close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JP2015233200

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 JP2015233200
To suppress the influence of noise. Means: The noise detection circuit 106 determines whether
or not noise is included in the first input audio signal D1a and the second input audio signal D1b.
The noise removal circuit 108 outputs a first intermediate audio signal D2a corresponding to the
first input audio signal D1a and a second intermediate audio signal D2b corresponding to the
second input audio signal D1b, in which noise is not detected. ii) When noise is detected,
predetermined noise correction processing is performed on the first input audio signal D1a and
the second input audio signal D1b to generate a third intermediate audio signal, and a first
intermediate signal including the third intermediate audio signal An audio signal D2a and a
second intermediate audio signal D2b including a third intermediate audio signal are output. The
beam forming circuit 110 performs a beam forming process using a differential signal of the first
intermediate audio signal D2a and the second intermediate audio signal D2 b. [Selected figure]
Figure 3
Audio signal processing circuit and electronic device using the same
[0001]
The present invention relates to audio signal processing.
[0002]
Various electronic devices such as digital video cameras, digital cameras, mobile phone terminals,
notebook computers, car navigation systems, and headsets are equipped with a recording
function and a telephone call function.
11-04-2019
1
A technique called beamforming has been developed in order to enhance directivity in a specific
direction in which a sound source is present, to intensively record a target sound, and to remove
non-target sound from other directions.
[0003]
FIGS. 1A to 1C are diagrams schematically showing beam forming. The recording system 1 r
includes a signal processing circuit 10 and a plurality of microphones 12 a and 12 b. The
microphones 12 a and 12 b are nondirectional, and are arranged in the direction of the
directional axis 14 in close proximity at predetermined intervals.
[0004]
The signal processing circuit 10 receives audio signals S1a and S1b converted into electric
signals by the microphones 12a and 12b, respectively. The signal processing circuit 10 includes
a delay element 11 for delaying one audio signal S1b, and performs beamforming processing to
extract a target sound from a direction centered on the directional axis 14 and record the sound
clearly. The delay amount τ of the delay element 11 is set so that the detection level of the
sound from the direction opposite to the pointing axis 14 is substantially zero. Since the
algorithm of the beamforming process is known, the detailed description will be omitted here and
its principle will be briefly described.
[0005]
1 (a) shows the case where the sound source 2 exists in the direction of the directional axis 14,
FIG. 1 (b) shows the case where the sound source 2 exists in the direction perpendicular to the
directional axis 14, and FIG. The case where the sound source 2 exists in the opposite direction
to the directional axis 14 is shown. FIGS. 2 (a) to 2 (c) are waveform diagrams of audio signals
obtained in FIGS. 1 (a) to 1 (c), respectively. The vertical and horizontal axes of the waveform
diagrams and time charts in the present specification are scaled up and down as appropriate to
facilitate understanding, and each waveform shown is also simplified for ease of understanding,
Or it is emphasized.
11-04-2019
2
[0006]
Since the two microphones 12a and 12b are arranged closely at intervals of several cm, the
sound 4 emitted from the sound source 2 is input to the two microphones with almost the same
amplitude, and their phase difference Δφ It changes according to the direction. That is, as
shown in FIG. 1A, when the sound source 2 exists in the direction of the directional axis 14, the
phase difference between the two audio signals S1a and S1b becomes large. On the contrary, as
shown in FIG. 1B, when the sound source 2 is present in the direction 16 perpendicular to the
directional axis 14, the phase difference Δφ between the two audio signals S1a and S1b
approaches zero.
[0007]
In beam forming, gain differences (amplitude differences) and / or phase differences of audio
signals S1a and S1b output from the microphones 12a and 12b are used. When the two
waveforms are identical, the gain difference or the phase difference can be regarded as
essentially equivalent and has a correlation with the difference (S1a-S1b) between the two audio
signals S1a and S1b. Therefore, the audio signal processing circuit 10 can intensively collect the
sound 4 from the direction of the directional axis 14 by performing arithmetic processing using
the difference signal (S1a-S1b).
[0008]
International Publication No. 09/025090 pamphlet International Publication No. 09/045456
pamphlet
[0009]
As a result of examining the recording system 1r whose directivity has been improved by
beamforming, the present inventors came to recognize the following problem.
