exposed to an air flow such as natural wind, exhalation and air conditioning.
[0002]
2. Description of the Related Art A microphone converts a change in sound pressure caused by a
sound wave into mechanical vibration of a diaphragm and operates an electroacoustic conversion
system based on the mechanical vibration to obtain an electrical signal. There are many.
Therefore, when mechanical sound is given to the diaphragm due to some factor other than the
desired voice when the desired voice is picked up by the microphone, this becomes noise for the
desired voice. In this case, if the above factor is wind, wind noise (hereinafter referred to as wind
noise) is generated.
[0003]
As a method of reducing wind noise generated in a microphone, (1) use of a windscreen
(windshield) (2) adoption of a configuration showing nondirectionality in the bass region (3)
electrical / acoustic high pass filter (cutoff The use of fixed frequency) is often used
conventionally.
03-05-2019
1
[0004]
However, according to the method (1), in general, the wind noise becomes smaller as the outside
dimension of the windscreen is larger and the distance between the microphone element and the
inner wall of the windscreen is larger. So, if you try to reduce wind noise sufficiently, there is a
problem that you have to prepare a big windscreen.
This problem is a major obstacle to downsizing and portability of the device.
[0005]
In addition, the method (2) is used because the wind noise is lower when the directivity of the
microphone is non-directional rather than directional. However, the wind noise reduction effect
by this method is not so large, and in addition, the wind noise does not reach a sufficiently low
level due to the influence of the case when actually configuring the microphone device.
[0006]
Moreover, although the method of using the electrical / acoustic high-pass filter of the fixed cutoff frequency of (3) above is effective because wind noise is noise mainly in the bass region, it
reduces wind noise. At the same time, the bass range of the desired voice is always reduced at
the same time. Therefore, even when there is almost no wind noise or even when the wind noise
level is low, the sound collection quality is lowered.
[0007]
The present invention basically uses the method (3) above, but it is possible to eliminate the
above-mentioned problems and to obtain as good sound collection quality as possible while
reducing wind noise. It is an object of the present invention to provide a wind noise reduction
device.
[0008]
SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the wind noise
reduction device according to the present invention comprises first and second microphones
03-05-2019
2
disposed in proximity to each other in correspondence with reference numerals of embodiments
described later. Wind noise detection means 15 for detecting a wind noise component by
calculating the difference between the elements 11 and 12 and the outputs of the first and
second microphone elements, and the output signal of the first and / or second microphone
elements And a control means 19 for controlling the cut-off frequency of the high-pass filter
based on the output of the wind noise detection means, the wind noise being reduced from the
high-pass filter A voice output signal is obtained.
[0009]
FIG. 5 shows the wind noise spectrum of the microphone. As can be seen from FIG. 5, the wind
noise is noise mainly composed of several hundred Hz or less.
On the other hand, FIG. 6 shows the spectrum of the speech signal, which is the spectrum of
female speech.
As apparent from FIG. 6, the audio signal is distributed at about 100 Hz or more.
[0010]
In the wind noise reduction device according to the present invention of the above configuration,
the cutoff frequency of the high pass filter for extracting the output sound signal is controlled in
accordance with the level of the wind noise.
[0011]
That is, when the wind noise level is high, the cutoff frequency of the high-pass filter is set to a
relatively high frequency of several hundred Hz, and the wind noise is removed.
When the wind noise level is low, it is not necessary to reduce the wind noise so much, so the
cutoff frequency of the high-pass filter is set to a low frequency inversely proportional to the
wind noise level of several hundred Hz or less. As much as possible, the low-frequency
component of speech is obtained as an output speech signal.
03-05-2019
3
[0012]
DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of the wind noise reduction
device according to the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing an embodiment in which a wind noise reduction device
according to the present invention is applied to a microphone device.
[0013]
In the example of FIG. 1, two microphone elements 11 and 12 are arranged close to each other.
The microphone elements 11 and 12 have the same characteristics in this example, and for
example, a nondirectional microphone unit is used.
[0014]
Then, the output of one of the microphone elements 11 is converted into a digital signal by the A
/ D converter 13, and the digital signal is supplied to the wind noise reduction high pass filter 16
whose cutoff frequency is variable. Then, the output signal of the high pass filter 16 is supplied
to the D / A converter 17 to be converted back to an analog signal and is led out to the output
terminal 20. The high pass filter 16 in this example is comprised of a first order IIR digital filter.
