JP3186411B2 - Sound pickup device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sound pickup apparatus suitable for use in, for example, a VTR-integrated video camera.

[0002]

2. Description of the Related Art For example, as a sound pickup device that performs stereo sound pickup suitable for use in a VTR integrated video camera or the like,
The applicant of the present application has previously filed Japanese Patent Application No.
-27298). That is, in this device, spatial and electrical processing is performed using two microphones provided relatively close to each other, so that good stereo sound pickup can be performed.

When such a VTR-integrated video camera or the like is used, for example, outdoors, there arises a problem of peculiar wind noise generated when the wind hits a microphone. In order to reduce such wind noise, the applicant of the present application has previously proposed Japanese Patent Application No. 3-153341 (see Japanese Patent Application Laid-Open No. Hei 5-73392).

[0004] That is, in this device, it is considered that the correlation between low-frequency sound signals is high in signals picked up by two microphones provided relatively close to each other.
On the other hand, wind noise is considered to have low correlation. Therefore, the difference component between the signals picked up by the two microphones is extracted, the peak value of the difference component is detected, and particularly the low-frequency component of the picked up signal is attenuated according to the peak value. It is.

Therefore, in this device, wind noise having low correlation is mainly extracted from the difference component, and the attenuation of the low-frequency component of the collected signal is controlled according to the peak value of the difference component. Accordingly, the low-frequency component of the signal collected only when the wind noise is large is attenuated, and the wind noise is reduced.

[0006]

However, in this device, when there is a lot of wind noise, all the low-frequency components are attenuated, so that the sound signal to be picked up is attenuated. Also, the fluctuation of the peak value of the wind noise changes sharply, and if the attenuation is controlled in accordance with this peak value, the attenuation is frequently controlled, which may give an unnatural sense of hearing. is there.

This application has been made in view of such a point, and the problem to be solved is that a conventional apparatus attenuates an original audio signal to be picked up.
Further, the amount of attenuation is frequently controlled, which may give an unnatural sense of hearing.

[0008]

A first means according to the present invention has two or more microphones 1A and 1B, and extracts a low-frequency difference component of a signal picked up by these microphones (operation unit 5). , A low-pass filter 6), and subtracts the difference component from the signals picked up by the microphones (arithmetic units 4A and 4B). The level of the subtracted difference component is determined by envelope detection of the difference component. An envelope detection circuit 9) is a sound collection device in which the detected signal is controlled (amplifier 7) by a signal in which a predetermined attack characteristic (characteristic setting circuit 10) and a recovery characteristic (characteristic setting circuit 11) are set.

According to a second aspect of the present invention, in the sound collection device according to the first aspect, a predetermined adaptive process is performed on the signal from which the difference component has been subtracted to further reduce the wind noise. This is a sound pickup device.

A third means according to the present invention is the sound pickup apparatus according to the second means, wherein said adaptive processing supplies said difference component to adaptive filters 17A and 17B, and obtains a signal of said difference component. This is a sound pickup device configured to subtract from low-frequency components of the subtracted signal (arithmetic units 15A and 15B).

According to a fourth aspect of the present invention, in the sound pickup apparatus according to the second means, the adaptive processing is performed using a control signal for controlling a level of the difference component. It is.

[0012]

According to this, a low-frequency difference component of a signal picked up by two or more microphones is extracted, and the difference component is subtracted from the signal picked up by each microphone. At the same time, by controlling the level of the subtracted difference component with a signal in which a predetermined attack characteristic and a recovery characteristic are set in the envelope detection signal of the difference component, the influence on the collected audio signal is minimized. Thus, only wind noise can be reduced, and a good sound signal can be collected.

Further, in addition to the above-described processing, by applying a predetermined adaptive processing to the signal from which the difference component has been subtracted, wind noise can be further reduced, and in this case, the signals are collected by the respective microphones. Since the process of subtracting the difference component from the sounded signal is performed first, the load of the adaptive process can be reduced, and a more favorable sound signal can be collected with a simple configuration.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In FIG. 1, 1A and 1B are, for example, VTRs.
Figure 2 shows two microphones provided in close proximity, such as a stereo microphone of an integrated video camera.
The signals picked up by the microphones 1A and 1B are passed through amplifiers 2A and 2B, respectively, to A / D converters 3A and 3B.
Supplied to Output signals of the A / D converters 3A and 3B are supplied to arithmetic units (Σ) 4A and 4B, respectively.

The output signals of the A / D converters 3A and 3B are supplied to a computing unit (Σ) 5 where the output signal of the A / D converter 3B is subtracted from the output signal of the A / D converter 3A. You.
The subtraction output of the arithmetic unit 5 is a low-pass filter (LPF)
6. As a result, the low-pass filter 6 extracts a low-frequency difference component of the signal collected by the two microphones 1A and 1B.