[0010]
For example, when an electronic device with a recording function using an ECM (Electret
Condenser Microphone) is used in a windy environment, the diaphragm of the ECM physically
11-04-2019
3
vibrates, so a noise called wind noise or wind noise is generated. It is recorded.
When such noise is mixed into the two audio signals S1a and S1b, the gain difference / phase
difference is greatly disturbed, which adversely affects the beam forming process.
Disturbances in gain difference / phase difference generate significant noise not only in the
frequency band where noise occurs but also in other bands. Similar problems can arise when
using a recording system in an environment where vibrations are applied.
[0011]
Such a problem should not be regarded as a general recognition of those skilled in the art, but is
uniquely recognized by the present inventors.
[0012]
The present invention has been made in view of such problems, and one of the exemplary objects
of an aspect thereof is to provide an audio signal processing circuit capable of suppressing the
influence of noise.
[0013]
One embodiment of the present invention relates to an audio signal processing circuit that
processes a first input audio signal and a second input audio signal collected by a first
microphone and a second microphone, respectively.
An audio signal processing circuit determines whether the first input audio signal and the second
input audio signal include noise exceeding tolerance, and generates a noise detection signal that
is asserted when the noise is included. And (i) when the noise detection signal is negated,
outputting a first intermediate audio signal according to the first input audio signal and a second
intermediate audio signal according to the second input audio signal, and (ii) noise detection
When the signal is asserted, the first intermediate audio signal including the third intermediate
audio signal is generated by performing a predetermined noise correction process on the first
input audio signal and the second input audio signal. A noise removal circuit that outputs a
second intermediate audio signal including the third intermediate audio signal, and Receiving a
first intermediate audio signal and the second intermediate audio signal output comprises a beam
11-04-2019
4
forming circuit which performs beam forming process by utilizing their difference signal.
[0014]
According to this aspect, when noise is generated, the first and second intermediate audio signals
are replaced with signals obtained using the third intermediate audio signal and provided to the
beam forming circuit in the subsequent stage. You can control the impact.
[0015]
The noise removal circuit may perform the noise correction process on predetermined bands of
the first input audio signal and the second input audio signal.
As a result, directivity can be maintained for bands outside the target of noise correction.
[0016]
The noise removal circuit may include a filter that divides each of the first input audio signal and
the second input audio signal into a plurality of bands.
[0017]
The noise removal circuit combines (i) the first input audio signal divided into a plurality of
bands when the noise detection signal is negated, and outputs a first intermediate audio signal
according to the combined signal. The second input audio signal divided into a plurality of bands
may be combined to generate a second intermediate audio signal according to the combined
signal.
Further, (ii) when the noise detection signal is asserted, the noise removal circuit performs noise
on a correction target band which is a predetermined one of the plurality of bands of the first
input audio signal and a correction target band of the second input audio signal. The third
intermediate audio signal is generated by performing correction processing, and the third
intermediate audio signal is combined with the other band of the first input audio signal to
generate a first intermediate audio signal, and the third intermediate audio signal is generated. ,
And other bands of the second input audio signal to generate a second intermediate audio signal.
11-04-2019
5
[0018]
The band to be corrected by the noise removal circuit may include a band of 0 to 500 Hz.
Thereby, noise due to vibration or wind can be suitably reduced.
[0019]
The noise removal circuit may correct the entire area of the first input audio signal and the
second input audio signal.
[0020]
The noise correction process may include a process of calculating an average value of two signals
to be corrected.
[0021]
The average value may be a simple average of two signals.
[0022]
The average value may be a weighted average of two signals.
In this case, the combination of weighting factors can adjust the trade-off relationship between
the effect of noise reduction and directivity.
The larger one weighting factor of the two signals may be smaller than the other smaller
weighting factor of the two signals.
Of the two signals, when the signal level is large, the level is likely to be increased by noise such
as wind noise. Therefore, by increasing the weighting factor of the smaller signal level, a large
11-04-2019
6
number of target sounds can be extracted to suppress noise.
[0023]
The noise correction process may further include at least one of a process of multiplying two
signals to be corrected by a predetermined coefficient and a process of multiplying an average
value by a predetermined coefficient.
[0024]
The coefficients may be variable depending on the level of noise detected.
[0025]
Another aspect of the present invention relates to an audio signal processing circuit that
processes a first input audio signal and a second input audio signal collected by a first
microphone and a second microphone, respectively.