[0015]
FIG. 2 is an example of the high pass filter 16 configured using this first order IIR digital filter.
This circuit comprises delay circuits 161 and 162 each having a delay time of a unit sample,
weighting circuits 163, 164 and 165 for filter coefficients, and addition circuits 166 and 167. a0,
a1, b1 are filter coefficients, and the cutoff frequency of the high pass filter 16 is changed by
changing the filter coefficients a0, a1, b1.
[0016]
03-05-2019
4
Then, the output of the other microphone element 12 is converted into a digital signal by the A /
D converter 14, and the digital signal is supplied to the subtraction circuit 15. In the subtraction
circuit 15, the difference between the digital signal of the output of the microphone element 11
from the A / D converter 13 and the digital signal of the output of the other microphone element
12 is obtained.
[0017]
When sound is collected by the apparatus shown in FIG. 1 in an environment where wind noise
occurs, the outputs of the microphone elements 11 and 12 include wind noise in the collected
sound signal. As described above, since the two microphone elements 11 and 12 are arranged
close to each other, the sound is collected by the two microphone elements 11 and 12 in a highly
correlated state. On the other hand, since the wind noise generated in the microphone element
11 and the microphone element 12 by the wind is unique to each microphone element, there is
no correlation between the wind noise of both microphone elements.
[0018]
Therefore, when the difference operation of the outputs of the two microphone elements 11 and
12 is performed in the subtractor circuit 15, the audio signal is canceled, and the component of
only wind noise is obtained from the subtractor circuit 15.
[0019]
The wind noise component from the subtraction circuit 15 is supplied to the averaging circuit
18.
The averaging circuit 18 is formed of a circuit that functions as a low pass filter or a low pass
filter. This makes the degree of change of the cut-off frequency of the wind noise reduction highpass filter 15 moderate by averaging the wind noise. Thereby, it is possible to suppress the
occurrence of distortion caused by high frequency components of noise and wind noise.
[0020]
03-05-2019
5
The output of the averaging circuit 18 is supplied to the control circuit 19. The control circuit 19
controls the cutoff frequency of the high pass filter 16 according to the output level of the
averaging circuit 18, that is, the amount (power) of wind noise. That is, the control circuit 19
obtains the values of the filter coefficients a0, a1, and b1 of the digital IIR filter according to the
algorithm represented by the coefficient updating formulas (1), (2), and (3) Control the cutoff
frequency according to the power of.
[0021]
a0 = -0.998692 + μ × X × X (1) a1 = 0.998692−μ × X × X (2) b1 = −0.997385 + 2 μ
××× (3) where μ is a constant and X is averaging It is the output signal level of the circuit 18.
[0022]
Therefore, when the wind noise power X is zero, the above equations (1) to (3) become a0 =
−0.998692 a1 = 0.998692 b1 = −0.997385.
The characteristic of the filter constituted by this value is a high pass filter having a cutoff
frequency of 20 Hz as shown in FIG.
[0023]
Then, when the wind noise comes to have a certain value, when the wind noise amount X is large,
the values of the coefficients a0, a1, and b1 change according to the amount X so that the cutoff
frequency becomes higher. . That is, when the amount of wind noise is small, the cutoff
frequency of the high pass filter 16 is low, and the low frequency component of the collected
voice of the microphone is also obtained at the output. On the other hand, when the amount of
wind noise is large, the cut-off frequency of the high-pass filter 16 is set to several hundreds Hz,
which mainly includes wind noise, and the wind noise is sufficiently reduced.
[0024]
FIG. 4 shows the values of the above-mentioned coefficients a0, a1 and b1 at several cut-off
frequencies of the high-pass filter 16 consisting of a first-order IIR digital filter as a table.
03-05-2019
6
[0025]
FIGS. 7 to 12 show experimentally the effects of the device of the above example, and the wind
noise and the female voice shown in FIG. 5 and FIG. 6 are simultaneously picked up as the
collected input speech. In the case of only the wind noise of FIG. 5, three cases of only the female
voice of FIG. 6 are prepared, and the spectra before and after the above-described wind noise
reduction processing are compared with each other.