The output signal of the low-pass filter 6 is supplied to an amplifier 7. The output signal of the low-pass filter 6 is supplied to an envelope detection circuit (DET) 9 through an amplifier 8, and the detection signal is supplied to an attack characteristic setting circuit 10 and a recovery characteristic setting circuit 11. The output signals of these characteristic setting circuits 10 and 11 are applied to a computing unit (Σ) 1
2 and added to each other.

The signal added by the arithmetic unit 12 is supplied to the amplifier 7, and the gain is controlled by the added signal. As a result, the level of the output signal of the low-pass filter 6 passing through the amplifier 7 is controlled. Then, the output signal of the amplifier 7 is supplied to the arithmetic unit 4A, and is subtracted from the output signal of the A / D converter 3A. The output signal of the amplifier 7 is supplied to the arithmetic unit 4B, and is added to the output signal of the A / D converter 3B.

Further, output signals of the arithmetic units 4A and 4B are supplied to low-pass filters (LPF) 13A and 13B and high-pass filters (HPF) 14A and 14B, respectively. Output signals of these low-pass filters 13A and 13B are supplied to arithmetic units (Σ) 15A and 15B, respectively. The output signals of the high-pass filters 14A and 14B are supplied to computing units (Σ) 16A and 16B, respectively.

The output signal of the low-pass filter 6 is supplied to adaptive filters 17A and 17B described later. Further, subtraction signals to be described later from the calculators 15A and 15B are supplied to the adaptive filters 17A and 17B, respectively. The addition signal from the arithmetic unit 12 is applied to the adaptive filters 17A and 17B.
Supplied to The coefficients of the adaptive filters 17A and 17B are controlled according to the addition signal and the subtraction signal.

Further, these adaptive filters 17A, 17
The output signal of B is supplied to arithmetic units 15A and 15B, respectively, and is subtracted from the output signals of low-pass filters 13A and 13B. The signals subtracted by these computing units 15A and 15B are supplied to computing units 16A and 16B, respectively, and added to the output signals of high-pass filters 14A and 14B.
Then, the added signals are output terminals 18A and 18B, respectively.
Is taken out.

Therefore, in this device, the low-pass filter 6 outputs a low-frequency difference component of the signal collected by the two microphones 1A and 1B provided in close proximity, that is, a component having low correlation in the low frequency band, that is, Wind noise is extracted. The level of the wind noise is controlled by the amplifier 7 and is subtracted from the original output signals of the A / D converters 3A and 3B by the calculators 4A and 4B. Since the output signal of the A / D converter 3B is subtracted by the arithmetic unit 5, the arithmetic unit 4B performs the subtraction by addition.

As a result, in the arithmetic units 4A and 4B, only the wind noise component in the signals collected by the microphones 1A and 1B is subtracted, and thus the original audio signal to be collected is attenuated. Without noise, wind noise can be eliminated.

In this device, the wind noise component from the low-pass filter 6 is envelope-detected by the detection circuit 9, and
The level of the wind noise subtracted by the arithmetic units 4A and 4B is controlled by the detected signal. Thus, only when wind noise is generated, wind noise is removed in accordance with the level.

Further, an attack characteristic and a recovery characteristic for controlling the level of the wind noise are set by setting circuits 10 and 11. That is, for example, at the beginning of wind noise, in order to improve the response to a gust or the like, an attack characteristic with a quick response is set in the setting circuit 10 so that the control characteristic follows this rise. Further, when the level of the wind noise becomes small, if the response of the control is made faster, the change point can be easily recognized by the audibility, so that a recovery characteristic that makes the response slower is set in the setting circuit 11.

Thus, according to this apparatus, two or more microphones 1A and 1B are provided, and a low-frequency difference component of a signal picked up by these microphones is extracted (arithmetic unit 5, low-pass filter 6). The difference component is subtracted from the signals picked up by the respective microphones (calculator 4A,
Along with 4B), the level of the subtracted difference component is set to a predetermined attack characteristic and a recovery characteristic in an envelope detection signal (detection circuit 9) of the difference component (characteristic setting circuits 10 and 1).
1) By controlling (amplifier 7) with the signal thus obtained, it is possible to minimize the influence on the sound signal to be picked up, reduce only wind noise, and perform sound pickup of a good sound signal. is there.

In this apparatus, the setting circuits 10 and 11 for setting the attack characteristic and the recovery characteristic are of, for example, a first-order time constant processing type. Processing methods can also be employed. This is a response to the above, in which the time constant corresponding to the recovery characteristic is changed according to the situation.