This audio signal processing includes filters for dividing each of the first input audio signal and
the second input audio signal into a plurality of bands, and whether or not the first input audio
signal and the second input audio signal include noise exceeding an allowable amount. A noise
detection circuit that determines a noise detection signal that is asserted when included, and (i)
when the noise detection signal is negated, all of the first input audio signal and the second input
audio signal respectively. Beamforming processing for the band is performed, and (ii) when the
noise detection signal is asserted, correction target bands of the first input audio signal and the
second input audio signal are excluded from the targets of the beamforming processing, A beam
forming circuit for performing beam forming processing on the remaining bands; That.
[0026]
According to this aspect, it is possible to prevent the beamforming process from being performed
based on the difference signal in which the gain difference / phase difference is disturbed by the
noise, and the noise can be suppressed.
[0027]
The audio signal processing circuit according to an aspect of the present invention comprises a
first amplifier for amplifying an output signal of a microphone of a first channel, a second
11-04-2019
7
amplifier for amplifying an output signal of a microphone of a second channel, and a digital
output signal of the first amplifier. It may further comprise a first A / D converter for converting
into a first input audio signal, and a second A / D converter for converting the output signal of
the second amplifier into a digital second input audio signal.
[0028]
In one aspect, the audio signal processing circuit may be integrally integrated on one
semiconductor substrate.
"Integrated integration" includes the case where all of the circuit components are formed on a
semiconductor substrate, and the case where the main components of the circuit are integrally
integrated. A resistor, a capacitor or the like may be provided outside the semiconductor
substrate.
[0029]
Another aspect of the present invention relates to an electronic device.
The electronic device comprises a microphone of the first channel, a microphone of the second
channel, and any of the audio signal processing circuits described above.
[0030]
It is to be noted that any combination of the above-described constituent elements, or one in
which the constituent elements and expressions of the present invention are mutually replaced
among methods, apparatuses, systems, etc. is also effective as an aspect of the present invention.
[0031]
According to the audio signal processing circuit of the present invention, the influence of noise
can be suppressed.
[0032]
11-04-2019
8
FIGS. 1A and 1B are diagrams schematically showing beam forming.
2 (a) and 2 (b) are waveform diagrams of audio signals obtained in FIGS. 1 (a) and 1 (b),
respectively.
It is a block diagram of a sound recording system provided with the audio signal processing
circuit concerning a 1st embodiment.
It is a functional block diagram of a noise removal circuit. It is a figure which shows the spectrum
of the audio signal obtained by an audio signal processing circuit. It is a block diagram of a sound
recording system provided with the audio signal processing circuit concerning a 2nd
embodiment. It is a perspective view of the electronic device carrying an audio signal processing
circuit.
[0033]
Hereinafter, the present invention will be described based on preferred embodiments with
reference to the drawings. The same or equivalent components, members, and processes shown
in the drawings are denoted by the same reference numerals, and duplicating descriptions will be
omitted as appropriate. In addition, the embodiments do not limit the invention and are merely
examples, and all the features and combinations thereof described in the embodiments are not
necessarily essential to the invention.
[0034]
In the present specification, “the state in which the member A is connected to the member B”
means that the members A and B are electrically connected in addition to the case where the
members A and B are physically and directly connected. It also includes the case of indirect
connection via other members that do not substantially affect the connection state of the
connection or do not impair the function or effect provided by the connection. Similarly, "a state
where the member C is provided between the member A and the member B" means that the
member A and the member C, or the member B and the member C are directly connected, and It
11-04-2019
9
also includes the case of indirect connection via other members that do not substantially affect
the connection state of the connection or do not impair the function or effect provided by the
connection.
[0035]
First Embodiment FIG. 3 is a block diagram of a recording system 1 provided with an audio
signal processing circuit 10 according to a first embodiment. The recording system 1 includes an
audio signal processing circuit 10, a first microphone 12a, and a second microphone 12b. The
first microphone 12a and the second microphone 12b are nondirectional as in FIGS. 1 (a) and 1
(b), and are disposed close to each other at predetermined intervals in the direction of the
directional axis 14.
[0036]
The signal processing circuit 10 receives the audio signals S1a and S1b converted into electric
signals by the first microphone 12a and the second microphone 12b, respectively, and performs
a beam forming process to obtain a target sound from a direction centered on the directivity axis.