[0026]
7 and 8 show the spectra before and after the above-described wind noise reduction processing
when wind noise and female voice are simultaneously picked up, respectively. Later, it turns out
that it has been improved.
[0027]
Also, FIG. 9 and FIG. 10 show the spectra before and after the above-described wind noise
reduction processing in the case of only wind noise, respectively, and from this, after processing,
it is improved by about 15 dB in the low frequency range I understand that.
[0028]
11 and 12 show the spectra before and after the above-described wind noise reduction
processing in the case of only female voice, respectively, from which it can be seen that there is
almost no difference between before and after processing, and wind noise When it is not, it is
understood that the bass component of the collected voice can be obtained without being
affected by the high pass filter.
[0029]
As the high pass filter 16, a second-order IIR digital filter 200 as shown in FIG. 13 can also be
used.
It comprises delay circuits 201, 202, 203 and 204 each having a delay time for a unit sample,
weighting circuits 206, 207, 208, 209 and 210 for filter coefficients, and addition circuits 211,
212 and 213. .
a0, a1, a2, b1, b2 are filter coefficients, and the cutoff frequency of the filter 200 is changed by
03-05-2019
7
changing the filter coefficients a0, a1, a2, b1, b2.
[0030]
In this case, the values of the filter coefficients a0, a1, a2, b1 and b2 are determined according to
the algorithm represented by the coefficient updating equations (4), (5), (6), (7) and (8) As
described above, the cutoff frequency is controlled according to the power of wind noise.
[0031]
a0 = 0.990786-μ × ε × X (4) a1 = −1.981573 + 2 μ × ε × X (5) a2 = 0.990786−μ × ε
× X (6) b1 = −1.981488 + 2 μ × ε × X 7) b2 = 0.981658-2 μ × ε × X (8) where μ is a
constant, X is an output signal level of the averaging circuit 18, and ε is a system output.
[0032]
FIG. 14 shows the values of the above-mentioned coefficients a0, a1, a2, b1 and b2 at several
cutoff frequencies of this second-order IIR digital filter 200 as a table.
[0033]
Here, the features of the first and second order IIR digital filters and equations (1) to (8) showing
the algorithm of updating the filter coefficients will be described.
[0034]
From the tables of FIG. 4 and FIG. 14, it can be seen that, for example, the change amounts
.DELTA.a0, .DELTA.a1, .DELTA.b1 of the respective coefficients when the cutoff frequency
changes from 100 Hz to 200 Hz have the following regularity.
Similarly, in the case of the second order, .DELTA.a0 =-. DELTA.a1 .apprxeq.2.DELTA.b1,
.DELTA.a0, .DELTA.a1, .DELTA.a2, .DELTA.b1, .DELTA.b2 have the following regularity.
Δa 0 = Δa 2 −2−2 × Δa 1 −2−2 × Δb 1 Δ 2 × Δb 2 The above relationship holds also
with other frequencies.
03-05-2019
8
Then, the algorithm of the coefficient update performed using these relationships is Formula (1)(8).
[0035]
The above example is the case where any one of a first-order IIR filter or a second-order IIR filter
is used as the high-pass filter 16, but a plurality of digital filters having different orders are
provided. By controlling the selection of one digital filter or the combination of a plurality of
digital filters that substantially functions as the above, it is also possible to easily obtain
appropriate blocking characteristics.
[0036]
FIG. 15 is a block diagram showing the configuration of one embodiment in that case.
The same parts as those in the above-described example are denoted by the same reference
numerals.
In this example, the high pass filter 16 includes a first digital filter 21, a second digital filter 22,
and a switching circuit 23.
Then, the average power of wind noise from the averaging circuit 18 is supplied to the control
circuit 24, and the switching circuit 23 is switched by the output of the control circuit 24.
In this example, the first digital filter 21 uses a first-order IIR filter, and the second digital filter
22 uses a second-order IIR filter.
[0037]
In this case, when the amount of wind noise is zero or very small, the switching circuit 23 is
switched to the input end A, and in the high pass filter 16, the first and second digital filters 21
and 22 are substantially bypassed. The output signal of the / D converter 13 is supplied to the D
/ A converter 17 as it is.