That is, for a large instantaneous noise input such as a plosive, for example, the input of the noise is completed in an instant. Therefore, in such a case, the response time is changed according to the polarity of the signal change so that the recovery time is shortened, and when the light enters stably for a certain time or more, the recovery time is lengthened. In this way, by applying a contrivance to the signal processing, it is possible to obtain a natural audible response characteristic to a change in a complicated level of wind noise.

Further, in this device, the operation units 4A, 4A
It is difficult to completely remove the wind noise from the output signal of B, and if the level of the wind noise component subtracted by the arithmetic units 4A and 4B is increased, the influence of the output signal of the microphone on the opposite side occurs. In some cases. Therefore, in this device, wind noise is further reduced by further performing adaptive processing on the output signals of the arithmetic units 4A and 4B.

That is, in this device, the characteristics of the low-pass filters 13A and 13B are the same as those of the low-pass filter 6 described above. On the other hand, the high-pass filter 14
The characteristics of A and 14B are the same as those of the low-pass filters 13A and 13B.
When the signals separated by these filters are added as they are, the signal is restored to the original signal.

The components of the wind noise from the low-pass filter 6 are supplied as reference signals to the adaptive filters 17A and 17B, and the output signals of the adaptive filters 17A and 17B are respectively processed by the arithmetic units 15A and 15B. It is subtracted from the output signals of 13A and 13B. Further, these subtraction signals are respectively applied to the adaptive filters 17.
A and 17B are fed back. The arithmetic unit 12
Are supplied to the adaptive filters 17A and 17B.

Therefore, these adaptive filters 17A, 17A
B, using the above-mentioned signals, respectively,
The output of 5B is configured to minimize the wind noise components that cannot be removed by the processing up to the arithmetic units 4A and 4B. An example of the configuration of such adaptive filters 17A and 17B is, for example, an LMS (Least Mesh) as shown in FIG.
an Square) algorithm can be applied.

That is, in FIG. 2, the wind noise component from the above-described low-pass filter 6 is supplied to a series circuit of arbitrary delay elements 71, 72... 7n. The input and output signals of these delay elements 71, 72... 7n are supplied to arithmetic units 80, 81. 8n are multiplied by coefficients W _0k , W _1k ... W _nk , respectively. Further, the output signals of these computing units 80, 81.
Is added.

Further, in this circuit, arithmetic units 80 and 8
Each of the coefficients W _0k , W _1k ... W _{nk of} 1... 8n is obtained by the arithmetic circuit 100. The arithmetic circuit 100 is supplied with the subtraction signal from the arithmetic unit 15A or 15B and a signal for controlling the level of the wind noise added by the arithmetic unit 12.

[0034] Then, in the arithmetic circuit 100, for example, _{W i (k + 1) =} W ik + 2με k X k However, k: Absolute performs processing time mu: a gain factor that determines the adaptation speed and stability, calculator 12 corresponds to a signal for controlling the level of the wind noise added. ε _k : An error one clock before, and corresponds to a subtraction signal from the computing unit 15A or 15B. .. 8n are obtained to obtain the coefficients W _0k , W _1k ... W _{nk of} the computing units 80, 81.

Thus, the adaptive filters 17A and 17B
, An output signal is extracted from the outputs of the computing units 15A and 15B so as to minimize the components of the wind noise that could not be removed by the processing up to the computing units 4A and 4B. The output signals of these adaptive filters 17A and 17B are supplied to arithmetic units 15A and 15B, respectively, and are subtracted from the output signals of low-pass filters 13A and 13B. The processing effect can be enhanced as the number of steps of the delay element increases.

Therefore, in this device, the adaptive filter 1
The output signals of 7A and 17B are output from arithmetic units 15A and 15A, respectively.
5B to be subtracted from the output signals of the low-pass filters 13A and 13B, and to minimize the wind noise component in the subtracted signal that could not be completely removed by the processing at the arithmetic units 4A and 4B. Adaptive filters 17A, 17
B is controlled.

In this case, in these adaptive filters 17A and 17B, since most of the wind noise is removed by the processing up to the arithmetic units 4A and 4B, it is sufficient to remove only a small amount of the remaining components. As compared with a device that removes all the wind noises by using only the adaptive filter, it is possible to remove good wind noises with a small-scale circuit.

According to this apparatus, in addition to the above-described process of subtracting the difference component from the signals collected by the respective microphones 1A and 1B, a predetermined adaptive process (the adaptive filter 17A) is performed on the signal from which the difference component has been subtracted. , 17B), the wind noise can be further reduced, and in this case, the load of the adaptive processing can be reduced because the processing for subtracting the difference component is performed first, With the simple configuration, it is possible to perform better sound signal collection.