Extract and record clearly.
[0037]
The audio signal processing circuit 10 includes a first amplifier 102a, a second amplifier 102b, a
first A / D converter 104a, a second A / D converter 104b, a noise detection circuit 106, a noise
removal circuit 108, and a beam forming circuit 110. It is a functional IC (Integrated Circuit)
integrated on a semiconductor substrate.
The audio signal processing circuit 10 may further include a microphone bias circuit (not shown)
that supplies a bias voltage to the first microphone 12a and the second microphone 12b.
[0038]
The input terminals INA and INB of the audio signal processing circuit 10 are connected to the
first microphone 12a and the second microphone 12b via the DC block capacitors C1a and C1b,
and the analog from the first microphone 12a and the second microphone 12b It receives audio
signals S1a and S1b.
11-04-2019
10
[0039]
The first amplifier 102a and the second amplifier 102b respectively amplify the analog audio
signals S1a and S1b.
The first A / D converter 104a and the second A / D converter 104b convert the analog audio
signals S2a and S2b output from the corresponding first amplifier 102a and second amplifier
102b into digital audio signals D1a and D2b, respectively. The audio signals D1a and D1b are
respectively referred to as a first input audio signal and a second input audio signal.
[0040]
The noise detection circuit 106 receives the first input audio signal D1a and the second input
audio signal D1b, determines whether the audio signals S1a and S1b include noise exceeding an
allowable amount, and detects noise indicating the determination result. Signal S4 is generated.
For example, when the noise detection circuit 106 determines that noise is included, the noise
detection circuit 106 asserts (for example, high level) the noise detection signal S4. The method
of detecting the noise is not particularly limited, and a known technique described in, for
example, JP-A-2014-060525 or a technique available in the future may be used. The noise to be
detected is preferably low frequency noise represented by wind noise or vibration, more
specifically noise in a band of 0 to 300 Hz, 0 to 500 Hz, or 0 to 1 kHz, but is not particularly
limited.
[0041]
The noise removal circuit 108 receives the first input audio signal D1a and the second input
audio signal D1b, and outputs a first intermediate audio signal D2a and a second intermediate
audio signal D2b.
[0042]
When (i) the noise detection signal S4 is negated, that is, when the noise detection circuit 106
detects no noise exceeding the allowable value, the noise removal circuit 108 performs the first
intermediate audio signal according to the first input audio signal D1a. A second intermediate
audio signal D2b corresponding to D2a and the second input audio signal D1b is output.
11-04-2019
11
[0043]
たとえばD2a=D1a、D2b=D1bであってもよい。
In this case, when the noise detection signal S4 is negated, the noise removal circuit 108 passes
the input audio signals D1a and D1b as it is.
Alternatively, functions f (x) = αx + β defined using parameters α and β are defined, and D1a
and D1b are respectively processed using function f (x), and D2a = f (D1a) and D2b = F (D1 b)
may be output.
[0044]
When (ii) the noise detection signal S4 is asserted, that is, when the noise detection circuit 106
detects noise exceeding the allowable value, the noise removal circuit 108 generates the first
input audio signal D1a and the second input audio signal D1b. A third intermediate audio signal
D3 obtained by performing predetermined noise correction processing is generated. Then, the
noise removal circuit 108 outputs a first intermediate audio signal D2a including the component
of the third intermediate audio signal D3 and a second intermediate audio signal D2b including
the component of the third intermediate audio signal D3.
[0045]
The noise correction processing here is defined so as to be able to remove the influence of low
frequency noise such as wind noise. The noise correction process may be performed on the
entire area of the two signals D1a and D1b, but preferably may be performed only on a
predetermined band including noise such as wind noise and vibration. The noise correction
process is a process of reducing the disturbance that noise causes in the gain difference (phase
difference) generated in the beam forming circuit 110, and therefore can be regarded as phase
correction or gain correction.
11-04-2019
12
[0046]
As an example, the noise correction process includes a process of calculating an average value of
two signals D1a and D1b to be corrected. The average value Y may be a simple average of two
signals D1a and D1b. Y=(D1a+D1b)/2 …(1)
[0047]
The noise removal circuit 108 may use the average value Y of Equation (1) as the third
intermediate audio signal D3.
[0048]
Alternatively, a value Y 'obtained by multiplying the average value Y by a predetermined
coefficient K may be output as the third intermediate audio signal D3.