03-05-2019
9
[0038]
In the case where the wind noise has a certain amount and is less than or equal to a
predetermined amount, the switching circuit 23 is switched to the input terminal B, and the
output signal of the A / D converter 13 is the first. The digital signal is supplied to the D / A
converter 17 only through the digital filter 21.
When the amount of wind noise increases and exceeds the predetermined amount, the switching
circuit 23 is switched to the input terminal C, and the second digital filter 22 is cascaded to the
first digital filter 21. It becomes a state.
The output of the A / D converter 13 is supplied to the D / A converter 17 through the two
digital filters 21 and 22.
[0039]
In the example of FIG. 15, the first digital filter 21 may be configured by a second-order IIR filter,
and the second digital filter 22 may be configured by a first-order IIR filter. Also, two or more
digital filters may be used in combination. Furthermore, the filter coefficients of the digital filters
21 and 22 may be controlled together by the control circuit 24.
[0040]
In the above example, the cutoff frequency of the high pass filter 16 is controlled according to
only the wind noise power level, but the output power of the desired voice obtained by passing
the output of the A / D converter 13 through the high pass filter The cut-off frequency of the
high pass filter 16 may be controlled depending on the frequency. In addition, by controlling the
cutoff frequency of the high-pass filter while considering both the amount of desired voice and
the amount of wind noise, it is possible to obtain a better quality output audio signal.
[0041]
03-05-2019
10
FIG. 16 is an embodiment of the present invention for controlling the cutoff frequency of the
high pass filter while considering both the amount of desired speech and the amount of wind
noise. That is, in the example of FIG. 16, the output signal of the A / D converter 13 is supplied to
the high pass filter 16 and to another high pass filter 31 mainly for extracting only the audio
signal. The output of the high pass filter 31 is supplied to a level ratio detection circuit 32.
Further, the component of the wind noise from the subtraction circuit 15 is supplied to the level
ratio detection circuit 32.
[0042]
The level ratio detection circuit 32 detects the ratio between the level of the audio signal from
the high pass filter 31 and the level of the wind noise signal from the subtraction circuit 15, and
the detected output is supplied to the control circuit 33. Then, the control circuit 33 controls the
cut-off frequency of the high pass filter 16 in accordance with the input level ratio. The high pass
filter 16 may have any of the configurations described above.
[0043]
In this example, the cutoff frequency of the high-pass filter is determined by the level ratio of the
audio signal and the wind noise signal. For example, even if the wind noise is large to a certain
extent, the cutoff frequency is selected when the desired audio level is large. By not making it so
high, the influence on the desired voice can be reduced.
[0044]
An audio signal obtained by passing the sum of the output of the microphone 11 and the output
of the microphone 12 through a high pass filter may be used instead of the output of the high
pass filter 31 described above.
[0045]
The above example is an embodiment in which the present invention is applied to a microphone
device, but the present invention can also be applied to a recording and reproducing device.
FIG. 17 is a block diagram of an embodiment of a tape recording and reproducing apparatus to
which the present invention is applied.
03-05-2019
11
[0046]
That is, in this example, the audio signals collected by the first and second microphones 11 and
12 are supplied to the recording heads 45 and 46 through the recording circuits 41 and 42 and
the recording amplifiers 43 and 44, respectively. Then, two channels are recorded on the
magnetic tape 47.
[0047]
The two-channel signals recorded on the tape 47 in this manner are picked up from the tape 47
by the reproducing heads 51 and 52, respectively.
Then, reproduction signals from the reproduction heads 51 and 52 are supplied to A / D
converters 55 and 56 through reproduction amplifiers 53 and 54 to be converted into digital
signals.
Then, the output of the A / D converter 55 is supplied to the D / A converter 58 through the high
pass filter 57 similar to the above-described high pass filter 16, and is converted back to an
analog signal and output to the output terminal 59.
[0048]
Further, the difference between the outputs of the A / D converters 55 and 56 is obtained in the
subtraction circuit 60, and the wind noise component contained in the reproduction signal is
extracted. The wind noise signal is supplied to the control circuit 62 through the averaging
circuit 61. The control circuit 62 controls the cut-off frequency of the high-pass filter 57 so that
the cut-off frequency increases as the amount of wind noise increases, in the same manner as the
above-described example. Also in this example, the cutoff frequency of the high pass filter 57
may be controlled in accordance with the level ratio between the output of the A / D converter
55 and the output of the subtraction circuit 60.