Further, the signals from which the low-frequency wind noise component has been removed by these adaptive processes are output to the arithmetic units 16 respectively.
A, 16B and supplied to the high-pass filters 14A, 14A, 1B.
It is added to the high frequency signal from 4B. As a result, the respective microphones 1 are connected to the output terminals 18A and 18B.
An audio signal from which the wind noise has been removed is extracted from the signals collected in A and 1B.

In the above-described apparatus, the low-pass filters 6, 13A and 13B and the high-pass filter 14
The cutoff frequencies of A and 14B are
Although the optimum value changes depending on A, 1B and the surrounding structure, it is generally effective to select a value between 200 and 500 Hz.

In the above-described apparatus, the adaptive filter 1
The algorithms applied to 7A and 17B are not limited to the LMS (Least Mean Square) algorithm described above, but include learning identification, affine projection, least squares, and RL.
Various algorithms such as S (Recursive Least Square) and high-speed Kalman can be applied.

Further, in the above embodiment, the signals collected by the microphones 1A and 1B are converted into A / D converters 3A and 3B.
Although the processing is carried out by digitizing in the present invention, the invention of the present application can be similarly implemented in analog processing. However, especially when adaptive processing is included, digital processing is advantageous and the possibility of implementation is high. When performing digital processing, a so-called digital signal processor (D
SP).

In the above-described embodiment, the case where there are two microphones has been described. However, when three microphones are used, the operation can be performed as shown in FIG.
That is, in FIG. 3, three microphones 1A, 1
The signals picked up by B and 1C are amplifiers 2A and 2A, respectively.
The signals are supplied to A / D converters 3A, 3B and 3C through B and 2C.

Further, for example, the signal collected by the central microphone 1C is supplied to the arithmetic unit 50 as it is. The signals picked up by the left and right microphones 1A and 1B are attenuated (bit-shifted) by 1/2 arithmetic units 51A and 51B and supplied to the arithmetic unit 50. Then, these attenuated signals are subtracted from the signal collected by the central microphone 1C, and the subtracted signal is supplied to the low-pass filter 6, and the same processing as described above is performed. In this way, the processing when three microphones are used is performed.

The signal processing shown in the above-mentioned Japanese Patent Application No. Hei 2-132051 (see Japanese Patent Application Laid-Open No. Hei 4-27298) can be performed, for example, at an arbitrary position in the circuit of the embodiment of the present invention shown in FIG. it can.

[0046]

According to the present invention, a signal having two or more microphones, a low-frequency difference component of a signal collected by these microphones is extracted, and the difference component is collected by each microphone. By subtracting the difference component from the signal and controlling the level of the subtracted difference component with a signal in which a predetermined attack characteristic and recovery characteristic are set to the envelope detection signal of the difference component, the influence on the collected sound is minimized. To reduce wind noise only, and achieve good sound signal collection.

In addition to the above-described processing, by applying a predetermined adaptive processing to the signal from which the difference component has been subtracted, wind noise can be further reduced, and in this case, the signals can be collected by the respective microphones. Since the processing of subtracting the difference component from the sounded signal is performed first, the load of the adaptive processing can be reduced, and a better sound signal can be collected with a simple configuration. Was.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an example of a sound collection device according to the present invention.

FIG. 2 is a configuration diagram of an example of an adaptive filter applied to the present invention.

FIG. 3 is a configuration diagram of another example of the sound collection device according to the present invention.

[Explanation of symbols]

1A, 1B microphone 2A, 2B, 7, 8 amplifier 3A, 3B A / D converter 4A, 4B, 5, 12 arithmetic unit 6, 13A, 13B, 15A, 15B, 16A, 16B
Low-pass filter 9 Envelope detection circuit 10 Attack characteristic setting circuit 11 Recovery characteristic setting circuit 14A, 14B High-pass filter 17A, 17B Adaptive filter 18A, 18B Output terminal

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int. Cl. ⁷ , DB name) H04R 3/00 320 H03H 21/00 H04N 5/225 H04R 1/40 320

Claims (4)

(57) [Claims] 1. A system comprising two or more microphones, extracting a low-frequency difference component of a signal collected by these microphones, subtracting the difference component from a signal collected by each of the microphones, A sound pickup device wherein the level of the difference component to be subtracted is controlled by a signal in which the difference component is envelope-detected and a predetermined attack characteristic and recovery characteristic are set to the detected signal. 2. The sound pickup device according to claim 1, wherein the signal from which the difference component has been subtracted is subjected to a predetermined adaptive process to further reduce the wind noise. 3. The sound collection device according to claim 2, wherein the adaptive processing supplies the difference component to an adaptive filter,
A sound collection device wherein the obtained signal is subtracted from a low-frequency component of the signal from which the difference component has been subtracted. 4. The sound pickup apparatus according to claim 2, wherein the adaptive processing is performed using a control signal for controlling a level of the difference component.