D3 = Y ′ = Y × K That is, in the noise correction processing, after performing processing of
multiplying each of two signals D1a and D1b to be corrected by a predetermined coefficient K, an
average value of equation (1) may be obtained Alternatively, a predetermined coefficient K may
be multiplied after the average value Y is obtained by the equation (1).
[0049]
The coefficient K may be fixed or variable. For example, switching may be performed in multiple
stages such as K = 1/2 when a weak wind is detected and K = 1/4 when a stronger wind is
detected. That is, the coefficient K may be variable according to the intensity of the wind, in other
words, according to the level of the detected noise. Furthermore, the coefficient K may be
changed based on other parameters other than the noise level.
[0050]
As the noise correction process, it is also possible to use root mean square (RMS) or other
processes.
11-04-2019
13
[0051]
The beam forming circuit 110 receives the first intermediate audio signal D2a and the second
intermediate audio signal D2b from the noise removing circuit 108, and performs a beam
forming process using at least a difference signal (D2a to D2b) thereof.
The algorithm of the beam forming process may be a known technique or a technique available
in the future, and is not particularly limited in the present invention.
[0052]
The output audio signals D4a and D4b subjected to the beamforming are supplied to a circuit at a
subsequent stage not shown. The processing of the latter stage of the beam forming circuit 110
is not particularly limited, and for example, it is written to a memory, etc. through digital signal
processing such as filtering and equalizing processing, compression processing and encoding
processing.
[0053]
FIG. 4 is a functional block diagram of the noise removal circuit 108. As shown in FIG. The noise
removal circuit 108 performs noise correction processing on a predetermined band (referred to
as a correction target band) including the noise frequency among the first input audio signal D1a
and the second input audio signal D1b.
[0054]
The noise removal circuit 108 is provided with a filter 112. The filter 112 divides each of the
first input audio signal D1a and the second input audio signal D1b into a plurality of bands. Here,
the first input audio signal D1a is divided into the high frequency component D1aH and the low
frequency component D1aL by the low pass filter and the high pass filter. Similarly, the low-pass
filter and the high-pass filter divide the second input audio signal D1b into a high frequency
component D1bH and a low frequency component D1bL. The low frequency component
corresponds to the correction target band.
11-04-2019
14
[0055]
When (i) the noise detection signal S4 is negated (low level, 0), the noise removal circuit 108
resynthesizes the first input audio signal D1a divided into the plurality of bands D1aL and D1aH
and is recombined. The first intermediate audio signal D2a corresponding to the signal is output.
The combination may be an addition operation. The second input audio signal D1b divided into
the plurality of bands D1bL and D1bH is recombined, and a second intermediate audio signal
D2b corresponding to the recombined signal is output.
[0056]
When (ii) the noise detection signal S4 is asserted (high level, 1), the noise removal circuit 108
corrects the correction target band D1aL of the plurality of bands of the first input audio signal
D1a and the second input audio signal D1b. The third intermediate audio signal D3 is generated
by performing the above-described noise correction process on the correction target band D1bL.
Then, the third intermediate audio signal D3 is combined with the other band D1aH of the first
input audio signal D1a to generate a first intermediate audio signal D2a. Similarly, the noise
removal circuit 108 combines the third intermediate audio signal D3 with the other band D1bH
of the second input audio signal D1b to generate a second intermediate audio signal D2b.
[0057]
These functions are realized by the noise correction unit 114, the first selector 116a, the second
selector 116b, the first combining unit 118a, and the second combining unit 118b.
[0058]
The noise correction unit 114 receives the correction target bands D1aL and D1bL of the first
input audio signal D1a and the second input audio signal D1b, respectively, and performs a
predetermined noise correction process on them to generate a third intermediate audio signal
D3. Do.
The noise correction process may be a simple average, as described above, in which case the
11-04-2019
15
third intermediate audio signal D3 is given by equation (2). D3=(D1aL+D1bL)/2
…(2)
[0059]
The first selector 116a receives the correction target band D1aL of the third intermediate audio
signal D3 and the first input audio signal D1a, selects D3 when the noise detection signal S4 is
asserted (1), and selects D1aL when negated. Do. Similarly, the second selector 116b receives the
correction target band D1bL of the third intermediate audio signal D3 and the second input
audio signal D1b, and when the noise detection signal S4 is asserted (1), D3 is negated, and D1bL
is negated. select.