[0049]
03-05-2019
12
As described above, according to the example of FIG. 17, the wind noise component contained in
the reproduction signal can be removed to obtain a good reproduction sound signal. At the time
of recording, the audio signals from the microphones 11 and 12 are not recorded in two
channels, but the output of one or both microphones and the difference between the outputs of
both microphones, that is, the wind noise component is recorded. It is also good. Furthermore,
wind noise may be recorded on a recording medium such as a tape together with an audio signal
as well as the output of a sensor that detects the flow rate and flow rate of wind.
[0050]
The above is the case where the high pass filter is configured by a digital filter, but the high pass
filter with variable cutoff frequency may be configured as an analog filter.
[0051]
FIG. 18 shows an embodiment in which the high-pass filter with variable cutoff frequency is
constituted by an analog filter, in which the present invention is applied to the above-mentioned
microphone device.
In this example, the output signal of the microphone 11 is supplied to a high pass filter 71 of an
analog configuration.
[0052]
The high pass filter 71 includes a capacitor 72 and a transistor 73 as an example of a variable
impedance element. Then, a wind noise signal from the subtraction circuit 74 for obtaining the
difference between the output of the microphone 11 and the output of the microphone 12 is
supplied to the base of the transistor 73 via the level detection circuit 75 as a control signal
formation circuit.
[0053]
When the amount of wind noise is a value close to zero, the base voltage of the transistor 73
03-05-2019
13
becomes zero or low level, so the transistor 73 becomes nonconductive. Therefore, the high pass
filter 71 has a cutoff frequency of 0 or a very low frequency, and the output sound signal of the
microphone 11 is derived to the output terminal 76 with almost the same frequency
characteristic.
[0054]
When the wind noise level increases, the base voltage of the transistor 73 increases according to
the size of the wind noise, so the impedance of the transistor 73 decreases, whereby the cutoff
frequency of the high-pass filter 73 It changes to the high frequency side according to the size.
[0055]
Needless to say, a photocoupler or the like can be used instead of the transistor 73 as the
variable impedance element.
Further, even in the case of using an analog high pass filter, a plurality of analog filters can be
combined to obtain a desired filter characteristic as in the example of FIG.
[0056]
In the above example, wind noise is detected as the difference between the output signals of the
two microphones, but the magnitude of wind noise is detected using a sensor that detects the
flow rate or flow rate of the wind. May be
[0057]
In addition, since two microphone elements can be allocated to the left and right channels, even
when configuring a stereo microphone, it can be realized without increasing the number of
microphone elements.
In that case, a high pass filter is inserted for each of the two microphone outputs, and the cutoff
frequency of each high pass filter is controlled according to the power of wind noise, the power
of audio signal, and the ratio of both. Make it
03-05-2019
14
[0058]
In addition, since the wind noise reduction operation is automatic, there is an advantage that the
user can concentrate on monitoring of a monitor image or the like while shooting with another
operation, for example, a camera integrated VTR.
[0059]
In the above-mentioned example, although two microphone elements 11 and 12 used a
nondirectional microphone, these microphone elements may have any directivity.
However, use of an omnidirectional microphone is easy to handle and inexpensive, so its
practical effect is large.
[0060]
In addition, it is easy to combine two microphone elements to obtain other directivity.
Furthermore, it is also easy to be almost omnidirectional in the low band and unidirectional in
the middle and high bands. Of course, three or more microphone elements can be used to obtain
a signal in accordance with the subject matter of the present invention.
[0061]
In addition, even if the steepness of the roll-off characteristic of the high-pass filter is controlled
instead of controlling the cut-off frequency of the high-pass filter, the effects of wind noise
reduction and quality improvement of the audio signal can be obtained.
[0062]
As described above, according to the present invention, the cut of the high-pass filter for
reducing the wind noise according to the size of the wind noise or according to the ratio between
the wind noise and the audio signal. Since the off frequency is changed, it is possible to
simultaneously obtain a good sound collection quality and a sufficient wind noise suppression
effect.
03-05-2019
15
[0063]
Further, according to the present invention, in the reproduction device, it is possible to minimize
the deterioration of the quality of the output sound signal and reduce wind noise contained in the
reproduction signal.
03-05-2019
16

Download Report