[0060]
The first combining unit 118a adds the output of the first selector 116a and the component
D1aH of the first input audio signal D1a other than the correction target band. Similarly, the
second combining unit 118b adds the output of the second selector 116b and the component
D1bH other than the correction target band in the second input audio signal D1b.
[0061]
The configuration of the noise removal circuit 108 is not limited to that shown in FIG. 4, and it is
understood by those skilled in the art that there can be various modifications having equivalent
functions. For example, the first selector 116a and the second selector 116b may be omitted, and
the operation of the noise correction unit 114 may be switched according to the noise detection
signal S4. For example, (i) when the noise detection signal S4 is asserted, the noise correction
unit 114 outputs the third intermediate audio signal D3 to the first combining unit 118a and the
second combining unit 118b, and (ii) the noise detection signal S4 is When negated, the noise
correction unit 114 may output D1aL to the first combining unit 118a and output D1bL to the
second combining unit 118b without performing noise correction processing.
[0062]
The function of the noise correction unit 114 may be realized by hardware or may be realized by
a combination of an embedded processor and software.
11-04-2019
16
[0063]
The above is the configuration of the audio signal processing circuit 10 according to the
embodiment.
Subsequently, the operation will be described separately when noise is not detected and when
noise is detected.
[0064]
1. When noise is not detected When noise due to wind noise or vibration does not exist, the noise
detection signal S4 is negated. At this time, the noise removal circuit 108 passes through the
inputs D1a and D1b and outputs the signal to the beam forming circuit 110 in the subsequent
stage as it is. That is, the noise removal circuit 108 does not affect the input audio signal. The
beam forming circuit 110 in the subsequent stage receives the original input audio signals D1a
and D1b and performs beam forming processing. That is, the operation when no noise is input is
the same as in the prior art.
[0065]
2. When noise is detected When noise due to wind noise or vibration is present, the noise
detection signal S4 is asserted by the noise detection circuit 106. Thereby, the noise removal
circuit 108 performs noise correction processing on the correction target band of each of the
first input audio signal D1a and the second input audio signal D1b, and replaces it with the third
intermediate audio signal D3. To go through.
[0066]
The above is the operation of the audio signal processing circuit 10. According to the audio
signal processing circuit 10, since the first intermediate audio signal D2a and the second
intermediate audio signal D2b become the same component D3 in the correction target band, the
11-04-2019
17
directivity is lost, but the beam forming circuit From the difference signals (D2a-D2b) generated
at 110, the effects of wind noise and vibration can be removed. In addition, for bands other than
the correction target band, the same processing as that in the related art is performed, so
directivity can be maintained. From another point of view, when noise is detected, it can be
understood that the band including the noise is excluded from the target of the beamforming
process.
[0067]
FIG. 5 is a diagram showing the spectrum of the audio signal D4a (D4b) obtained by the audio
signal processing circuit 10. As shown in FIG. In the spectrum, the target sound 4 is input with
the wind of 4.5 m / s applied to the first microphone 12a and the second microphone 12b, and
the output audio signal D4a (D4b) obtained at this time is It is a fast Fourier transform (FFT).
[0068]
Graph (i) of FIG. 5 is a spectrum obtained by the audio signal processing circuit 10 according to
the embodiment. For comparison, a spectrum (ii) obtained when only the beamforming process is
performed, and a spectrum (iii) obtained when the noise correction process and the beam
forming process are not performed in the audio signal processing circuit It shows together.
[0069]
The spectrum (ii) of the audio signal processing circuit is equivalent to that obtained by forcibly
negating the noise detection signal S4 regardless of the detection result of the noise detection
circuit 106. The spectrum (iii) is obtained by forcibly negating the noise detection signal S4, and
stopping (through) the beam forming circuit 110.
[0070]
As can be seen from the comparison of graphs (ii) and (iii), when beam forming processing is
performed in a state where noise such as wind noise is input, the gain difference between the
first intermediate audio signal D2a and the second intermediate audio signal D2b ( Because the
11-04-2019
18
phase difference is greatly disturbed, the noise level in the 100 to 1 kHz band becomes very
large (ii). Further, the disturbance of the gain difference (phase difference) due to noise also
deteriorates the noise level in the band of 1 kHz or more.
[0071]
On the other hand, according to the audio signal processing circuit 10 according to the
embodiment, as shown in the graph (i), the beam forming process is performed on the noise level
although the beam forming process is performed. If not, it can be suppressed to a lower level
than in (iii). It should be noted that the noise reduction effect extends not only to the correction
target band (0 to 1 kHz) by the noise correction unit 114 but also to higher bands.
[0072]
Subsequently, several modifications of the audio signal processing circuit 10 according to the
first embodiment will be described.
[0073]
(First Modification) In the embodiment, the averaging process, more specifically, the simple
averaging is used as the noise correction process. However, the present invention is not limited
thereto.
The average value Y may be a weighted average of the two signals D1a and D1b. Y = (Ka × D1a
+ Kb × D1b) / (Ka + Kb) (3) Ka and Kb are weighting coefficients. If the condition of Ka + Kb = 1
is imposed, equation (3 ') is obtained. Y=(Ka×D1a+Kb×D1b) …(3’)
[0074]
The combination of Ka and Kb makes it possible to adjust the trade-off relationship between the
effect of noise reduction and directivity. For example, the coefficients Ka and Kb may be set
according to the signal levels | D1a | and | D1b | of the first input audio signal D1a and the
second input audio signal D1b, respectively.
11-04-2019
19
[0075]
Specifically, of the two signals D1a, D1b (or D1aL, D1bL), one of the weighting factors with the
larger signal level may be smaller than the other weighting factor with the smaller signal level of
the two signals. preferable. When | D1a |> | D1b |, it is estimated that a large amount of wind
noise is included on the first input audio signal D1a side. And in this case, the possibility that the
ratio indicated by the target sound is higher on the second input audio signal D1b side is high. In
this case, by setting Ka <Ka, a large number of target sounds included in the second input audio
signal D1b can be extracted to generate the third intermediate audio signal D3. On the contrary,
when | D1a | <| D1b |, it may be set as Ka> Kb.
[0076]
Second Modified Example Although the first input audio signal D1a and the second input audio
signal D1b are each divided into two bands in the embodiment, the present invention is not
limited thereto, and three or more bands may be used. It may be divided into
[0077]
In the embodiment, the noise removal circuit 108 targets only a predetermined band for noise
correction processing. However, the present invention is not limited to this, and all bands may be
subjected to noise correction.
[0078]
Third Modification In the embodiment, when noise is detected, the low frequency component
D2aL of the first intermediate audio signal D2a and the low frequency component D2bL of the
second intermediate audio signal D2b are the same signal, that is, the third intermediate
Although the case of the audio signal D3 has been described, the present invention is not limited
thereto.
The low frequency component D2aL of the first intermediate audio signal D2a and the low
frequency component D2bL of the second intermediate audio signal D2b do not necessarily have
to be the same signal and may include at least the third intermediate audio signal D3.
11-04-2019
20
D2aL is also the sum of D3 and a signal corresponding to the low frequency component D1aL of
the first input audio signal D1a, and D2bL is also the sum of D3 and a signal corresponding to
the low frequency component D1bL of the second input audio signal D1b Good. In this case,
directivity can be maintained also for the correction target band.
[0079]
Second Embodiment In the first embodiment, when noise is detected, the correction target band
of each of the first intermediate audio signal D2a and the second intermediate audio signal D2b
is set to the third intermediate audio signal D3. It can be understood that the processing of
replacing these bands is to exclude those bands from the target of beamforming processing by
the beamforming circuit 110.
[0080]
FIG. 6 is a block diagram of a recording system 1a including an audio signal processing circuit
10a according to the second embodiment.
The audio signal processing circuit 10a includes a first amplifier 102a, a second amplifier 102b,
a first A / D converter 104a, a second A / D converter 104b, a noise detection circuit 106, a filter
112, and a beam forming circuit 110a.
[0081]
The filter 112 divides each of the first input audio signal D1a and the second input audio signal
D1b into a plurality of bands.
[0082]
The noise detection circuit 106 determines whether the first input audio signal D1a and the
second input audio signal D1b include noise exceeding the allowable amount.
The noise detection signal S4 is asserted when it is determined that noise is included.
11-04-2019
21
[0083]
When (i) the noise detection signal S4 is negated, the beam forming circuit 110a performs beam
forming processing on all the bands of the first input audio signal D1a and the second input
audio signal D1b. Further, (ii) when the noise detection signal S4 is asserted, the beamforming
circuit 110a performs beamforming processing on the correction target band (low frequency
region) of each of the first input audio signal D1a and the second input audio signal D1b. And the
beamforming processing is performed for the remaining band (high frequency region).
[0084]
The beam forming circuit 110a includes a beam forming circuit 110, a third combining unit
120a, a fourth combining unit 120b, a fifth combining unit 122a, and a sixth combining unit
122b. The function of the beamforming circuit 110 is similar to that of FIG.
[0085]
When the noise detection signal S4 is negated, the third synthesis unit 120a resynthesizes the
plurality of bands D1aL and D1aH of the first input audio signal D1a divided by the filter 112,
and outputs the result to the beam forming circuit 110. When the noise detection signal S4 is
asserted, the third combining unit 120a outputs only D1aH to the beam forming circuit 110.
[0086]
Similarly, when the noise detection signal S4 is negated, the fourth synthesis unit 120b
resynthesizes the plurality of bands D1bL and D1bH of the second input audio signal D1b
divided by the filter 112 and outputs the result to the beam forming circuit 110. Do. When the
noise detection signal S4 is asserted, the fourth combining unit 120b outputs only D1bH to the
beam forming circuit 110.
[0087]
11-04-2019
22
The fifth combining unit 122a outputs the output D5a of the beamforming circuit 110 as it is
when the noise detection signal S4 is negated, and outputs the output D5a of the beamforming
circuit 110 when the noise detection signal S4 is asserted. Synthesize D1aL. Similarly, the sixth
combining unit 122b outputs the output D5b of the beamforming circuit 110 as it is when the
noise detection signal S4 is negated, and the output D5b of the beamforming circuit 110 when
the noise detection signal S4 is asserted. To synthesize D1bL.
[0088]
According to this audio signal processing circuit 10a, the same effect as that of the first
embodiment can be obtained.
[0089]
Finally, the application of the audio signal processing circuit 10 will be described.
FIG. 7 is a perspective view of an electronic device on which the audio signal processing circuit
10 is mounted. FIG. 7 is a digital video camera (camcorder) which is an example of the electronic
device.
[0090]
A digital video camera 800 includes a housing 802, a lens 804, an imaging device (not shown),
an image processing processor, and a recording medium. In addition to that, the digital video
camera 800 includes a first microphone 12 a, a second microphone 12 b, and an audio signal
processing circuit 10. The first microphone 12 a and the second microphone 12 b are disposed
along the direction of the directional axis 14.
[0091]
In addition, electronic devices include digital cameras, voice recorders, mobile phone terminals,
smart phones, PHS (Personal Handy-phone System), PDAs (Personal Digital Assistants), notebook
computers, tablet terminals, audio players, car navigation systems, heads It may be a set or the
like.
11-04-2019
23
[0092]
While the present invention has been described using specific terms based on the embodiments,
the embodiments merely show the principles and applications of the present invention, and the
embodiments are defined in the claims. Many variations and modifications of the arrangement
can be made without departing from the concept of the present invention.
[0093]
DESCRIPTION OF SYMBOLS 1 ... Recording system, 2 ... sound source, 4 ... sound, 10 ... Audio
signal processing circuit, 12a ... 1st microphone, 12b ... 2nd microphone, 14 ... Directional axis,
102a ... 1st amplifier, 102b ... 2nd amplifier, 104a ... 1st A / D converter, 104b ... 2nd A / D
converter, 106 ... noise detection circuit, 108 ... noise removal circuit, 110 ... beam forming
circuit, 112 ... filter, 114 ... noise correction unit, 116a ... first selector, 116b ... second selector
118a first combining unit 118b second combining unit 120a third combining unit 120b fourth
combining unit 122a fifth combining unit 122b sixth combining unit D1a sixth 1 input audio
signal, D1 b: second input audio signal, D2 a: first intermediate audio signal, D2 b: second
intermediate audio signal, D3: third During audio signal, S4 ... noise detection signal, 800 ...
digital video camera.
11-04-2019
24
Документ
Категория
Без категории
Просмотров
0
Размер файла
36 Кб
Теги
description, jp2015233200
1/--страниц
Пожаловаться на содержимое документа